install script: remove newlines from KEY (#1139)

* add cron

* update the install script

.github/ISSUE_TEMPLATE/bug_report.yml

@@ -1,8 +1,19 @@
 name: 🐛 Bug report
 description: Report a new bug or issue.
 title: '[Bug]: '
-labels: ['bug']
+labels: ['bug', "needs confirmation"]
 body:
+  - type: dropdown
+    id: component
+    attributes:
+      label: Component
+      description: Which part of Beszel is this about?
+      options:
+        - Hub
+        - Agent
+        - Hub & Agent
+    validations:
+      required: true
   - type: markdown
     attributes:
       value: |
@@ -43,6 +54,39 @@ body:
         3. Pour it into a cup.
     validations:
       required: true
+  - type: dropdown
+    id: category
+    attributes:
+      label: Category
+      description: Which category does this relate to most?
+      options:
+        - Metrics
+        - Charts & Visualization
+        - Settings & Configuration
+        - Notifications & Alerts
+        - Authentication
+        - Installation
+        - Performance
+        - UI / UX
+        - Other
+    validations:
+      required: true
+  - type: dropdown
+    id: metrics
+    attributes:
+      label: Affected Metrics
+      description: If applicable, which specific metric does this relate to most?
+      options:
+        - CPU
+        - Memory
+        - Storage
+        - Network
+        - Containers
+        - GPU
+        - Sensors
+        - Other
+    validations:
+      required: true
   - type: input
     id: system
     attributes:
@@ -61,7 +105,6 @@ body:
     id: install-method
     attributes:
       label: Installation method
-      default: 0
       options:
         - Docker
         - Binary

.github/ISSUE_TEMPLATE/feature_request.yml

@@ -1,8 +1,19 @@
 name: 🚀 Feature request
 description: Request a new feature or change.
 title: "[Feature]: "
-labels: ["enhancement"]
+labels: ["enhancement", "needs review"]
 body:
+  - type: dropdown
+    id: component
+    attributes:
+      label: Component
+      description: Which part of Beszel is this about?
+      options:
+        - Hub
+        - Agent
+        - Hub & Agent
+    validations:
+      required: true
   - type: markdown
     attributes:
       value: Before submitting, please search existing [issues](https://github.com/henrygd/beszel/issues) and [discussions](https://github.com/henrygd/beszel/discussions) (including closed).
@@ -11,8 +22,55 @@ body:
       label: Describe the feature you would like to see
     validations:
       required: true
+  - type: textarea
+    id: motivation
+    attributes:
+      label: Motivation / Use Case
+      description: Why do you want this feature? What problem does it solve?
+    validations:
+      required: true
   - type: textarea
     attributes:
       label: Describe how you would like to see this feature implemented
     validations:
       required: true
+  - type: textarea
+    id: logs
+    attributes:
+      label: Screenshots
+      description: Please attach any relevant screenshots, such as images from your current solution or similar implementations.
+    validations:
+      required: false
+  - type: dropdown
+    id: category
+    attributes:
+      label: Category
+      description: Which category does this relate to most?
+      options:
+        - Metrics
+        - Charts & Visualization
+        - Settings & Configuration
+        - Notifications & Alerts
+        - Authentication
+        - Installation
+        - Performance
+        - UI / UX
+        - Other
+    validations:
+      required: true
+  - type: dropdown
+    id: metrics
+    attributes:
+      label: Affected Metrics
+      description: If applicable, which specific metric does this relate to most?
+      options:
+        - CPU
+        - Memory
+        - Storage
+        - Network
+        - Containers
+        - GPU
+        - Sensors
+        - Other
+    validations:
+      required: true

.github/pull_request_template.md

@@ -0,0 +1,33 @@
+## 📃 Description
+
+A short description of the pull request changes should go here and the sections below should list in detail all changes. You can remove the sections you don't need.
+
+## 📖 Documentation
+
+Add a link to the PR for [documentation](https://github.com/henrygd/beszel-docs) changes.
+
+## 🪵 Changelog
+
+### ➕ Added
+
+- one
+- two
+
+### ✏️ Changed
+
+- one
+- two
+
+### 🔧 Fixed
+
+- one
+- two
+
+### 🗑️ Removed
+
+- one
+- two
+
+## 📷 Screenshots
+
+If this PR has any UI/UX changes it's strongly suggested you add screenshots here.

.github/workflows/docker-images.yml

@@ -3,7 +3,7 @@ name: Make docker images
 on:
   push:
     tags:
-      - 'xv*'
+      - "v*"
 
 jobs:
   build:
@@ -14,28 +14,48 @@ jobs:
         include:
           - image: henrygd/beszel
             context: ./beszel
-            dockerfile: ./beszel/dockerfile_Hub
+            dockerfile: ./beszel/dockerfile_hub
             registry: docker.io
             username_secret: DOCKERHUB_USERNAME
             password_secret: DOCKERHUB_TOKEN
+
           - image: henrygd/beszel-agent
             context: ./beszel
-            dockerfile: ./beszel/dockerfile_Agent
+            dockerfile: ./beszel/dockerfile_agent
             registry: docker.io
             username_secret: DOCKERHUB_USERNAME
             password_secret: DOCKERHUB_TOKEN
+
+          - image: henrygd/beszel-agent-nvidia
+            context: ./beszel
+            dockerfile: ./beszel/dockerfile_agent_nvidia
+            platforms: linux/amd64
+            registry: docker.io
+            username_secret: DOCKERHUB_USERNAME
+            password_secret: DOCKERHUB_TOKEN
+
           - image: ghcr.io/${{ github.repository }}/beszel
             context: ./beszel
-            dockerfile: ./beszel/dockerfile_Hub
+            dockerfile: ./beszel/dockerfile_hub
             registry: ghcr.io
             username: ${{ github.actor }}
             password_secret: GITHUB_TOKEN
+
           - image: ghcr.io/${{ github.repository }}/beszel-agent
             context: ./beszel
-            dockerfile: ./beszel/dockerfile_Agent
+            dockerfile: ./beszel/dockerfile_agent
             registry: ghcr.io
             username: ${{ github.actor }}
             password_secret: GITHUB_TOKEN
+
+          - image: ghcr.io/${{ github.repository }}/beszel-agent-nvidia
+            context: ./beszel
+            dockerfile: ./beszel/dockerfile_agent_nvidia
+            platforms: linux/amd64
+            registry: ghcr.io
+            username: ${{ github.actor }}
+            password_secret: GITHUB_TOKEN
+
     permissions:
       contents: read
       packages: write
@@ -65,6 +85,7 @@ jobs:
         with:
           images: ${{ matrix.image }}
           tags: |
+            type=raw,value=edge
             type=semver,pattern={{version}}
             type=semver,pattern={{major}}.{{minor}}
             type=semver,pattern={{major}}
@@ -72,7 +93,9 @@ jobs:
 
       # https://github.com/docker/login-action
       - name: Login to Docker Hub
-        if: github.event_name != 'pull_request'
+        env: 
+            password_secret_exists: ${{ secrets[matrix.password_secret] != '' && 'true' || 'false' }}
+        if: github.event_name != 'pull_request' && env.password_secret_exists == 'true'
         uses: docker/login-action@v3
         with:
           username: ${{ matrix.username || secrets[matrix.username_secret] }}
@@ -84,9 +107,9 @@ jobs:
       - name: Build and push Docker image
         uses: docker/build-push-action@v5
         with:
-          context: '${{ matrix.context }}'
+          context: "${{ matrix.context }}"
           file: ${{ matrix.dockerfile }}
-          platforms: linux/amd64,linux/arm64,linux/arm/v7
+          platforms: ${{ matrix.platforms || 'linux/amd64,linux/arm64,linux/arm/v7' }}
-          push: ${{ github.ref_type == 'tag' }}
+          push: ${{ github.ref_type == 'tag' && secrets[matrix.password_secret] != '' }}
           tags: ${{ steps.metadata.outputs.tags }}
           labels: ${{ steps.metadata.outputs.labels }}

.github/workflows/inactivity-actions.yml

@@ -0,0 +1,43 @@
+name: 'Issue and PR Maintenance'
+
+on:
+  schedule:
+    - cron: '0 0 * * *'   # runs at midnight UTC
+  workflow_dispatch:
+
+permissions:
+  issues: write
+  pull-requests: write
+
+jobs:
+  close-stale:
+    name: Close Stale Issues
+    runs-on: ubuntu-24.04
+    steps:
+      - name: Close Stale Issues
+        uses: actions/stale@v9
+        with:
+          repo-token: ${{ secrets.GITHUB_TOKEN }}
+
+          # Messaging
+          stale-issue-message: >
+            👋 This issue has been automatically marked as stale due to inactivity.
+            If this issue is still relevant, please comment to keep it open.
+            Without activity, it will be closed in 7 days.
+
+          close-issue-message: >
+            🔒 This issue has been automatically closed due to prolonged inactivity.
+            Feel free to open a new issue if you have further questions or concerns.
+
+          # Timing
+          days-before-issue-stale: 14
+          days-before-issue-close: 7
+
+          # Labels
+          stale-issue-label: 'stale'
+          remove-stale-when-updated: true
+          only-issue-labels: 'awaiting-requester'
+
+          # Exemptions
+          exempt-assignees: true
+          exempt-milestones: true

.github/workflows/label-from-dropdown.yml

@@ -0,0 +1,82 @@
+name: Label issues from dropdowns
+
+on:
+  issues:
+    types: [opened]
+
+jobs:
+  label_from_dropdown:
+    runs-on: ubuntu-latest
+    permissions:
+      issues: write
+    steps:
+      - name: Apply labels based on dropdown choices
+        uses: actions/github-script@v7
+        with:
+          script: |
+
+            const issueNumber = context.issue.number;
+            const owner = context.repo.owner;
+            const repo = context.repo.repo;
+
+            // Get the issue body
+            const body = context.payload.issue.body;
+
+            // Helper to find dropdown value in the body (assuming markdown format)
+            function extractSectionValue(heading) {
+              const regex = new RegExp(`### ${heading}\\s+([\\s\\S]*?)(?:\\n###|$)`, 'i');
+              const match = body.match(regex);
+              if (match) {
+                // Get the first non-empty line after the heading
+                const lines = match[1].split('\n').map(l => l.trim()).filter(Boolean);
+                return lines[0] || null;
+              }
+              return null;
+            }
+
+            // Extract dropdown selections
+            const category = extractSectionValue('Category');
+            const metrics = extractSectionValue('Affected Metrics');
+            const component = extractSectionValue('Component');
+
+            // Build labels to add
+            let labelsToAdd = [];
+            if (category) labelsToAdd.push(category);
+            if (metrics) labelsToAdd.push(metrics);
+            if (component) labelsToAdd.push(component);
+
+            // Get existing labels in the repo
+            const { data: existingLabels } = await github.rest.issues.listLabelsForRepo({
+              owner,
+              repo,
+              per_page: 100
+            });
+            const existingLabelNames = existingLabels.map(l => l.name);
+
+            // Find labels that need to be created
+            const labelsToCreate = labelsToAdd.filter(label => !existingLabelNames.includes(label));
+
+            // Create missing labels (with a default color)
+            for (const label of labelsToCreate) {
+              try {
+                await github.rest.issues.createLabel({
+                  owner,
+                  repo,
+                  name: label,
+                  color: 'ededed' // light gray, you can pick any hex color
+                });
+              } catch (e) {
+                // Ignore if label already exists (race condition), otherwise rethrow
+                if (!e || e.status !== 422) throw e;
+              }
+            }
+
+            // Now apply all labels (they all exist now)
+            if (labelsToAdd.length > 0) {
+              await github.rest.issues.addLabels({
+                owner,
+                repo,
+                issue_number: issueNumber,
+                labels: labelsToAdd
+              });
+            } 

.github/workflows/release.yml

@@ -3,7 +3,7 @@ name: Make release and binaries
 on:
   push:
     tags:
-      - 'v*'
+      - "v*"
 
 permissions:
   contents: write
@@ -29,7 +29,17 @@ jobs:
       - name: Set up Go
         uses: actions/setup-go@v5
         with:
-          go-version: '^1.22.1'
+          go-version: "^1.22.1"
+
+      - name: Set up .NET
+        uses: actions/setup-dotnet@v4
+        with:
+          dotnet-version: "9.0.x"
+
+      - name: Build .NET LHM executable for Windows sensors
+        run: |
+          dotnet build -c Release ./beszel/internal/agent/lhm/beszel_lhm.csproj
+        shell: bash
 
       - name: GoReleaser beszel
         uses: goreleaser/goreleaser-action@v6
@@ -39,4 +49,6 @@ jobs:
           version: latest
           args: release --clean
         env:
-          GITHUB_TOKEN: ${{ secrets.TOKEN }}
+          GITHUB_TOKEN: ${{ secrets.TOKEN || secrets.GITHUB_TOKEN }}
+          WINGET_TOKEN: ${{ secrets.WINGET_TOKEN }}
+          IS_FORK: ${{ github.repository_owner != 'henrygd' }}

.github/workflows/vulncheck.yml

@@ -0,0 +1,33 @@
+# https://github.com/minio/minio/blob/master/.github/workflows/vulncheck.yml
+
+name: VulnCheck
+on:
+  pull_request:
+    branches:
+      - main
+
+  push:
+    branches:
+      - main
+
+permissions:
+  contents: read # to fetch code (actions/checkout)
+
+jobs:
+  vulncheck:
+    name: Analysis
+    runs-on: ubuntu-latest
+    steps:
+      - name: Check out code into the Go module directory
+        uses: actions/checkout@v4
+      - name: Set up Go
+        uses: actions/setup-go@v5
+        with:
+          go-version: 1.24.x
+          cached: false
+      - name: Get official govulncheck
+        run: go install golang.org/x/vuln/cmd/govulncheck@latest
+        shell: bash
+      - name: Run govulncheck
+        run: govulncheck -C ./beszel -show verbose ./...
+        shell: bash

.gitignore

@@ -17,3 +17,6 @@ beszel/build
 beszel/site/src/locales/**/*.ts
 *.bak
 __debug_*
+beszel/internal/agent/lhm/obj
+beszel/internal/agent/lhm/bin
+dockerfile_agent_dev

beszel/.goreleaser.yml

@@ -37,11 +37,26 @@ builds:
       - arm
       - mips64
       - riscv64
+      - mipsle
+      - mips
+      - ppc64le
+    gomips:
+      - hardfloat
+      - softfloat
     ignore:
       - goos: freebsd
         goarch: arm
       - goos: openbsd
         goarch: arm
+      - goos: linux
+        goarch: mips64
+        gomips: softfloat
+      - goos: linux
+        goarch: mipsle
+        gomips: hardfloat
+      - goos: linux
+        goarch: mips
+        gomips: hardfloat
       - goos: windows
         goarch: arm
       - goos: darwin
@@ -51,8 +66,8 @@ builds:
 
 archives:
   - id: beszel-agent
-    format: tar.gz
+    formats: [tar.gz]
-    builds:
+    ids:
       - beszel-agent
     name_template: >-
       {{ .Binary }}_
@@ -60,11 +75,11 @@ archives:
       {{- .Arch }}
     format_overrides:
       - goos: windows
-        format: zip
+        formats: [zip]
 
   - id: beszel
-    format: tar.gz
+    formats: [tar.gz]
-    builds:
+    ids:
       - beszel
     name_template: >-
       {{ .Binary }}_
@@ -83,13 +98,10 @@ nfpms:
       API access.
     maintainer: henrygd <hank@henrygd.me>
     section: net
-    builds:
+    ids:
       - beszel-agent
     formats:
       - deb
-    # don't think this is needed with CGO_ENABLED=0
-    # dependencies:
-    #   - libc6
     contents:
       - src: ../supplemental/debian/beszel-agent.service
         dst: lib/systemd/system/beszel-agent.service
@@ -120,9 +132,10 @@ scoops:
     repository:
       owner: henrygd
       name: beszel-scoops
-    homepage: 'https://beszel.dev'
+    homepage: "https://beszel.dev"
-    description: 'Agent for Beszel, a lightweight server monitoring platform.'
+    description: "Agent for Beszel, a lightweight server monitoring platform."
     license: MIT
+    skip_upload: "{{ if .Env.IS_FORK }}true{{ else }}auto{{ end }}"
 
 # # Needs choco installed, so doesn't build on linux / default gh workflow :(
 # chocolateys:
@@ -153,9 +166,10 @@ brews:
     repository:
       owner: henrygd
       name: homebrew-beszel
-    homepage: 'https://beszel.dev'
+    homepage: "https://beszel.dev"
-    description: 'Agent for Beszel, a lightweight server monitoring platform.'
+    description: "Agent for Beszel, a lightweight server monitoring platform."
     license: MIT
+    skip_upload: "{{ if .Env.IS_FORK }}true{{ else }}auto{{ end }}"
     extra_install: |
       (bin/"beszel-agent-launcher").write <<~EOS
         #!/bin/bash
@@ -172,6 +186,44 @@ brews:
       log_path "#{Dir.home}/.cache/beszel/beszel-agent.log"
       error_log_path "#{Dir.home}/.cache/beszel/beszel-agent.log"
       keep_alive true
+      restart_delay 5
+      process_type :background
+
+winget:
+  - ids: [beszel-agent]
+    name: beszel-agent
+    package_identifier: henrygd.beszel-agent
+    publisher: henrygd
+    license: MIT
+    license_url: "https://github.com/henrygd/beszel/blob/main/LICENSE"
+    copyright: "2025 henrygd"
+    homepage: "https://beszel.dev"
+    release_notes_url: "https://github.com/henrygd/beszel/releases/tag/v{{ .Version }}"
+    publisher_support_url: "https://github.com/henrygd/beszel/issues"
+    short_description: "Agent for Beszel, a lightweight server monitoring platform."
+    skip_upload: "{{ if .Env.IS_FORK }}true{{ else }}auto{{ end }}"
+    description: |
+      Beszel is a lightweight server monitoring platform that includes Docker
+      statistics, historical data, and alert functions. It has a friendly web
+      interface, simple configuration, and is ready to use out of the box.
+      It supports automatic backup, multi-user, OAuth authentication, and
+      API access.
+    tags:
+      - homelab
+      - monitoring
+      - self-hosted
+    repository:
+      owner: henrygd
+      name: beszel-winget
+      branch: henrygd.beszel-agent-{{ .Version }}
+      token: "{{ .Env.WINGET_TOKEN }}"
+      # pull_request:
+      #   enabled: true
+      #   draft: false
+      #   base:
+      #     owner: microsoft
+      #     name: winget-pkgs
+      #     branch: master
 
 release:
   draft: true
@@ -181,5 +233,5 @@ changelog:
   sort: asc
   filters:
     exclude:
-      - '^docs:'
+      - "^docs:"
-      - '^test:'
+      - "^test:"

beszel/Makefile

@@ -4,7 +4,10 @@ ARCH ?= $(shell go env GOARCH)
 # Skip building the web UI if true
 SKIP_WEB ?= false
 
-.PHONY: tidy build-agent build-hub build clean lint dev-server dev-agent dev-hub dev generate-locales
+# Set executable extension based on target OS
+EXE_EXT := $(if $(filter windows,$(OS)),.exe,)
+
+.PHONY: tidy build-agent build-hub build-hub-dev build clean lint dev-server dev-agent dev-hub dev generate-locales
 .DEFAULT_GOAL := build
 
 clean:
@@ -30,11 +33,29 @@ build-web-ui:
 		npm run --prefix ./site build; \
 	fi
 
-build-agent: tidy
+# Conditional .NET build - only for Windows
-	GOOS=$(OS) GOARCH=$(ARCH) go build -o ./build/beszel-agent_$(OS)_$(ARCH) -ldflags "-w -s" beszel/cmd/agent
+build-dotnet-conditional:
+	@if [ "$(OS)" = "windows" ]; then \
+		echo "Building .NET executable for Windows..."; \
+		if command -v dotnet >/dev/null 2>&1; then \
+			rm -rf ./internal/agent/lhm/bin; \
+			dotnet build -c Release ./internal/agent/lhm/beszel_lhm.csproj; \
+		else \
+			echo "Error: dotnet not found. Install .NET SDK to build Windows agent."; \
+			exit 1; \
+		fi; \
+	fi
+
+# Update build-agent to include conditional .NET build
+build-agent: tidy build-dotnet-conditional
+	GOOS=$(OS) GOARCH=$(ARCH) go build -o ./build/beszel-agent_$(OS)_$(ARCH)$(EXE_EXT) -ldflags "-w -s" beszel/cmd/agent
 
 build-hub: tidy $(if $(filter false,$(SKIP_WEB)),build-web-ui)
-	GOOS=$(OS) GOARCH=$(ARCH) go build -o ./build/beszel_$(OS)_$(ARCH) -ldflags "-w -s" beszel/cmd/hub
+	GOOS=$(OS) GOARCH=$(ARCH) go build -o ./build/beszel_$(OS)_$(ARCH)$(EXE_EXT) -ldflags "-w -s" beszel/cmd/hub
+
+build-hub-dev: tidy
+	mkdir -p ./site/dist && touch ./site/dist/index.html
+	GOOS=$(OS) GOARCH=$(ARCH) go build -tags development -o ./build/beszel-dev_$(OS)_$(ARCH)$(EXE_EXT) -ldflags "-w -s" beszel/cmd/hub
 
 build: build-agent build-hub
 
@@ -47,18 +68,18 @@ generate-locales:
 dev-server: generate-locales
 	cd ./site
 	@if command -v bun >/dev/null 2>&1; then \
-		cd ./site && bun run dev; \
+		cd ./site && bun run dev --host 0.0.0.0; \
 	else \
-		cd ./site && npm run dev; \
+		cd ./site && npm run dev --host 0.0.0.0; \
 	fi
 
 dev-hub: export ENV=dev
 dev-hub:
 	mkdir -p ./site/dist && touch ./site/dist/index.html
 	@if command -v entr >/dev/null 2>&1; then \
-		find ./cmd/hub/*.go ./internal/{alerts,hub,records,users}/*.go | entr -r -s "cd ./cmd/hub && go run . serve"; \
+		find ./cmd/hub/*.go ./internal/{alerts,hub,records,users}/*.go | entr -r -s "cd ./cmd/hub && go run -tags development . serve --http 0.0.0.0:8090"; \
 	else \
-		cd ./cmd/hub && go run . serve; \
+		cd ./cmd/hub && go run -tags development . serve --http 0.0.0.0:8090; \
 	fi
 
 dev-agent:
@@ -67,6 +88,15 @@ dev-agent:
 	else \
 		go run beszel/cmd/agent; \
 	fi
+	
+build-dotnet:
+	@if command -v dotnet >/dev/null 2>&1; then \
+		rm -rf ./internal/agent/lhm/bin; \
+		dotnet build -c Release ./internal/agent/lhm/beszel_lhm.csproj; \
+	else \
+		echo "dotnet not found"; \
+	fi
+
 
 # KEY="..." make -j dev
 dev: dev-server dev-hub dev-agent

beszel/cmd/agent/agent.go

@@ -3,11 +3,13 @@ package main
 import (
 	"beszel"
 	"beszel/internal/agent"
-	"flag"
+	"beszel/internal/agent/health"
 	"fmt"
 	"log"
 	"os"
+	"strings"
 
+	"github.com/spf13/pflag"
 	"golang.org/x/crypto/ssh"
 )
 
@@ -15,47 +17,26 @@ import (
 type cmdOptions struct {
 	key    string // key is the public key(s) for SSH authentication.
 	listen string // listen is the address or port to listen on.
+	// TODO: add hubURL and token
+	// hubURL string // hubURL is the URL of the hub to use.
+	// token  string // token is the token to use for authentication.
 }
 
 // parse parses the command line flags and populates the config struct.
 // It returns true if a subcommand was handled and the program should exit.
 func (opts *cmdOptions) parse() bool {
-	flag.StringVar(&opts.key, "key", "", "Public key(s) for SSH authentication")
-	flag.StringVar(&opts.listen, "listen", "", "Address or port to listen on")
-
-	flag.Usage = func() {
-		fmt.Printf("Usage: %s [command] [flags]\n", os.Args[0])
-		fmt.Println("\nCommands:")
-		fmt.Println("  health    Check if the agent is running")
-		fmt.Println("  help      Display this help message")
-		fmt.Println("  update    Update to the latest version")
-		fmt.Println("  version   Display the version")
-		fmt.Println("\nFlags:")
-		flag.PrintDefaults()
-	}
-
 	subcommand := ""
 	if len(os.Args) > 1 {
 		subcommand = os.Args[1]
 	}
 
+	// Subcommands that don't require any pflag parsing
 	switch subcommand {
 	case "-v", "version":
 		fmt.Println(beszel.AppName+"-agent", beszel.Version)
 		return true
-	case "help":
-		flag.Usage()
-		return true
-	case "update":
-		agent.Update()
-		return true
 	case "health":
-		// for health, we need to parse flags first to get the listen address
+		err := health.Check()
-		args := append(os.Args[2:], subcommand)
-		flag.CommandLine.Parse(args)
-		addr := opts.getAddress()
-		network := agent.GetNetwork(addr)
-		err := agent.Health(addr, network)
 		if err != nil {
 			log.Fatal(err)
 		}
@@ -63,7 +44,57 @@ func (opts *cmdOptions) parse() bool {
 		return true
 	}
 
-	flag.Parse()
+	// pflag.CommandLine.ParseErrorsWhitelist.UnknownFlags = true
+	pflag.StringVarP(&opts.key, "key", "k", "", "Public key(s) for SSH authentication")
+	pflag.StringVarP(&opts.listen, "listen", "l", "", "Address or port to listen on")
+	// pflag.StringVarP(&opts.hubURL, "hub-url", "u", "", "URL of the hub to use")
+	// pflag.StringVarP(&opts.token, "token", "t", "", "Token to use for authentication")
+	chinaMirrors := pflag.BoolP("china-mirrors", "c", false, "Use mirror for update (gh.beszel.dev) instead of GitHub")
+	help := pflag.BoolP("help", "h", false, "Show this help message")
+
+	// Convert old single-dash long flags to double-dash for backward compatibility
+	flagsToConvert := []string{"key", "listen"}
+	for i, arg := range os.Args {
+		for _, flag := range flagsToConvert {
+			singleDash := "-" + flag
+			doubleDash := "--" + flag
+			if arg == singleDash {
+				os.Args[i] = doubleDash
+				break
+			} else if strings.HasPrefix(arg, singleDash+"=") {
+				os.Args[i] = doubleDash + arg[len(singleDash):]
+				break
+			}
+		}
+	}
+
+	pflag.Usage = func() {
+		builder := strings.Builder{}
+		builder.WriteString("Usage: ")
+		builder.WriteString(os.Args[0])
+		builder.WriteString(" [command] [flags]\n")
+		builder.WriteString("\nCommands:\n")
+		builder.WriteString("  health    Check if the agent is running\n")
+		// builder.WriteString("  help      Display this help message\n")
+		builder.WriteString("  update    Update to the latest version\n")
+		builder.WriteString("\nFlags:\n")
+		fmt.Print(builder.String())
+		pflag.PrintDefaults()
+	}
+
+	// Parse all arguments with pflag
+	pflag.Parse()
+
+	// Must run after pflag.Parse()
+	switch {
+	case *help || subcommand == "help":
+		pflag.Usage()
+		return true
+	case subcommand == "update":
+		agent.Update(*chinaMirrors)
+		return true
+	}
+
 	return false
 }
 
@@ -115,8 +146,12 @@ func main() {
 	serverConfig.Addr = addr
 	serverConfig.Network = agent.GetNetwork(addr)
 
-	agent := agent.NewAgent()
+	a, err := agent.NewAgent()
-	if err := agent.StartServer(serverConfig); err != nil {
+	if err != nil {
-		log.Fatal("Failed to start server:", err)
+		log.Fatal("Failed to create agent: ", err)
+	}
+
+	if err := a.Start(serverConfig); err != nil {
+		log.Fatal("Failed to start server: ", err)
 	}
 }

beszel/cmd/agent/agent_test.go

@@ -3,11 +3,11 @@ package main
 import (
 	"beszel/internal/agent"
 	"crypto/ed25519"
-	"flag"
 	"os"
 	"path/filepath"
 	"testing"
 
+	"github.com/spf13/pflag"
 	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"
 	"golang.org/x/crypto/ssh"
@@ -245,7 +245,7 @@ func TestParseFlags(t *testing.T) {
 	oldArgs := os.Args
 	defer func() {
 		os.Args = oldArgs
-		flag.CommandLine = flag.NewFlagSet(os.Args[0], flag.ExitOnError)
+		pflag.CommandLine = pflag.NewFlagSet(os.Args[0], pflag.ExitOnError)
 	}()
 
 	tests := []struct {
@@ -269,6 +269,22 @@ func TestParseFlags(t *testing.T) {
 				listen: "",
 			},
 		},
+		{
+			name: "key flag double dash",
+			args: []string{"cmd", "--key", "testkey"},
+			expected: cmdOptions{
+				key:    "testkey",
+				listen: "",
+			},
+		},
+		{
+			name: "key flag short",
+			args: []string{"cmd", "-k", "testkey"},
+			expected: cmdOptions{
+				key:    "testkey",
+				listen: "",
+			},
+		},
 		{
 			name: "addr flag only",
 			args: []string{"cmd", "-listen", ":8080"},
@@ -277,6 +293,22 @@ func TestParseFlags(t *testing.T) {
 				listen: ":8080",
 			},
 		},
+		{
+			name: "addr flag double dash",
+			args: []string{"cmd", "--listen", ":8080"},
+			expected: cmdOptions{
+				key:    "",
+				listen: ":8080",
+			},
+		},
+		{
+			name: "addr flag short",
+			args: []string{"cmd", "-l", ":8080"},
+			expected: cmdOptions{
+				key:    "",
+				listen: ":8080",
+			},
+		},
 		{
 			name: "both flags",
 			args: []string{"cmd", "-key", "testkey", "-listen", ":8080"},
@@ -290,12 +322,12 @@ func TestParseFlags(t *testing.T) {
 	for _, tt := range tests {
 		t.Run(tt.name, func(t *testing.T) {
 			// Reset flags for each test
-			flag.CommandLine = flag.NewFlagSet(tt.args[0], flag.ExitOnError)
+			pflag.CommandLine = pflag.NewFlagSet(tt.args[0], pflag.ExitOnError)
 			os.Args = tt.args
 
 			var opts cmdOptions
 			opts.parse()
-			flag.Parse()
+			pflag.Parse()
 
 			assert.Equal(t, tt.expected, opts)
 		})

beszel/cmd/hub/hub.go

@@ -45,11 +45,13 @@ func getBaseApp() *pocketbase.PocketBase {
 	baseApp.RootCmd.Use = beszel.AppName
 	baseApp.RootCmd.Short = ""
 	// add update command
-	baseApp.RootCmd.AddCommand(&cobra.Command{
+	updateCmd := &cobra.Command{
 		Use:   "update",
 		Short: "Update " + beszel.AppName + " to the latest version",
 		Run:   hub.Update,
-	})
+	}
+	updateCmd.Flags().Bool("china-mirrors", false, "Use mirror (gh.beszel.dev) instead of GitHub")
+	baseApp.RootCmd.AddCommand(updateCmd)
 	// add health command
 	baseApp.RootCmd.AddCommand(newHealthCmd())
 

beszel/dockerfile_agent

@@ -0,0 +1,26 @@
+FROM --platform=$BUILDPLATFORM golang:alpine AS builder
+
+WORKDIR /app
+
+COPY go.mod go.sum ./
+# RUN go mod download
+COPY *.go ./
+COPY cmd ./cmd
+COPY internal ./internal
+
+# Build
+ARG TARGETOS TARGETARCH
+RUN CGO_ENABLED=0 GOGC=75 GOOS=$TARGETOS GOARCH=$TARGETARCH go build -ldflags "-w -s" -o /agent ./cmd/agent
+
+RUN rm -rf /tmp/*
+
+# --------------------------
+# Final image: default scratch-based agent
+# --------------------------
+FROM scratch
+COPY --from=builder /agent /agent
+
+# this is so we don't need to create the /tmp directory in the scratch container
+COPY --from=builder /tmp /tmp
+
+ENTRYPOINT ["/agent"]

beszel/dockerfile_agent_nvidia

@@ -12,9 +12,10 @@ COPY internal ./internal
 ARG TARGETOS TARGETARCH
 RUN CGO_ENABLED=0 GOGC=75 GOOS=$TARGETOS GOARCH=$TARGETARCH go build -ldflags "-w -s" -o /agent ./cmd/agent
 
-# ? -------------------------
+# --------------------------
-FROM scratch
+# Final image: GPU-enabled agent with nvidia-smi
-
+# --------------------------
+FROM nvidia/cuda:12.2.2-base-ubuntu22.04
 COPY --from=builder /agent /agent
 
-ENTRYPOINT ["/agent"]
+ENTRYPOINT ["/agent"]

beszel/go.mod

@@ -1,24 +1,27 @@
 module beszel
 
-go 1.24.2
+go 1.24.4
 
 // lock shoutrrr to specific version to allow review before updating
 replace github.com/nicholas-fedor/shoutrrr => github.com/nicholas-fedor/shoutrrr v0.8.8
 
 require (
 	github.com/blang/semver v3.5.1+incompatible
+	github.com/distatus/battery v0.11.0
+	github.com/fxamacker/cbor/v2 v2.9.0
 	github.com/gliderlabs/ssh v0.3.8
-	github.com/goccy/go-json v0.10.5
+	github.com/google/uuid v1.6.0
-	github.com/nicholas-fedor/shoutrrr v0.8.8
+	github.com/lxzan/gws v1.8.9
+	github.com/nicholas-fedor/shoutrrr v0.8.17
 	github.com/pocketbase/dbx v1.11.0
-	github.com/pocketbase/pocketbase v0.27.1
+	github.com/pocketbase/pocketbase v0.29.3
-	github.com/rhysd/go-github-selfupdate v1.2.3
+	github.com/shirou/gopsutil/v4 v4.25.6
-	github.com/shirou/gopsutil/v4 v4.25.3
+	github.com/spf13/cast v1.9.2
-	github.com/spf13/cast v1.7.1
 	github.com/spf13/cobra v1.9.1
-	github.com/stretchr/testify v1.10.0
+	github.com/spf13/pflag v1.0.7
-	golang.org/x/crypto v0.37.0
+	github.com/stretchr/testify v1.11.0
-	golang.org/x/exp v0.0.0-20250408133849-7e4ce0ab07d0
+	golang.org/x/crypto v0.41.0
+	golang.org/x/exp v0.0.0-20250819193227-8b4c13bb791b
 	gopkg.in/yaml.v3 v3.0.1
 )
 
@@ -27,42 +30,40 @@ require (
 	github.com/asaskevich/govalidator v0.0.0-20230301143203-a9d515a09cc2 // indirect
 	github.com/davecgh/go-spew v1.1.2-0.20180830191138-d8f796af33cc // indirect
 	github.com/disintegration/imaging v1.6.2 // indirect
+	github.com/dolthub/maphash v0.1.0 // indirect
 	github.com/domodwyer/mailyak/v3 v3.6.2 // indirect
 	github.com/dustin/go-humanize v1.0.1 // indirect
-	github.com/ebitengine/purego v0.8.2 // indirect
+	github.com/ebitengine/purego v0.8.4 // indirect
 	github.com/fatih/color v1.18.0 // indirect
 	github.com/gabriel-vasile/mimetype v1.4.9 // indirect
 	github.com/ganigeorgiev/fexpr v0.5.0 // indirect
 	github.com/go-ole/go-ole v1.3.0 // indirect
 	github.com/go-ozzo/ozzo-validation/v4 v4.3.0 // indirect
 	github.com/go-sql-driver/mysql v1.9.1 // indirect
-	github.com/golang-jwt/jwt/v5 v5.2.2 // indirect
+	github.com/golang-jwt/jwt/v5 v5.3.0 // indirect
-	github.com/google/go-github/v30 v30.1.0 // indirect
-	github.com/google/go-querystring v1.1.0 // indirect
-	github.com/google/uuid v1.6.0 // indirect
-	github.com/inconshreveable/go-update v0.0.0-20160112193335-8152e7eb6ccf // indirect
 	github.com/inconshreveable/mousetrap v1.1.0 // indirect
-	github.com/lufia/plan9stats v0.0.0-20250317134145-8bc96cf8fc35 // indirect
+	github.com/klauspost/compress v1.18.0 // indirect
+	github.com/lufia/plan9stats v0.0.0-20250821153705-5981dea3221d // indirect
 	github.com/mattn/go-colorable v0.1.14 // indirect
 	github.com/mattn/go-isatty v0.0.20 // indirect
 	github.com/ncruces/go-strftime v0.1.9 // indirect
 	github.com/pmezard/go-difflib v1.0.1-0.20181226105442-5d4384ee4fb2 // indirect
 	github.com/power-devops/perfstat v0.0.0-20240221224432-82ca36839d55 // indirect
 	github.com/remyoudompheng/bigfft v0.0.0-20230129092748-24d4a6f8daec // indirect
-	github.com/spf13/pflag v1.0.6 // indirect
-	github.com/tcnksm/go-gitconfig v0.1.2 // indirect
 	github.com/tklauser/go-sysconf v0.3.15 // indirect
 	github.com/tklauser/numcpus v0.10.0 // indirect
-	github.com/ulikunitz/xz v0.5.12 // indirect
+	github.com/x448/float16 v0.8.4 // indirect
 	github.com/yusufpapurcu/wmi v1.2.4 // indirect
-	golang.org/x/image v0.26.0 // indirect
+	golang.org/x/image v0.30.0 // indirect
-	golang.org/x/net v0.39.0 // indirect
+	golang.org/x/net v0.43.0 // indirect
-	golang.org/x/oauth2 v0.29.0 // indirect
+	golang.org/x/oauth2 v0.30.0 // indirect
-	golang.org/x/sync v0.13.0 // indirect
+	golang.org/x/sync v0.16.0 // indirect
-	golang.org/x/sys v0.32.0 // indirect
+	golang.org/x/sys v0.35.0 // indirect
-	golang.org/x/text v0.24.0 // indirect
+	golang.org/x/text v0.28.0 // indirect
-	modernc.org/libc v1.64.0 // indirect
+	gopkg.in/check.v1 v1.0.0-20180628173108-788fd7840127 // indirect
+	howett.net/plist v1.0.1 // indirect
+	modernc.org/libc v1.66.3 // indirect
 	modernc.org/mathutil v1.7.1 // indirect
-	modernc.org/memory v1.10.0 // indirect
+	modernc.org/memory v1.11.0 // indirect
-	modernc.org/sqlite v1.37.0 // indirect
+	modernc.org/sqlite v1.38.2 // indirect
 )

beszel/go.sum

@@ -13,17 +13,22 @@ github.com/davecgh/go-spew v1.1.2-0.20180830191138-d8f796af33cc h1:U9qPSI2PIWSS1
 github.com/davecgh/go-spew v1.1.2-0.20180830191138-d8f796af33cc/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
 github.com/disintegration/imaging v1.6.2 h1:w1LecBlG2Lnp8B3jk5zSuNqd7b4DXhcjwek1ei82L+c=
 github.com/disintegration/imaging v1.6.2/go.mod h1:44/5580QXChDfwIclfc/PCwrr44amcmDAg8hxG0Ewe4=
+github.com/distatus/battery v0.11.0 h1:KJk89gz90Iq/wJtbjjM9yUzBXV+ASV/EG2WOOL7N8lc=
+github.com/distatus/battery v0.11.0/go.mod h1:KmVkE8A8hpIX4T78QRdMktYpEp35QfOL8A8dwZBxq2k=
+github.com/dolthub/maphash v0.1.0 h1:bsQ7JsF4FkkWyrP3oCnFJgrCUAFbFf3kOl4L/QxPDyQ=
+github.com/dolthub/maphash v0.1.0/go.mod h1:gkg4Ch4CdCDu5h6PMriVLawB7koZ+5ijb9puGMV50a4=
 github.com/domodwyer/mailyak/v3 v3.6.2 h1:x3tGMsyFhTCaxp6ycgR0FE/bu5QiNp+hetUuCOBXMn8=
 github.com/domodwyer/mailyak/v3 v3.6.2/go.mod h1:lOm/u9CyCVWHeaAmHIdF4RiKVxKUT/H5XX10lIKAL6c=
 github.com/dustin/go-humanize v1.0.1 h1:GzkhY7T5VNhEkwH0PVJgjz+fX1rhBrR7pRT3mDkpeCY=
 github.com/dustin/go-humanize v1.0.1/go.mod h1:Mu1zIs6XwVuF/gI1OepvI0qD18qycQx+mFykh5fBlto=
-github.com/ebitengine/purego v0.8.2 h1:jPPGWs2sZ1UgOSgD2bClL0MJIqu58nOmIcBuXr62z1I=
+github.com/ebitengine/purego v0.8.4 h1:CF7LEKg5FFOsASUj0+QwaXf8Ht6TlFxg09+S9wz0omw=
-github.com/ebitengine/purego v0.8.2/go.mod h1:iIjxzd6CiRiOG0UyXP+V1+jWqUXVjPKLAI0mRfJZTmQ=
+github.com/ebitengine/purego v0.8.4/go.mod h1:iIjxzd6CiRiOG0UyXP+V1+jWqUXVjPKLAI0mRfJZTmQ=
 github.com/fatih/color v1.18.0 h1:S8gINlzdQ840/4pfAwic/ZE0djQEH3wM94VfqLTZcOM=
 github.com/fatih/color v1.18.0/go.mod h1:4FelSpRwEGDpQ12mAdzqdOukCy4u8WUtOY6lkT/6HfU=
 github.com/frankban/quicktest v1.14.6 h1:7Xjx+VpznH+oBnejlPUj8oUpdxnVs4f8XU8WnHkI4W8=
 github.com/frankban/quicktest v1.14.6/go.mod h1:4ptaffx2x8+WTWXmUCuVU6aPUX1/Mz7zb5vbUoiM6w0=
-github.com/fsnotify/fsnotify v1.4.7/go.mod h1:jwhsz4b93w/PPRr/qN1Yymfu8t87LnFCMoQvtojpjFo=
+github.com/fxamacker/cbor/v2 v2.9.0 h1:NpKPmjDBgUfBms6tr6JZkTHtfFGcMKsw3eGcmD/sapM=
+github.com/fxamacker/cbor/v2 v2.9.0/go.mod h1:vM4b+DJCtHn+zz7h3FFp/hDAI9WNWCsZj23V5ytsSxQ=
 github.com/gabriel-vasile/mimetype v1.4.9 h1:5k+WDwEsD9eTLL8Tz3L0VnmVh9QxGjRmjBvAG7U/oYY=
 github.com/gabriel-vasile/mimetype v1.4.9/go.mod h1:WnSQhFKJuBlRyLiKohA/2DtIlPFAbguNaG7QCHcyGok=
 github.com/ganigeorgiev/fexpr v0.5.0 h1:XA9JxtTE/Xm+g/JFI6RfZEHSiQlk+1glLvRK1Lpv/Tk=
@@ -42,41 +47,30 @@ github.com/go-sql-driver/mysql v1.9.1 h1:FrjNGn/BsJQjVRuSa8CBrM5BWA9BWoXXat3KrtS
 github.com/go-sql-driver/mysql v1.9.1/go.mod h1:qn46aNg1333BRMNU69Lq93t8du/dwxI64Gl8i5p1WMU=
 github.com/go-task/slim-sprig/v3 v3.0.0 h1:sUs3vkvUymDpBKi3qH1YSqBQk9+9D/8M2mN1vB6EwHI=
 github.com/go-task/slim-sprig/v3 v3.0.0/go.mod h1:W848ghGpv3Qj3dhTPRyJypKRiqCdHZiAzKg9hl15HA8=
-github.com/goccy/go-json v0.10.5 h1:Fq85nIqj+gXn/S5ahsiTlK3TmC85qgirsdTP/+DeaC4=
+github.com/golang-jwt/jwt/v5 v5.3.0 h1:pv4AsKCKKZuqlgs5sUmn4x8UlGa0kEVt/puTpKx9vvo=
-github.com/goccy/go-json v0.10.5/go.mod h1:oq7eo15ShAhp70Anwd5lgX2pLfOS3QCiwU/PULtXL6M=
+github.com/golang-jwt/jwt/v5 v5.3.0/go.mod h1:fxCRLWMO43lRc8nhHWY6LGqRcf+1gQWArsqaEUEa5bE=
-github.com/golang-jwt/jwt/v5 v5.2.2 h1:Rl4B7itRWVtYIHFrSNd7vhTiz9UpLdi6gZhZ3wEeDy8=
-github.com/golang-jwt/jwt/v5 v5.2.2/go.mod h1:pqrtFR0X4osieyHYxtmOUWsAWrfe1Q5UVIyoH402zdk=
-github.com/golang/protobuf v1.2.0/go.mod h1:6lQm79b+lXiMfvg/cZm0SGofjICqVBUtrP5yJMmIC1U=
 github.com/golang/protobuf v1.3.1/go.mod h1:6lQm79b+lXiMfvg/cZm0SGofjICqVBUtrP5yJMmIC1U=
-github.com/golang/protobuf v1.3.2/go.mod h1:6lQm79b+lXiMfvg/cZm0SGofjICqVBUtrP5yJMmIC1U=
-github.com/google/go-cmp v0.5.2/go.mod h1:v8dTdLbMG2kIc/vJvl+f65V22dbkXbowE6jgT/gNBxE=
 github.com/google/go-cmp v0.7.0 h1:wk8382ETsv4JYUZwIsn6YpYiWiBsYLSJiTsyBybVuN8=
 github.com/google/go-cmp v0.7.0/go.mod h1:pXiqmnSA92OHEEa9HXL2W4E7lf9JzCmGVUdgjX3N/iU=
-github.com/google/go-github/v30 v30.1.0 h1:VLDx+UolQICEOKu2m4uAoMti1SxuEBAl7RSEG16L+Oo=
-github.com/google/go-github/v30 v30.1.0/go.mod h1:n8jBpHl45a/rlBUtRJMOG4GhNADUQFEufcolZ95JfU8=
-github.com/google/go-querystring v1.0.0/go.mod h1:odCYkC5MyYFN7vkCjXpyrEuKhc/BUO6wN/zVPAxq5ck=
-github.com/google/go-querystring v1.1.0 h1:AnCroh3fv4ZBgVIf1Iwtovgjaw/GiKJo8M8yD/fhyJ8=
-github.com/google/go-querystring v1.1.0/go.mod h1:Kcdr2DB4koayq7X8pmAG4sNG59So17icRSOU623lUBU=
 github.com/google/pprof v0.0.0-20250403155104-27863c87afa6 h1:BHT72Gu3keYf3ZEu2J0b1vyeLSOYI8bm5wbJM/8yDe8=
 github.com/google/pprof v0.0.0-20250403155104-27863c87afa6/go.mod h1:boTsfXsheKC2y+lKOCMpSfarhxDeIzfZG1jqGcPl3cA=
 github.com/google/uuid v1.6.0 h1:NIvaJDMOsjHA8n1jAhLSgzrAzy1Hgr+hNrb57e+94F0=
 github.com/google/uuid v1.6.0/go.mod h1:TIyPZe4MgqvfeYDBFedMoGGpEw/LqOeaOT+nhxU+yHo=
-github.com/hpcloud/tail v1.0.0/go.mod h1:ab1qPbhIpdTxEkNHXyeSf5vhxWSCs/tWer42PpOxQnU=
-github.com/inconshreveable/go-update v0.0.0-20160112193335-8152e7eb6ccf h1:WfD7VjIE6z8dIvMsI4/s+1qr5EL+zoIGev1BQj1eoJ8=
-github.com/inconshreveable/go-update v0.0.0-20160112193335-8152e7eb6ccf/go.mod h1:hyb9oH7vZsitZCiBt0ZvifOrB+qc8PS5IiilCIb87rg=
 github.com/inconshreveable/mousetrap v1.1.0 h1:wN+x4NVGpMsO7ErUn/mUI3vEoE6Jt13X2s0bqwp9tc8=
 github.com/inconshreveable/mousetrap v1.1.0/go.mod h1:vpF70FUmC8bwa3OWnCshd2FqLfsEA9PFc4w1p2J65bw=
 github.com/jarcoal/httpmock v1.4.0 h1:BvhqnH0JAYbNudL2GMJKgOHe2CtKlzJ/5rWKyp+hc2k=
 github.com/jarcoal/httpmock v1.4.0/go.mod h1:ftW1xULwo+j0R0JJkJIIi7UKigZUXCLLanykgjwBXL0=
-github.com/kr/pretty v0.1.0/go.mod h1:dAy3ld7l9f0ibDNOQOHHMYYIIbhfbHSm3C4ZsoJORNo=
+github.com/jessevdk/go-flags v1.4.0/go.mod h1:4FA24M0QyGHXBuZZK/XkWh8h0e1EYbRYJSGM75WSRxI=
+github.com/klauspost/compress v1.18.0 h1:c/Cqfb0r+Yi+JtIEq73FWXVkRonBlf0CRNYc8Zttxdo=
+github.com/klauspost/compress v1.18.0/go.mod h1:2Pp+KzxcywXVXMr50+X0Q/Lsb43OQHYWRCY2AiWywWQ=
 github.com/kr/pretty v0.3.1 h1:flRD4NNwYAUpkphVc1HcthR4KEIFJ65n8Mw5qdRn3LE=
 github.com/kr/pretty v0.3.1/go.mod h1:hoEshYVHaxMs3cyo3Yncou5ZscifuDolrwPKZanG3xk=
-github.com/kr/pty v1.1.1/go.mod h1:pFQYn66WHrOpPYNljwOMqo10TkYh1fy3cYio2l3bCsQ=
-github.com/kr/text v0.1.0/go.mod h1:4Jbv+DJW3UT/LiOwJeYQe1efqtUx/iVham/4vfdArNI=
 github.com/kr/text v0.2.0 h1:5Nx0Ya0ZqY2ygV366QzturHI13Jq95ApcVaJBhpS+AY=
 github.com/kr/text v0.2.0/go.mod h1:eLer722TekiGuMkidMxC/pM04lWEeraHUUmBw8l2grE=
-github.com/lufia/plan9stats v0.0.0-20250317134145-8bc96cf8fc35 h1:PpXWgLPs+Fqr325bN2FD2ISlRRztXibcX6e8f5FR5Dc=
+github.com/lufia/plan9stats v0.0.0-20250821153705-5981dea3221d h1:vFzYZc8yji+9DmNRhpEbs8VBK4CgV/DPfGzeVJSSp/8=
-github.com/lufia/plan9stats v0.0.0-20250317134145-8bc96cf8fc35/go.mod h1:autxFIvghDt3jPTLoqZ9OZ7s9qTGNAWmYCjVFWPX/zg=
+github.com/lufia/plan9stats v0.0.0-20250821153705-5981dea3221d/go.mod h1:autxFIvghDt3jPTLoqZ9OZ7s9qTGNAWmYCjVFWPX/zg=
+github.com/lxzan/gws v1.8.9 h1:VU3SGUeWlQrEwfUSfokcZep8mdg/BrUF+y73YYshdBM=
+github.com/lxzan/gws v1.8.9/go.mod h1:d9yHaR1eDTBHagQC6KY7ycUOaz5KWeqQtP3xu7aMK8Y=
 github.com/mattn/go-colorable v0.1.14 h1:9A9LHSqF/7dyVVX6g0U9cwm9pG3kP9gSzcuIPHPsaIE=
 github.com/mattn/go-colorable v0.1.14/go.mod h1:6LmQG8QLFO4G5z1gPvYEzlUgJ2wF+stgPZH1UqBm1s8=
 github.com/mattn/go-isatty v0.0.20 h1:xfD0iDuEKnDkl03q4limB+vH+GxLEtL/jb4xVJSWWEY=
@@ -85,11 +79,8 @@ github.com/ncruces/go-strftime v0.1.9 h1:bY0MQC28UADQmHmaF5dgpLmImcShSi2kHU9XLdh
 github.com/ncruces/go-strftime v0.1.9/go.mod h1:Fwc5htZGVVkseilnfgOVb9mKy6w1naJmn9CehxcKcls=
 github.com/nicholas-fedor/shoutrrr v0.8.8 h1:F/oyoatWK5cbHPPgkjRZrA0262TP7KWuUQz9KskRtR8=
 github.com/nicholas-fedor/shoutrrr v0.8.8/go.mod h1:T30Y+eoZFEjDk4HtOItcHQioZSOe3Z6a6aNfSz6jc5c=
-github.com/onsi/ginkgo v1.6.0 h1:Ix8l273rp3QzYgXSR+c8d1fTG7UPgYkOSELPhiY/YGw=
-github.com/onsi/ginkgo v1.6.0/go.mod h1:lLunBs/Ym6LB5Z9jYTR76FiuTmxDTDusOGeTQH+WWjE=
 github.com/onsi/ginkgo/v2 v2.23.4 h1:ktYTpKJAVZnDT4VjxSbiBenUjmlL/5QkBEocaWXiQus=
 github.com/onsi/ginkgo/v2 v2.23.4/go.mod h1:Bt66ApGPBFzHyR+JO10Zbt0Gsp4uWxu5mIOTusL46e8=
-github.com/onsi/gomega v1.4.2/go.mod h1:ex+gbHU/CVuBBDIJjb2X0qEXbFg53c61hWP/1CpauHY=
 github.com/onsi/gomega v1.37.0 h1:CdEG8g0S133B4OswTDC/5XPSzE1OeP29QOioj2PID2Y=
 github.com/onsi/gomega v1.37.0/go.mod h1:8D9+Txp43QWKhM24yyOBEdpkzN8FvJyAwecBgsU4KU0=
 github.com/pmezard/go-difflib v1.0.0/go.mod h1:iKH77koFhYxTK1pcRnkKkqfTogsbg7gZNVY4sRDYZ/4=
@@ -97,122 +88,105 @@ github.com/pmezard/go-difflib v1.0.1-0.20181226105442-5d4384ee4fb2 h1:Jamvg5psRI
 github.com/pmezard/go-difflib v1.0.1-0.20181226105442-5d4384ee4fb2/go.mod h1:iKH77koFhYxTK1pcRnkKkqfTogsbg7gZNVY4sRDYZ/4=
 github.com/pocketbase/dbx v1.11.0 h1:LpZezioMfT3K4tLrqA55wWFw1EtH1pM4tzSVa7kgszU=
 github.com/pocketbase/dbx v1.11.0/go.mod h1:xXRCIAKTHMgUCyCKZm55pUOdvFziJjQfXaWKhu2vhMs=
-github.com/pocketbase/pocketbase v0.27.1 h1:KGCsS8idUVTC5QHxTj91qHDhIXOb5Yb50wwHhNvJRTQ=
+github.com/pocketbase/pocketbase v0.29.3 h1:Mj8o5awsbVJIdIoTuQNhfC2oL/c4aImQ3RyfFZlzFVg=
-github.com/pocketbase/pocketbase v0.27.1/go.mod h1:aTpwwloVJzeJ7MlwTRrbI/x62QNR2/kkCrovmyrXpqs=
+github.com/pocketbase/pocketbase v0.29.3/go.mod h1:oGpT67LObxCFK4V2fSL7J9YnPbBnnshOpJ5v3zcneww=
 github.com/power-devops/perfstat v0.0.0-20240221224432-82ca36839d55 h1:o4JXh1EVt9k/+g42oCprj/FisM4qX9L3sZB3upGN2ZU=
 github.com/power-devops/perfstat v0.0.0-20240221224432-82ca36839d55/go.mod h1:OmDBASR4679mdNQnz2pUhc2G8CO2JrUAVFDRBDP/hJE=
 github.com/remyoudompheng/bigfft v0.0.0-20230129092748-24d4a6f8daec h1:W09IVJc94icq4NjY3clb7Lk8O1qJ8BdBEF8z0ibU0rE=
 github.com/remyoudompheng/bigfft v0.0.0-20230129092748-24d4a6f8daec/go.mod h1:qqbHyh8v60DhA7CoWK5oRCqLrMHRGoxYCSS9EjAz6Eo=
-github.com/rhysd/go-github-selfupdate v1.2.3 h1:iaa+J202f+Nc+A8zi75uccC8Wg3omaM7HDeimXA22Ag=
-github.com/rhysd/go-github-selfupdate v1.2.3/go.mod h1:mp/N8zj6jFfBQy/XMYoWsmfzxazpPAODuqarmPDe2Rg=
 github.com/rogpeppe/go-internal v1.9.0 h1:73kH8U+JUqXU8lRuOHeVHaa/SZPifC7BkcraZVejAe8=
 github.com/rogpeppe/go-internal v1.9.0/go.mod h1:WtVeX8xhTBvf0smdhujwtBcq4Qrzq/fJaraNFVN+nFs=
 github.com/russross/blackfriday/v2 v2.1.0/go.mod h1:+Rmxgy9KzJVeS9/2gXHxylqXiyQDYRxCVz55jmeOWTM=
-github.com/shirou/gopsutil/v4 v4.25.3 h1:SeA68lsu8gLggyMbmCn8cmp97V1TI9ld9sVzAUcKcKE=
+github.com/shirou/gopsutil/v4 v4.25.6 h1:kLysI2JsKorfaFPcYmcJqbzROzsBWEOAtw6A7dIfqXs=
-github.com/shirou/gopsutil/v4 v4.25.3/go.mod h1:xbuxyoZj+UsgnZrENu3lQivsngRR5BdjbJwf2fv4szA=
+github.com/shirou/gopsutil/v4 v4.25.6/go.mod h1:PfybzyydfZcN+JMMjkF6Zb8Mq1A/VcogFFg7hj50W9c=
-github.com/spf13/cast v1.7.1 h1:cuNEagBQEHWN1FnbGEjCXL2szYEXqfJPbP2HNUaca9Y=
+github.com/spf13/cast v1.9.2 h1:SsGfm7M8QOFtEzumm7UZrZdLLquNdzFYfIbEXntcFbE=
-github.com/spf13/cast v1.7.1/go.mod h1:ancEpBxwJDODSW/UG4rDrAqiKolqNNh2DX3mk86cAdo=
+github.com/spf13/cast v1.9.2/go.mod h1:jNfB8QC9IA6ZuY2ZjDp0KtFO2LZZlg4S/7bzP6qqeHo=
 github.com/spf13/cobra v1.9.1 h1:CXSaggrXdbHK9CF+8ywj8Amf7PBRmPCOJugH954Nnlo=
 github.com/spf13/cobra v1.9.1/go.mod h1:nDyEzZ8ogv936Cinf6g1RU9MRY64Ir93oCnqb9wxYW0=
-github.com/spf13/pflag v1.0.6 h1:jFzHGLGAlb3ruxLB8MhbI6A8+AQX/2eW4qeyNZXNp2o=
 github.com/spf13/pflag v1.0.6/go.mod h1:McXfInJRrz4CZXVZOBLb0bTZqETkiAhM9Iw0y3An2Bg=
+github.com/spf13/pflag v1.0.7 h1:vN6T9TfwStFPFM5XzjsvmzZkLuaLX+HS+0SeFLRgU6M=
+github.com/spf13/pflag v1.0.7/go.mod h1:McXfInJRrz4CZXVZOBLb0bTZqETkiAhM9Iw0y3An2Bg=
 github.com/stretchr/objx v0.1.0/go.mod h1:HFkY916IF+rwdDfMAkV7OtwuqBVzrE8GR6GFx+wExME=
 github.com/stretchr/testify v1.4.0/go.mod h1:j7eGeouHqKxXV5pUuKE4zz7dFj8WfuZ+81PSLYec5m4=
-github.com/stretchr/testify v1.10.0 h1:Xv5erBjTwe/5IxqUQTdXv5kgmIvbHo3QQyRwhJsOfJA=
+github.com/stretchr/testify v1.11.0 h1:ib4sjIrwZKxE5u/Japgo/7SJV3PvgjGiRNAvTVGqQl8=
-github.com/stretchr/testify v1.10.0/go.mod h1:r2ic/lqez/lEtzL7wO/rwa5dbSLXVDPFyf8C91i36aY=
+github.com/stretchr/testify v1.11.0/go.mod h1:wZwfW3scLgRK+23gO65QZefKpKQRnfz6sD981Nm4B6U=
-github.com/tcnksm/go-gitconfig v0.1.2 h1:iiDhRitByXAEyjgBqsKi9QU4o2TNtv9kPP3RgPgXBPw=
-github.com/tcnksm/go-gitconfig v0.1.2/go.mod h1:/8EhP4H7oJZdIPyT+/UIsG87kTzrzM4UsLGSItWYCpE=
 github.com/tklauser/go-sysconf v0.3.15 h1:VE89k0criAymJ/Os65CSn1IXaol+1wrsFHEB8Ol49K4=
 github.com/tklauser/go-sysconf v0.3.15/go.mod h1:Dmjwr6tYFIseJw7a3dRLJfsHAMXZ3nEnL/aZY+0IuI4=
 github.com/tklauser/numcpus v0.10.0 h1:18njr6LDBk1zuna922MgdjQuJFjrdppsZG60sHGfjso=
 github.com/tklauser/numcpus v0.10.0/go.mod h1:BiTKazU708GQTYF4mB+cmlpT2Is1gLk7XVuEeem8LsQ=
-github.com/ulikunitz/xz v0.5.9/go.mod h1:nbz6k7qbPmH4IRqmfOplQw/tblSgqTqBwxkY0oWt/14=
+github.com/x448/float16 v0.8.4 h1:qLwI1I70+NjRFUR3zs1JPUCgaCXSh3SW62uAKT1mSBM=
-github.com/ulikunitz/xz v0.5.12 h1:37Nm15o69RwBkXM0J6A5OlE67RZTfzUxTj8fB3dfcsc=
+github.com/x448/float16 v0.8.4/go.mod h1:14CWIYCyZA/cWjXOioeEpHeN/83MdbZDRQHoFcYsOfg=
-github.com/ulikunitz/xz v0.5.12/go.mod h1:nbz6k7qbPmH4IRqmfOplQw/tblSgqTqBwxkY0oWt/14=
 github.com/yusufpapurcu/wmi v1.2.4 h1:zFUKzehAFReQwLys1b/iSMl+JQGSCSjtVqQn9bBrPo0=
 github.com/yusufpapurcu/wmi v1.2.4/go.mod h1:SBZ9tNy3G9/m5Oi98Zks0QjeHVDvuK0qfxQmPyzfmi0=
 go.uber.org/automaxprocs v1.6.0 h1:O3y2/QNTOdbF+e/dpXNNW7Rx2hZ4sTIPyybbxyNqTUs=
 go.uber.org/automaxprocs v1.6.0/go.mod h1:ifeIMSnPZuznNm6jmdzmU3/bfk01Fe2fotchwEFJ8r8=
 golang.org/x/crypto v0.0.0-20190308221718-c2843e01d9a2/go.mod h1:djNgcEr1/C05ACkg1iLfiJU5Ep61QUkGW8qpdssI0+w=
-golang.org/x/crypto v0.0.0-20201221181555-eec23a3978ad/go.mod h1:jdWPYTVW3xRLrWPugEBEK3UY2ZEsg3UU495nc5E+M+I=
+golang.org/x/crypto v0.41.0 h1:WKYxWedPGCTVVl5+WHSSrOBT0O8lx32+zxmHxijgXp4=
-golang.org/x/crypto v0.37.0 h1:kJNSjF/Xp7kU0iB2Z+9viTPMW4EqqsrywMXLJOOsXSE=
+golang.org/x/crypto v0.41.0/go.mod h1:pO5AFd7FA68rFak7rOAGVuygIISepHftHnr8dr6+sUc=
-golang.org/x/crypto v0.37.0/go.mod h1:vg+k43peMZ0pUMhYmVAWysMK35e6ioLh3wB8ZCAfbVc=
+golang.org/x/exp v0.0.0-20250819193227-8b4c13bb791b h1:DXr+pvt3nC887026GRP39Ej11UATqWDmWuS99x26cD0=
-golang.org/x/exp v0.0.0-20250408133849-7e4ce0ab07d0 h1:R84qjqJb5nVJMxqWYb3np9L5ZsaDtB+a39EqjV0JSUM=
+golang.org/x/exp v0.0.0-20250819193227-8b4c13bb791b/go.mod h1:4QTo5u+SEIbbKW1RacMZq1YEfOBqeXa19JeshGi+zc4=
-golang.org/x/exp v0.0.0-20250408133849-7e4ce0ab07d0/go.mod h1:S9Xr4PYopiDyqSyp5NjCrhFrqg6A5zA2E/iPHPhqnS8=
 golang.org/x/image v0.0.0-20191009234506-e7c1f5e7dbb8/go.mod h1:FeLwcggjj3mMvU+oOTbSwawSJRM1uh48EjtB4UJZlP0=
-golang.org/x/image v0.26.0 h1:4XjIFEZWQmCZi6Wv8BoxsDhRU3RVnLX04dToTDAEPlY=
+golang.org/x/image v0.30.0 h1:jD5RhkmVAnjqaCUXfbGBrn3lpxbknfN9w2UhHHU+5B4=
-golang.org/x/image v0.26.0/go.mod h1:lcxbMFAovzpnJxzXS3nyL83K27tmqtKzIJpctK8YO5c=
+golang.org/x/image v0.30.0/go.mod h1:SAEUTxCCMWSrJcCy/4HwavEsfZZJlYxeHLc6tTiAe/c=
-golang.org/x/mod v0.24.0 h1:ZfthKaKaT4NrhGVZHO1/WDTwGES4De8KtWO0SIbNJMU=
+golang.org/x/mod v0.27.0 h1:kb+q2PyFnEADO2IEF935ehFUXlWiNjJWtRNgBLSfbxQ=
-golang.org/x/mod v0.24.0/go.mod h1:IXM97Txy2VM4PJ3gI61r1YEk/gAj6zAHN3AdZt6S9Ww=
+golang.org/x/mod v0.27.0/go.mod h1:rWI627Fq0DEoudcK+MBkNkCe0EetEaDSwJJkCcjpazc=
-golang.org/x/net v0.0.0-20180724234803-3673e40ba225/go.mod h1:mL1N/T3taQHkDXs73rZJwtUhF3w3ftmwwsq0BUmARs4=
-golang.org/x/net v0.0.0-20180906233101-161cd47e91fd/go.mod h1:mL1N/T3taQHkDXs73rZJwtUhF3w3ftmwwsq0BUmARs4=
-golang.org/x/net v0.0.0-20190311183353-d8887717615a/go.mod h1:t9HGtf8HONx5eT2rtn7q6eTqICYqUVnKs3thJo3Qplg=
-golang.org/x/net v0.0.0-20190404232315-eb5bcb51f2a3/go.mod h1:t9HGtf8HONx5eT2rtn7q6eTqICYqUVnKs3thJo3Qplg=
 golang.org/x/net v0.0.0-20190603091049-60506f45cf65/go.mod h1:HSz+uSET+XFnRR8LxR5pz3Of3rY3CfYBVs4xY44aLks=
-golang.org/x/net v0.39.0 h1:ZCu7HMWDxpXpaiKdhzIfaltL9Lp31x/3fCP11bc6/fY=
+golang.org/x/net v0.43.0 h1:lat02VYK2j4aLzMzecihNvTlJNQUq316m2Mr9rnM6YE=
-golang.org/x/net v0.39.0/go.mod h1:X7NRbYVEA+ewNkCNyJ513WmMdQ3BineSwVtN2zD/d+E=
+golang.org/x/net v0.43.0/go.mod h1:vhO1fvI4dGsIjh73sWfUVjj3N7CA9WkKJNQm2svM6Jg=
-golang.org/x/oauth2 v0.0.0-20180821212333-d2e6202438be/go.mod h1:N/0e6XlmueqKjAGxoOufVs8QHGRruUQn6yWY3a++T0U=
+golang.org/x/oauth2 v0.30.0 h1:dnDm7JmhM45NNpd8FDDeLhK6FwqbOf4MLCM9zb1BOHI=
-golang.org/x/oauth2 v0.0.0-20181106182150-f42d05182288/go.mod h1:N/0e6XlmueqKjAGxoOufVs8QHGRruUQn6yWY3a++T0U=
+golang.org/x/oauth2 v0.30.0/go.mod h1:B++QgG3ZKulg6sRPGD/mqlHQs5rB3Ml9erfeDY7xKlU=
-golang.org/x/oauth2 v0.29.0 h1:WdYw2tdTK1S8olAzWHdgeqfy+Mtm9XNhv/xJsY65d98=
+golang.org/x/sync v0.16.0 h1:ycBJEhp9p4vXvUZNszeOq0kGTPghopOL8q0fq3vstxw=
-golang.org/x/oauth2 v0.29.0/go.mod h1:onh5ek6nERTohokkhCD/y2cV4Do3fxFHFuAejCkRWT8=
+golang.org/x/sync v0.16.0/go.mod h1:1dzgHSNfp02xaA81J2MS99Qcpr2w7fw1gpm99rleRqA=
-golang.org/x/sync v0.0.0-20180314180146-1d60e4601c6f/go.mod h1:RxMgew5VJxzue5/jJTE5uejpjVlOe/izrB70Jof72aM=
-golang.org/x/sync v0.13.0 h1:AauUjRAJ9OSnvULf/ARrrVywoJDy0YS2AwQ98I37610=
-golang.org/x/sync v0.13.0/go.mod h1:1dzgHSNfp02xaA81J2MS99Qcpr2w7fw1gpm99rleRqA=
-golang.org/x/sys v0.0.0-20180909124046-d0be0721c37e/go.mod h1:STP8DvDyc/dI5b8T5hshtkjS+E42TnysNCUPdjciGhY=
 golang.org/x/sys v0.0.0-20190215142949-d0b11bdaac8a/go.mod h1:STP8DvDyc/dI5b8T5hshtkjS+E42TnysNCUPdjciGhY=
 golang.org/x/sys v0.0.0-20190916202348-b4ddaad3f8a3/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
-golang.org/x/sys v0.0.0-20191026070338-33540a1f6037/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.0.0-20201204225414-ed752295db88/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.1.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.6.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
-golang.org/x/sys v0.32.0 h1:s77OFDvIQeibCmezSnk/q6iAfkdiQaJi4VzroCFrN20=
+golang.org/x/sys v0.35.0 h1:vz1N37gP5bs89s7He8XuIYXpyY0+QlsKmzipCbUtyxI=
-golang.org/x/sys v0.32.0/go.mod h1:BJP2sWEmIv4KK5OTEluFJCKSidICx8ciO85XgH3Ak8k=
+golang.org/x/sys v0.35.0/go.mod h1:BJP2sWEmIv4KK5OTEluFJCKSidICx8ciO85XgH3Ak8k=
-golang.org/x/term v0.0.0-20201117132131-f5c789dd3221/go.mod h1:Nr5EML6q2oocZ2LXRh80K7BxOlk5/8JxuGnuhpl+muw=
+golang.org/x/term v0.34.0 h1:O/2T7POpk0ZZ7MAzMeWFSg6S5IpWd/RXDlM9hgM3DR4=
-golang.org/x/term v0.31.0 h1:erwDkOK1Msy6offm1mOgvspSkslFnIGsFnxOKoufg3o=
+golang.org/x/term v0.34.0/go.mod h1:5jC53AEywhIVebHgPVeg0mj8OD3VO9OzclacVrqpaAw=
-golang.org/x/term v0.31.0/go.mod h1:R4BeIy7D95HzImkxGkTW1UQTtP54tio2RyHz7PwK0aw=
 golang.org/x/text v0.3.0/go.mod h1:NqM8EUOU14njkJ3fqMW+pc6Ldnwhi/IjpwHt7yyuwOQ=
 golang.org/x/text v0.3.2/go.mod h1:bEr9sfX3Q8Zfm5fL9x+3itogRgK3+ptLWKqgva+5dAk=
-golang.org/x/text v0.3.5/go.mod h1:5Zoc/QRtKVWzQhOtBMvqHzDpF6irO9z98xDceosuGiQ=
+golang.org/x/text v0.28.0 h1:rhazDwis8INMIwQ4tpjLDzUhx6RlXqZNPEM0huQojng=
-golang.org/x/text v0.24.0 h1:dd5Bzh4yt5KYA8f9CJHCP4FB4D51c2c6JvN37xJJkJ0=
+golang.org/x/text v0.28.0/go.mod h1:U8nCwOR8jO/marOQ0QbDiOngZVEBB7MAiitBuMjXiNU=
-golang.org/x/text v0.24.0/go.mod h1:L8rBsPeo2pSS+xqN0d5u2ikmjtmoJbDBT1b7nHvFCdU=
 golang.org/x/tools v0.0.0-20180917221912-90fa682c2a6e/go.mod h1:n7NCudcB/nEzxVGmLbDWY5pfWTLqBcC2KZ6jyYvM4mQ=
-golang.org/x/tools v0.32.0 h1:Q7N1vhpkQv7ybVzLFtTjvQya2ewbwNDZzUgfXGqtMWU=
+golang.org/x/tools v0.36.0 h1:kWS0uv/zsvHEle1LbV5LE8QujrxB3wfQyxHfhOk0Qkg=
-golang.org/x/tools v0.32.0/go.mod h1:ZxrU41P/wAbZD8EDa6dDCa6XfpkhJ7HFMjHJXfBDu8s=
+golang.org/x/tools v0.36.0/go.mod h1:WBDiHKJK8YgLHlcQPYQzNCkUxUypCaa5ZegCVutKm+s=
-golang.org/x/xerrors v0.0.0-20191204190536-9bdfabe68543/go.mod h1:I/5z698sn9Ka8TeJc9MKroUUfqBBauWjQqLJ2OPfmY0=
-google.golang.org/appengine v1.1.0/go.mod h1:EbEs0AVv82hx2wNQdGPgUI5lhzA/G0D9YwlJXL52JkM=
-google.golang.org/appengine v1.3.0/go.mod h1:xpcJRLb0r/rnEns0DIKYYv+WjYCduHsrkT7/EB5XEv4=
 google.golang.org/appengine v1.6.5/go.mod h1:8WjMMxjGQR8xUklV/ARdw2HLXBOI7O7uCIDZVag1xfc=
 google.golang.org/protobuf v1.36.6 h1:z1NpPI8ku2WgiWnf+t9wTPsn6eP1L7ksHUlkfLvd9xY=
 google.golang.org/protobuf v1.36.6/go.mod h1:jduwjTPXsFjZGTmRluh+L6NjiWu7pchiJ2/5YcXBHnY=
 gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0=
 gopkg.in/check.v1 v1.0.0-20180628173108-788fd7840127 h1:qIbj1fsPNlZgppZ+VLlY7N33q108Sa+fhmuc+sWQYwY=
 gopkg.in/check.v1 v1.0.0-20180628173108-788fd7840127/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0=
-gopkg.in/fsnotify.v1 v1.4.7/go.mod h1:Tz8NjZHkW78fSQdbUxIjBTcgA1z1m8ZHf0WmKUhAMys=
+gopkg.in/yaml.v1 v1.0.0-20140924161607-9f9df34309c0/go.mod h1:WDnlLJ4WF5VGsH/HVa3CI79GS0ol3YnhVnKP89i0kNg=
-gopkg.in/tomb.v1 v1.0.0-20141024135613-dd632973f1e7/go.mod h1:dt/ZhP58zS4L8KSrWDmTeBkI65Dw0HsyUHuEVlX15mw=
-gopkg.in/yaml.v2 v2.2.1/go.mod h1:hI93XBmqTisBFMUTm0b8Fm+jr3Dg1NNxqwp+5A1VGuI=
 gopkg.in/yaml.v2 v2.2.2/go.mod h1:hI93XBmqTisBFMUTm0b8Fm+jr3Dg1NNxqwp+5A1VGuI=
 gopkg.in/yaml.v3 v3.0.1 h1:fxVm/GzAzEWqLHuvctI91KS9hhNmmWOoWu0XTYJS7CA=
 gopkg.in/yaml.v3 v3.0.1/go.mod h1:K4uyk7z7BCEPqu6E+C64Yfv1cQ7kz7rIZviUmN+EgEM=
-modernc.org/cc/v4 v4.26.0 h1:QMYvbVduUGH0rrO+5mqF/PSPPRZNpRtg2CLELy7vUpA=
+howett.net/plist v1.0.1 h1:37GdZ8tP09Q35o9ych3ehygcsL+HqKSwzctveSlarvM=
-modernc.org/cc/v4 v4.26.0/go.mod h1:uVtb5OGqUKpoLWhqwNQo/8LwvoiEBLvZXIQ/SmO6mL0=
+howett.net/plist v1.0.1/go.mod h1:lqaXoTrLY4hg8tnEzNru53gicrbv7rrk+2xJA/7hw9g=
-modernc.org/ccgo/v4 v4.26.0 h1:gVzXaDzGeBYJ2uXTOpR8FR7OlksDOe9jxnjhIKCsiTc=
+modernc.org/cc/v4 v4.26.2 h1:991HMkLjJzYBIfha6ECZdjrIYz2/1ayr+FL8GN+CNzM=
-modernc.org/ccgo/v4 v4.26.0/go.mod h1:Sem8f7TFUtVXkG2fiaChQtyyfkqhJBg/zjEJBkmuAVY=
+modernc.org/cc/v4 v4.26.2/go.mod h1:uVtb5OGqUKpoLWhqwNQo/8LwvoiEBLvZXIQ/SmO6mL0=
-modernc.org/fileutil v1.3.1 h1:8vq5fe7jdtEvoCf3Zf9Nm0Q05sH6kGx0Op2CPx1wTC8=
+modernc.org/ccgo/v4 v4.28.0 h1:rjznn6WWehKq7dG4JtLRKxb52Ecv8OUGah8+Z/SfpNU=
-modernc.org/fileutil v1.3.1/go.mod h1:HxmghZSZVAz/LXcMNwZPA/DRrQZEVP9VX0V4LQGQFOc=
+modernc.org/ccgo/v4 v4.28.0/go.mod h1:JygV3+9AV6SmPhDasu4JgquwU81XAKLd3OKTUDNOiKE=
+modernc.org/fileutil v1.3.8 h1:qtzNm7ED75pd1C7WgAGcK4edm4fvhtBsEiI/0NQ54YM=
+modernc.org/fileutil v1.3.8/go.mod h1:HxmghZSZVAz/LXcMNwZPA/DRrQZEVP9VX0V4LQGQFOc=
 modernc.org/gc/v2 v2.6.5 h1:nyqdV8q46KvTpZlsw66kWqwXRHdjIlJOhG6kxiV/9xI=
 modernc.org/gc/v2 v2.6.5/go.mod h1:YgIahr1ypgfe7chRuJi2gD7DBQiKSLMPgBQe9oIiito=
-modernc.org/libc v1.64.0 h1:U0k8BD2d3cD3e9I8RLcZgJBHAcsJzbXx5mKGSb5pyJA=
+modernc.org/goabi0 v0.2.0 h1:HvEowk7LxcPd0eq6mVOAEMai46V+i7Jrj13t4AzuNks=
-modernc.org/libc v1.64.0/go.mod h1:7m9VzGq7APssBTydds2zBcxGREwvIGpuUBaKTXdm2Qs=
+modernc.org/goabi0 v0.2.0/go.mod h1:CEFRnnJhKvWT1c1JTI3Avm+tgOWbkOu5oPA8eH8LnMI=
+modernc.org/libc v1.66.3 h1:cfCbjTUcdsKyyZZfEUKfoHcP3S0Wkvz3jgSzByEWVCQ=
+modernc.org/libc v1.66.3/go.mod h1:XD9zO8kt59cANKvHPXpx7yS2ELPheAey0vjIuZOhOU8=
 modernc.org/mathutil v1.7.1 h1:GCZVGXdaN8gTqB1Mf/usp1Y/hSqgI2vAGGP4jZMCxOU=
 modernc.org/mathutil v1.7.1/go.mod h1:4p5IwJITfppl0G4sUEDtCr4DthTaT47/N3aT6MhfgJg=
-modernc.org/memory v1.10.0 h1:fzumd51yQ1DxcOxSO+S6X7+QTuVU+n8/Aj7swYjFfC4=
+modernc.org/memory v1.11.0 h1:o4QC8aMQzmcwCK3t3Ux/ZHmwFPzE6hf2Y5LbkRs+hbI=
-modernc.org/memory v1.10.0/go.mod h1:/JP4VbVC+K5sU2wZi9bHoq2MAkCnrt2r98UGeSK7Mjw=
+modernc.org/memory v1.11.0/go.mod h1:/JP4VbVC+K5sU2wZi9bHoq2MAkCnrt2r98UGeSK7Mjw=
 modernc.org/opt v0.1.4 h1:2kNGMRiUjrp4LcaPuLY2PzUfqM/w9N23quVwhKt5Qm8=
 modernc.org/opt v0.1.4/go.mod h1:03fq9lsNfvkYSfxrfUhZCWPk1lm4cq4N+Bh//bEtgns=
 modernc.org/sortutil v1.2.1 h1:+xyoGf15mM3NMlPDnFqrteY07klSFxLElE2PVuWIJ7w=
 modernc.org/sortutil v1.2.1/go.mod h1:7ZI3a3REbai7gzCLcotuw9AC4VZVpYMjDzETGsSMqJE=
-modernc.org/sqlite v1.37.0 h1:s1TMe7T3Q3ovQiK2Ouz4Jwh7dw4ZDqbebSDTlSJdfjI=
+modernc.org/sqlite v1.38.2 h1:Aclu7+tgjgcQVShZqim41Bbw9Cho0y/7WzYptXqkEek=
-modernc.org/sqlite v1.37.0/go.mod h1:5YiWv+YviqGMuGw4V+PNplcyaJ5v+vQd7TQOgkACoJM=
+modernc.org/sqlite v1.38.2/go.mod h1:cPTJYSlgg3Sfg046yBShXENNtPrWrDX8bsbAQBzgQ5E=
 modernc.org/strutil v1.2.1 h1:UneZBkQA+DX2Rp35KcM69cSsNES9ly8mQWD71HKlOA0=
 modernc.org/strutil v1.2.1/go.mod h1:EHkiggD70koQxjVdSBM3JKM7k6L0FbGE5eymy9i3B9A=
 modernc.org/token v1.1.0 h1:Xl7Ap9dKaEs5kLoOQeQmPWevfnk/DM5qcLcYlA8ys6Y=

beszel/internal/agent/agent.go

@@ -4,34 +4,55 @@ package agent
 import (
 	"beszel"
 	"beszel/internal/entities/system"
+	"crypto/sha256"
+	"encoding/hex"
 	"log/slog"
 	"os"
+	"path/filepath"
 	"strings"
 	"sync"
 	"time"
+
+	"github.com/gliderlabs/ssh"
+	"github.com/shirou/gopsutil/v4/host"
+	gossh "golang.org/x/crypto/ssh"
 )
 
 type Agent struct {
-	sync.Mutex                               // Used to lock agent while collecting data
+	sync.Mutex                                   // Used to lock agent while collecting data
-	debug         bool                       // true if LOG_LEVEL is set to debug
+	debug             bool                       // true if LOG_LEVEL is set to debug
-	zfs           bool                       // true if system has arcstats
+	zfs               bool                       // true if system has arcstats
-	memCalc       string                     // Memory calculation formula
+	memCalc           string                     // Memory calculation formula
-	fsNames       []string                   // List of filesystem device names being monitored
+	fsNames           []string                   // List of filesystem device names being monitored
-	fsStats       map[string]*system.FsStats // Keeps track of disk stats for each filesystem
+	fsStats           map[string]*system.FsStats // Keeps track of disk stats for each filesystem
-	netInterfaces map[string]struct{}        // Stores all valid network interfaces
+	netInterfaces     map[string]struct{}        // Stores all valid network interfaces
-	netIoStats    system.NetIoStats          // Keeps track of bandwidth usage
+	netIoStats        system.NetIoStats          // Keeps track of bandwidth usage
-	dockerManager *dockerManager             // Manages Docker API requests
+	dockerManager     *dockerManager             // Manages Docker API requests
-	sensorConfig  *SensorConfig              // Sensors config
+	sensorConfig      *SensorConfig              // Sensors config
-	systemInfo    system.Info                // Host system info
+	systemInfo        system.Info                // Host system info
-	gpuManager    *GPUManager                // Manages GPU data
+	gpuManager        *GPUManager                // Manages GPU data
-	cache         *SessionCache              // Cache for system stats based on primary session ID
+	cache             *SessionCache              // Cache for system stats based on primary session ID
+	connectionManager *ConnectionManager         // Channel to signal connection events
+	server            *ssh.Server                // SSH server
+	dataDir           string                     // Directory for persisting data
+	keys              []gossh.PublicKey          // SSH public keys
 }
 
-func NewAgent() *Agent {
+// NewAgent creates a new agent with the given data directory for persisting data.
-	agent := &Agent{
+// If the data directory is not set, it will attempt to find the optimal directory.
+func NewAgent(dataDir ...string) (agent *Agent, err error) {
+	agent = &Agent{
 		fsStats: make(map[string]*system.FsStats),
 		cache:   NewSessionCache(69 * time.Second),
 	}
+
+	agent.dataDir, err = getDataDir(dataDir...)
+	if err != nil {
+		slog.Warn("Data directory not found")
+	} else {
+		slog.Info("Data directory", "path", agent.dataDir)
+	}
+
 	agent.memCalc, _ = GetEnv("MEM_CALC")
 	agent.sensorConfig = agent.newSensorConfig()
 	// Set up slog with a log level determined by the LOG_LEVEL env var
@@ -49,10 +70,19 @@ func NewAgent() *Agent {
 
 	slog.Debug(beszel.Version)
 
-	// initialize system info / docker manager
+	// initialize system info
 	agent.initializeSystemInfo()
+
+	// initialize connection manager
+	agent.connectionManager = newConnectionManager(agent)
+
+	// initialize disk info
 	agent.initializeDiskInfo()
+
+	// initialize net io stats
 	agent.initializeNetIoStats()
+
+	// initialize docker manager
 	agent.dockerManager = newDockerManager(agent)
 
 	// initialize GPU manager
@@ -67,7 +97,7 @@ func NewAgent() *Agent {
 		slog.Debug("Stats", "data", agent.gatherStats(""))
 	}
 
-	return agent
+	return agent, nil
 }
 
 // GetEnv retrieves an environment variable with a "BESZEL_AGENT_" prefix, or falls back to the unprefixed key.
@@ -83,35 +113,70 @@ func (a *Agent) gatherStats(sessionID string) *system.CombinedData {
 	a.Lock()
 	defer a.Unlock()
 
-	cachedData, ok := a.cache.Get(sessionID)
+	data, isCached := a.cache.Get(sessionID)
-	if ok {
+	if isCached {
-		slog.Debug("Cached stats", "session", sessionID)
+		slog.Debug("Cached data", "session", sessionID)
-		return cachedData
+		return data
 	}
 
-	*cachedData = system.CombinedData{
+	*data = system.CombinedData{
 		Stats: a.getSystemStats(),
 		Info:  a.systemInfo,
 	}
-	slog.Debug("System stats", "data", cachedData)
+	slog.Debug("System data", "data", data)
 
 	if a.dockerManager != nil {
 		if containerStats, err := a.dockerManager.getDockerStats(); err == nil {
-			cachedData.Containers = containerStats
+			data.Containers = containerStats
-			slog.Debug("Docker stats", "data", cachedData.Containers)
+			slog.Debug("Containers", "data", data.Containers)
 		} else {
-			slog.Debug("Docker stats", "err", err)
+			slog.Debug("Containers", "err", err)
 		}
 	}
 
-	cachedData.Stats.ExtraFs = make(map[string]*system.FsStats)
+	data.Stats.ExtraFs = make(map[string]*system.FsStats)
 	for name, stats := range a.fsStats {
 		if !stats.Root && stats.DiskTotal > 0 {
-			cachedData.Stats.ExtraFs[name] = stats
+			data.Stats.ExtraFs[name] = stats
 		}
 	}
-	slog.Debug("Extra filesystems", "data", cachedData.Stats.ExtraFs)
+	slog.Debug("Extra FS", "data", data.Stats.ExtraFs)
 
-	a.cache.Set(sessionID, cachedData)
+	a.cache.Set(sessionID, data)
-	return cachedData
+	return data
+}
+
+// StartAgent initializes and starts the agent with optional WebSocket connection
+func (a *Agent) Start(serverOptions ServerOptions) error {
+	a.keys = serverOptions.Keys
+	return a.connectionManager.Start(serverOptions)
+}
+
+func (a *Agent) getFingerprint() string {
+	// first look for a fingerprint in the data directory
+	if a.dataDir != "" {
+		if fp, err := os.ReadFile(filepath.Join(a.dataDir, "fingerprint")); err == nil {
+			return string(fp)
+		}
+	}
+
+	// if no fingerprint is found, generate one
+	fingerprint, err := host.HostID()
+	if err != nil || fingerprint == "" {
+		fingerprint = a.systemInfo.Hostname + a.systemInfo.CpuModel
+	}
+
+	// hash fingerprint
+	sum := sha256.Sum256([]byte(fingerprint))
+	fingerprint = hex.EncodeToString(sum[:24])
+
+	// save fingerprint to data directory
+	if a.dataDir != "" {
+		err = os.WriteFile(filepath.Join(a.dataDir, "fingerprint"), []byte(fingerprint), 0644)
+		if err != nil {
+			slog.Warn("Failed to save fingerprint", "err", err)
+		}
+	}
+
+	return fingerprint
 }

beszel/internal/agent/agent_cache_test.go

@@ -1,3 +1,6 @@
+//go:build testing
+// +build testing
+
 package agent
 
 import (
@@ -11,7 +14,7 @@ import (
 )
 
 func TestSessionCache_GetSet(t *testing.T) {
-	synctest.Run(func() {
+	synctest.Test(t, func(t *testing.T) {
 		cache := NewSessionCache(69 * time.Second)
 
 		testData := &system.CombinedData{

beszel/internal/agent/agent_test_helpers.go

@@ -0,0 +1,9 @@
+//go:build testing
+// +build testing
+
+package agent
+
+// TESTING ONLY: GetConnectionManager is a helper function to get the connection manager for testing.
+func (a *Agent) GetConnectionManager() *ConnectionManager {
+	return a.connectionManager
+}

beszel/internal/agent/battery/battery.go

@@ -0,0 +1,53 @@
+//go:build !freebsd
+
+// Package battery provides functions to check if the system has a battery and to get the battery stats.
+package battery
+
+import (
+	"errors"
+	"log/slog"
+
+	"github.com/distatus/battery"
+)
+
+var systemHasBattery = false
+var haveCheckedBattery = false
+
+// HasReadableBattery checks if the system has a battery and returns true if it does.
+func HasReadableBattery() bool {
+	if haveCheckedBattery {
+		return systemHasBattery
+	}
+	haveCheckedBattery = true
+	bat, err := battery.Get(0)
+	if err == nil && bat != nil {
+		systemHasBattery = true
+	} else {
+		slog.Debug("No battery found", "err", err)
+	}
+	return systemHasBattery
+}
+
+// GetBatteryStats returns the current battery percent and charge state
+func GetBatteryStats() (batteryPercent uint8, batteryState uint8, err error) {
+	if !systemHasBattery {
+		return batteryPercent, batteryState, errors.ErrUnsupported
+	}
+	batteries, err := battery.GetAll()
+	if err != nil || len(batteries) == 0 {
+		return batteryPercent, batteryState, err
+	}
+	totalCapacity := float64(0)
+	totalCharge := float64(0)
+	for _, bat := range batteries {
+		if bat.Design != 0 {
+			totalCapacity += bat.Design
+		} else {
+			totalCapacity += bat.Full
+		}
+		totalCharge += bat.Current
+	}
+	batteryPercent = uint8(totalCharge / totalCapacity * 100)
+	batteryState = uint8(batteries[0].State.Raw)
+	return batteryPercent, batteryState, nil
+}

beszel/internal/agent/battery/battery_freebsd.go

@@ -0,0 +1,13 @@
+//go:build freebsd
+
+package battery
+
+import "errors"
+
+func HasReadableBattery() bool {
+	return false
+}
+
+func GetBatteryStats() (uint8, uint8, error) {
+	return 0, 0, errors.ErrUnsupported
+}

beszel/internal/agent/client.go

@@ -0,0 +1,265 @@
+package agent
+
+import (
+	"beszel"
+	"beszel/internal/common"
+	"crypto/tls"
+	"errors"
+	"fmt"
+	"log/slog"
+	"net"
+	"net/http"
+	"net/url"
+	"os"
+	"path"
+	"strings"
+	"time"
+
+	"github.com/fxamacker/cbor/v2"
+	"github.com/lxzan/gws"
+	"golang.org/x/crypto/ssh"
+)
+
+const (
+	wsDeadline = 70 * time.Second
+)
+
+// WebSocketClient manages the WebSocket connection between the agent and hub.
+// It handles authentication, message routing, and connection lifecycle management.
+type WebSocketClient struct {
+	gws.BuiltinEventHandler
+	options            *gws.ClientOption                   // WebSocket client configuration options
+	agent              *Agent                              // Reference to the parent agent
+	Conn               *gws.Conn                           // Active WebSocket connection
+	hubURL             *url.URL                            // Parsed hub URL for connection
+	token              string                              // Authentication token for hub registration
+	fingerprint        string                              // System fingerprint for identification
+	hubRequest         *common.HubRequest[cbor.RawMessage] // Reusable request structure for message parsing
+	lastConnectAttempt time.Time                           // Timestamp of last connection attempt
+	hubVerified        bool                                // Whether the hub has been cryptographically verified
+}
+
+// newWebSocketClient creates a new WebSocket client for the given agent.
+// It reads configuration from environment variables and validates the hub URL.
+func newWebSocketClient(agent *Agent) (client *WebSocketClient, err error) {
+	hubURLStr, exists := GetEnv("HUB_URL")
+	if !exists {
+		return nil, errors.New("HUB_URL environment variable not set")
+	}
+
+	client = &WebSocketClient{}
+
+	client.hubURL, err = url.Parse(hubURLStr)
+	if err != nil {
+		return nil, errors.New("invalid hub URL")
+	}
+	// get registration token
+	client.token, err = getToken()
+	if err != nil {
+		return nil, err
+	}
+
+	client.agent = agent
+	client.hubRequest = &common.HubRequest[cbor.RawMessage]{}
+	client.fingerprint = agent.getFingerprint()
+
+	return client, nil
+}
+
+// getToken returns the token for the WebSocket client.
+// It first checks the TOKEN environment variable, then the TOKEN_FILE environment variable.
+// If neither is set, it returns an error.
+func getToken() (string, error) {
+	// get token from env var
+	token, _ := GetEnv("TOKEN")
+	if token != "" {
+		return token, nil
+	}
+	// get token from file
+	tokenFile, _ := GetEnv("TOKEN_FILE")
+	if tokenFile == "" {
+		return "", errors.New("must set TOKEN or TOKEN_FILE")
+	}
+	tokenBytes, err := os.ReadFile(tokenFile)
+	if err != nil {
+		return "", err
+	}
+	return string(tokenBytes), nil
+}
+
+// getOptions returns the WebSocket client options, creating them if necessary.
+// It configures the connection URL, TLS settings, and authentication headers.
+func (client *WebSocketClient) getOptions() *gws.ClientOption {
+	if client.options != nil {
+		return client.options
+	}
+
+	// update the hub url to use websocket scheme and api path
+	if client.hubURL.Scheme == "https" {
+		client.hubURL.Scheme = "wss"
+	} else {
+		client.hubURL.Scheme = "ws"
+	}
+	client.hubURL.Path = path.Join(client.hubURL.Path, "api/beszel/agent-connect")
+
+	client.options = &gws.ClientOption{
+		Addr:      client.hubURL.String(),
+		TlsConfig: &tls.Config{InsecureSkipVerify: true},
+		RequestHeader: http.Header{
+			"User-Agent": []string{getUserAgent()},
+			"X-Token":    []string{client.token},
+			"X-Beszel":   []string{beszel.Version},
+		},
+	}
+	return client.options
+}
+
+// Connect establishes a WebSocket connection to the hub.
+// It closes any existing connection before attempting to reconnect.
+func (client *WebSocketClient) Connect() (err error) {
+	client.lastConnectAttempt = time.Now()
+
+	// make sure previous connection is closed
+	client.Close()
+
+	client.Conn, _, err = gws.NewClient(client, client.getOptions())
+	if err != nil {
+		return err
+	}
+
+	go client.Conn.ReadLoop()
+
+	return nil
+}
+
+// OnOpen handles WebSocket connection establishment.
+// It sets a deadline for the connection to prevent hanging.
+func (client *WebSocketClient) OnOpen(conn *gws.Conn) {
+	conn.SetDeadline(time.Now().Add(wsDeadline))
+}
+
+// OnClose handles WebSocket connection closure.
+// It logs the closure reason and notifies the connection manager.
+func (client *WebSocketClient) OnClose(conn *gws.Conn, err error) {
+	slog.Warn("Connection closed", "err", strings.TrimPrefix(err.Error(), "gws: "))
+	client.agent.connectionManager.eventChan <- WebSocketDisconnect
+}
+
+// OnMessage handles incoming WebSocket messages from the hub.
+// It decodes CBOR messages and routes them to appropriate handlers.
+func (client *WebSocketClient) OnMessage(conn *gws.Conn, message *gws.Message) {
+	defer message.Close()
+	conn.SetDeadline(time.Now().Add(wsDeadline))
+
+	if message.Opcode != gws.OpcodeBinary {
+		return
+	}
+
+	if err := cbor.NewDecoder(message.Data).Decode(client.hubRequest); err != nil {
+		slog.Error("Error parsing message", "err", err)
+		return
+	}
+	if err := client.handleHubRequest(client.hubRequest); err != nil {
+		slog.Error("Error handling message", "err", err)
+	}
+}
+
+// OnPing handles WebSocket ping frames.
+// It responds with a pong and updates the connection deadline.
+func (client *WebSocketClient) OnPing(conn *gws.Conn, message []byte) {
+	conn.SetDeadline(time.Now().Add(wsDeadline))
+	conn.WritePong(message)
+}
+
+// handleAuthChallenge verifies the authenticity of the hub and returns the system's fingerprint.
+func (client *WebSocketClient) handleAuthChallenge(msg *common.HubRequest[cbor.RawMessage]) (err error) {
+	var authRequest common.FingerprintRequest
+	if err := cbor.Unmarshal(msg.Data, &authRequest); err != nil {
+		return err
+	}
+
+	if err := client.verifySignature(authRequest.Signature); err != nil {
+		return err
+	}
+
+	client.hubVerified = true
+	client.agent.connectionManager.eventChan <- WebSocketConnect
+
+	response := &common.FingerprintResponse{
+		Fingerprint: client.fingerprint,
+	}
+
+	if authRequest.NeedSysInfo {
+		response.Hostname = client.agent.systemInfo.Hostname
+		serverAddr := client.agent.connectionManager.serverOptions.Addr
+		_, response.Port, _ = net.SplitHostPort(serverAddr)
+	}
+
+	return client.sendMessage(response)
+}
+
+// verifySignature verifies the signature of the token using the public keys.
+func (client *WebSocketClient) verifySignature(signature []byte) (err error) {
+	for _, pubKey := range client.agent.keys {
+		sig := ssh.Signature{
+			Format: pubKey.Type(),
+			Blob:   signature,
+		}
+		if err = pubKey.Verify([]byte(client.token), &sig); err == nil {
+			return nil
+		}
+	}
+	return errors.New("invalid signature - check KEY value")
+}
+
+// Close closes the WebSocket connection gracefully.
+// This method is safe to call multiple times.
+func (client *WebSocketClient) Close() {
+	if client.Conn != nil {
+		_ = client.Conn.WriteClose(1000, nil)
+	}
+}
+
+// handleHubRequest routes the request to the appropriate handler.
+// It ensures the hub is verified before processing most requests.
+func (client *WebSocketClient) handleHubRequest(msg *common.HubRequest[cbor.RawMessage]) error {
+	if !client.hubVerified && msg.Action != common.CheckFingerprint {
+		return errors.New("hub not verified")
+	}
+	switch msg.Action {
+	case common.GetData:
+		return client.sendSystemData()
+	case common.CheckFingerprint:
+		return client.handleAuthChallenge(msg)
+	}
+	return nil
+}
+
+// sendSystemData gathers and sends current system statistics to the hub.
+func (client *WebSocketClient) sendSystemData() error {
+	sysStats := client.agent.gatherStats(client.token)
+	return client.sendMessage(sysStats)
+}
+
+// sendMessage encodes the given data to CBOR and sends it as a binary message over the WebSocket connection to the hub.
+func (client *WebSocketClient) sendMessage(data any) error {
+	bytes, err := cbor.Marshal(data)
+	if err != nil {
+		return err
+	}
+	return client.Conn.WriteMessage(gws.OpcodeBinary, bytes)
+}
+
+// getUserAgent returns one of two User-Agent strings based on current time.
+// This is used to avoid being blocked by Cloudflare or other anti-bot measures.
+func getUserAgent() string {
+	const (
+		uaBase    = "Mozilla/5.0 (%s) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/124.0.0.0 Safari/537.36"
+		uaWindows = "Windows NT 11.0; Win64; x64"
+		uaMac     = "Macintosh; Intel Mac OS X 14_0_0"
+	)
+	if time.Now().UnixNano()%2 == 0 {
+		return fmt.Sprintf(uaBase, uaWindows)
+	}
+	return fmt.Sprintf(uaBase, uaMac)
+}

beszel/internal/agent/client_test.go

@@ -0,0 +1,538 @@
+//go:build testing
+// +build testing
+
+package agent
+
+import (
+	"beszel"
+	"beszel/internal/common"
+	"crypto/ed25519"
+	"net/url"
+	"os"
+	"strings"
+	"testing"
+	"time"
+
+	"github.com/fxamacker/cbor/v2"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+	"golang.org/x/crypto/ssh"
+)
+
+// TestNewWebSocketClient tests WebSocket client creation
+func TestNewWebSocketClient(t *testing.T) {
+	agent := createTestAgent(t)
+
+	testCases := []struct {
+		name        string
+		hubURL      string
+		token       string
+		expectError bool
+		errorMsg    string
+	}{
+		{
+			name:        "valid configuration",
+			hubURL:      "http://localhost:8080",
+			token:       "test-token-123",
+			expectError: false,
+		},
+		{
+			name:        "valid https URL",
+			hubURL:      "https://hub.example.com",
+			token:       "secure-token",
+			expectError: false,
+		},
+		{
+			name:        "missing hub URL",
+			hubURL:      "",
+			token:       "test-token",
+			expectError: true,
+			errorMsg:    "HUB_URL environment variable not set",
+		},
+		{
+			name:        "invalid URL",
+			hubURL:      "ht\ttp://invalid",
+			token:       "test-token",
+			expectError: true,
+			errorMsg:    "invalid hub URL",
+		},
+		{
+			name:        "missing token",
+			hubURL:      "http://localhost:8080",
+			token:       "",
+			expectError: true,
+			errorMsg:    "must set TOKEN or TOKEN_FILE",
+		},
+	}
+
+	for _, tc := range testCases {
+		t.Run(tc.name, func(t *testing.T) {
+			// Set up environment
+			if tc.hubURL != "" {
+				os.Setenv("BESZEL_AGENT_HUB_URL", tc.hubURL)
+			} else {
+				os.Unsetenv("BESZEL_AGENT_HUB_URL")
+			}
+			if tc.token != "" {
+				os.Setenv("BESZEL_AGENT_TOKEN", tc.token)
+			} else {
+				os.Unsetenv("BESZEL_AGENT_TOKEN")
+			}
+			defer func() {
+				os.Unsetenv("BESZEL_AGENT_HUB_URL")
+				os.Unsetenv("BESZEL_AGENT_TOKEN")
+			}()
+
+			client, err := newWebSocketClient(agent)
+
+			if tc.expectError {
+				assert.Error(t, err)
+				if err != nil && tc.errorMsg != "" {
+					assert.Contains(t, err.Error(), tc.errorMsg)
+				}
+				assert.Nil(t, client)
+			} else {
+				require.NoError(t, err)
+				assert.NotNil(t, client)
+				assert.Equal(t, agent, client.agent)
+				assert.Equal(t, tc.token, client.token)
+				assert.Equal(t, tc.hubURL, client.hubURL.String())
+				assert.NotEmpty(t, client.fingerprint)
+				assert.NotNil(t, client.hubRequest)
+			}
+		})
+	}
+}
+
+// TestWebSocketClient_GetOptions tests WebSocket client options configuration
+func TestWebSocketClient_GetOptions(t *testing.T) {
+	agent := createTestAgent(t)
+
+	testCases := []struct {
+		name           string
+		inputURL       string
+		expectedScheme string
+		expectedPath   string
+	}{
+		{
+			name:           "http to ws conversion",
+			inputURL:       "http://localhost:8080",
+			expectedScheme: "ws",
+			expectedPath:   "/api/beszel/agent-connect",
+		},
+		{
+			name:           "https to wss conversion",
+			inputURL:       "https://hub.example.com",
+			expectedScheme: "wss",
+			expectedPath:   "/api/beszel/agent-connect",
+		},
+		{
+			name:           "existing path preservation",
+			inputURL:       "http://localhost:8080/custom/path",
+			expectedScheme: "ws",
+			expectedPath:   "/custom/path/api/beszel/agent-connect",
+		},
+	}
+
+	for _, tc := range testCases {
+		t.Run(tc.name, func(t *testing.T) {
+			// Set up environment
+			os.Setenv("BESZEL_AGENT_HUB_URL", tc.inputURL)
+			os.Setenv("BESZEL_AGENT_TOKEN", "test-token")
+			defer func() {
+				os.Unsetenv("BESZEL_AGENT_HUB_URL")
+				os.Unsetenv("BESZEL_AGENT_TOKEN")
+			}()
+
+			client, err := newWebSocketClient(agent)
+			require.NoError(t, err)
+
+			options := client.getOptions()
+
+			// Parse the WebSocket URL
+			wsURL, err := url.Parse(options.Addr)
+			require.NoError(t, err)
+
+			assert.Equal(t, tc.expectedScheme, wsURL.Scheme)
+			assert.Equal(t, tc.expectedPath, wsURL.Path)
+
+			// Check headers
+			assert.Equal(t, "test-token", options.RequestHeader.Get("X-Token"))
+			assert.Equal(t, beszel.Version, options.RequestHeader.Get("X-Beszel"))
+			assert.Contains(t, options.RequestHeader.Get("User-Agent"), "Mozilla/5.0")
+
+			// Test options caching
+			options2 := client.getOptions()
+			assert.Same(t, options, options2, "Options should be cached")
+		})
+	}
+}
+
+// TestWebSocketClient_VerifySignature tests signature verification
+func TestWebSocketClient_VerifySignature(t *testing.T) {
+	agent := createTestAgent(t)
+
+	// Generate test key pairs
+	_, goodPrivKey, err := ed25519.GenerateKey(nil)
+	require.NoError(t, err)
+	goodPubKey, err := ssh.NewPublicKey(goodPrivKey.Public().(ed25519.PublicKey))
+	require.NoError(t, err)
+
+	_, badPrivKey, err := ed25519.GenerateKey(nil)
+	require.NoError(t, err)
+	badPubKey, err := ssh.NewPublicKey(badPrivKey.Public().(ed25519.PublicKey))
+	require.NoError(t, err)
+
+	// Set up environment
+	os.Setenv("BESZEL_AGENT_HUB_URL", "http://localhost:8080")
+	os.Setenv("BESZEL_AGENT_TOKEN", "test-token")
+	defer func() {
+		os.Unsetenv("BESZEL_AGENT_HUB_URL")
+		os.Unsetenv("BESZEL_AGENT_TOKEN")
+	}()
+
+	client, err := newWebSocketClient(agent)
+	require.NoError(t, err)
+
+	testCases := []struct {
+		name        string
+		keys        []ssh.PublicKey
+		token       string
+		signWith    ed25519.PrivateKey
+		expectError bool
+	}{
+		{
+			name:        "valid signature with correct key",
+			keys:        []ssh.PublicKey{goodPubKey},
+			token:       "test-token",
+			signWith:    goodPrivKey,
+			expectError: false,
+		},
+		{
+			name:        "invalid signature with wrong key",
+			keys:        []ssh.PublicKey{goodPubKey},
+			token:       "test-token",
+			signWith:    badPrivKey,
+			expectError: true,
+		},
+		{
+			name:        "valid signature with multiple keys",
+			keys:        []ssh.PublicKey{badPubKey, goodPubKey},
+			token:       "test-token",
+			signWith:    goodPrivKey,
+			expectError: false,
+		},
+		{
+			name:        "no valid keys",
+			keys:        []ssh.PublicKey{badPubKey},
+			token:       "test-token",
+			signWith:    goodPrivKey,
+			expectError: true,
+		},
+	}
+
+	for _, tc := range testCases {
+		t.Run(tc.name, func(t *testing.T) {
+			// Set up agent with test keys
+			agent.keys = tc.keys
+			client.token = tc.token
+
+			// Create signature
+			signature := ed25519.Sign(tc.signWith, []byte(tc.token))
+
+			err := client.verifySignature(signature)
+
+			if tc.expectError {
+				assert.Error(t, err)
+				assert.Contains(t, err.Error(), "invalid signature")
+			} else {
+				assert.NoError(t, err)
+			}
+		})
+	}
+}
+
+// TestWebSocketClient_HandleHubRequest tests hub request routing (basic verification logic)
+func TestWebSocketClient_HandleHubRequest(t *testing.T) {
+	agent := createTestAgent(t)
+
+	// Set up environment
+	os.Setenv("BESZEL_AGENT_HUB_URL", "http://localhost:8080")
+	os.Setenv("BESZEL_AGENT_TOKEN", "test-token")
+	defer func() {
+		os.Unsetenv("BESZEL_AGENT_HUB_URL")
+		os.Unsetenv("BESZEL_AGENT_TOKEN")
+	}()
+
+	client, err := newWebSocketClient(agent)
+	require.NoError(t, err)
+
+	testCases := []struct {
+		name        string
+		action      common.WebSocketAction
+		hubVerified bool
+		expectError bool
+		errorMsg    string
+	}{
+		{
+			name:        "CheckFingerprint without verification",
+			action:      common.CheckFingerprint,
+			hubVerified: false,
+			expectError: false, // CheckFingerprint is allowed without verification
+		},
+		{
+			name:        "GetData without verification",
+			action:      common.GetData,
+			hubVerified: false,
+			expectError: true,
+			errorMsg:    "hub not verified",
+		},
+	}
+
+	for _, tc := range testCases {
+		t.Run(tc.name, func(t *testing.T) {
+			client.hubVerified = tc.hubVerified
+
+			// Create minimal request
+			hubRequest := &common.HubRequest[cbor.RawMessage]{
+				Action: tc.action,
+				Data:   cbor.RawMessage{},
+			}
+
+			err := client.handleHubRequest(hubRequest)
+
+			if tc.expectError {
+				assert.Error(t, err)
+				if tc.errorMsg != "" {
+					assert.Contains(t, err.Error(), tc.errorMsg)
+				}
+			} else {
+				// For CheckFingerprint, we expect a decode error since we're not providing valid data,
+				// but it shouldn't be the "hub not verified" error
+				if err != nil && tc.errorMsg != "" {
+					assert.NotContains(t, err.Error(), tc.errorMsg)
+				}
+			}
+		})
+	}
+}
+
+// TestWebSocketClient_GetUserAgent tests user agent generation
+func TestGetUserAgent(t *testing.T) {
+	// Run multiple times to check both variants
+	userAgents := make(map[string]bool)
+
+	for range 20 {
+		ua := getUserAgent()
+		userAgents[ua] = true
+
+		// Check that it's a valid Mozilla user agent
+		assert.Contains(t, ua, "Mozilla/5.0")
+		assert.Contains(t, ua, "AppleWebKit/537.36")
+		assert.Contains(t, ua, "Chrome/124.0.0.0")
+		assert.Contains(t, ua, "Safari/537.36")
+
+		// Should contain either Windows or Mac
+		isWindows := strings.Contains(ua, "Windows NT 11.0")
+		isMac := strings.Contains(ua, "Macintosh; Intel Mac OS X 14_0_0")
+		assert.True(t, isWindows || isMac, "User agent should contain either Windows or Mac identifier")
+	}
+
+	// With enough iterations, we should see both variants
+	// though this might occasionally fail
+	if len(userAgents) == 1 {
+		t.Log("Note: Only one user agent variant was generated in this test run")
+	}
+}
+
+// TestWebSocketClient_Close tests connection closing
+func TestWebSocketClient_Close(t *testing.T) {
+	agent := createTestAgent(t)
+
+	// Set up environment
+	os.Setenv("BESZEL_AGENT_HUB_URL", "http://localhost:8080")
+	os.Setenv("BESZEL_AGENT_TOKEN", "test-token")
+	defer func() {
+		os.Unsetenv("BESZEL_AGENT_HUB_URL")
+		os.Unsetenv("BESZEL_AGENT_TOKEN")
+	}()
+
+	client, err := newWebSocketClient(agent)
+	require.NoError(t, err)
+
+	// Test closing with nil connection (should not panic)
+	assert.NotPanics(t, func() {
+		client.Close()
+	})
+}
+
+// TestWebSocketClient_ConnectRateLimit tests connection rate limiting
+func TestWebSocketClient_ConnectRateLimit(t *testing.T) {
+	agent := createTestAgent(t)
+
+	// Set up environment
+	os.Setenv("BESZEL_AGENT_HUB_URL", "http://localhost:8080")
+	os.Setenv("BESZEL_AGENT_TOKEN", "test-token")
+	defer func() {
+		os.Unsetenv("BESZEL_AGENT_HUB_URL")
+		os.Unsetenv("BESZEL_AGENT_TOKEN")
+	}()
+
+	client, err := newWebSocketClient(agent)
+	require.NoError(t, err)
+
+	// Set recent connection attempt
+	client.lastConnectAttempt = time.Now()
+
+	// Test that connection fails quickly due to rate limiting
+	// This won't actually connect but should fail fast
+	err = client.Connect()
+	assert.Error(t, err, "Connection should fail but not hang")
+}
+
+// TestGetToken tests the getToken function with various scenarios
+func TestGetToken(t *testing.T) {
+	unsetEnvVars := func() {
+		os.Unsetenv("BESZEL_AGENT_TOKEN")
+		os.Unsetenv("TOKEN")
+		os.Unsetenv("BESZEL_AGENT_TOKEN_FILE")
+		os.Unsetenv("TOKEN_FILE")
+	}
+
+	t.Run("token from TOKEN environment variable", func(t *testing.T) {
+		unsetEnvVars()
+
+		// Set TOKEN env var
+		expectedToken := "test-token-from-env"
+		os.Setenv("TOKEN", expectedToken)
+		defer os.Unsetenv("TOKEN")
+
+		token, err := getToken()
+		assert.NoError(t, err)
+		assert.Equal(t, expectedToken, token)
+	})
+
+	t.Run("token from BESZEL_AGENT_TOKEN environment variable", func(t *testing.T) {
+		unsetEnvVars()
+
+		// Set BESZEL_AGENT_TOKEN env var (should take precedence)
+		expectedToken := "test-token-from-beszel-env"
+		os.Setenv("BESZEL_AGENT_TOKEN", expectedToken)
+		defer os.Unsetenv("BESZEL_AGENT_TOKEN")
+
+		token, err := getToken()
+		assert.NoError(t, err)
+		assert.Equal(t, expectedToken, token)
+	})
+
+	t.Run("token from TOKEN_FILE", func(t *testing.T) {
+		unsetEnvVars()
+
+		// Create a temporary token file
+		expectedToken := "test-token-from-file"
+		tokenFile, err := os.CreateTemp("", "token-test-*.txt")
+		require.NoError(t, err)
+		defer os.Remove(tokenFile.Name())
+
+		_, err = tokenFile.WriteString(expectedToken)
+		require.NoError(t, err)
+		tokenFile.Close()
+
+		// Set TOKEN_FILE env var
+		os.Setenv("TOKEN_FILE", tokenFile.Name())
+		defer os.Unsetenv("TOKEN_FILE")
+
+		token, err := getToken()
+		assert.NoError(t, err)
+		assert.Equal(t, expectedToken, token)
+	})
+
+	t.Run("token from BESZEL_AGENT_TOKEN_FILE", func(t *testing.T) {
+		unsetEnvVars()
+
+		// Create a temporary token file
+		expectedToken := "test-token-from-beszel-file"
+		tokenFile, err := os.CreateTemp("", "token-test-*.txt")
+		require.NoError(t, err)
+		defer os.Remove(tokenFile.Name())
+
+		_, err = tokenFile.WriteString(expectedToken)
+		require.NoError(t, err)
+		tokenFile.Close()
+
+		// Set BESZEL_AGENT_TOKEN_FILE env var (should take precedence)
+		os.Setenv("BESZEL_AGENT_TOKEN_FILE", tokenFile.Name())
+		defer os.Unsetenv("BESZEL_AGENT_TOKEN_FILE")
+
+		token, err := getToken()
+		assert.NoError(t, err)
+		assert.Equal(t, expectedToken, token)
+	})
+
+	t.Run("TOKEN takes precedence over TOKEN_FILE", func(t *testing.T) {
+		unsetEnvVars()
+
+		// Create a temporary token file
+		fileToken := "token-from-file"
+		tokenFile, err := os.CreateTemp("", "token-test-*.txt")
+		require.NoError(t, err)
+		defer os.Remove(tokenFile.Name())
+
+		_, err = tokenFile.WriteString(fileToken)
+		require.NoError(t, err)
+		tokenFile.Close()
+
+		// Set both TOKEN and TOKEN_FILE
+		envToken := "token-from-env"
+		os.Setenv("TOKEN", envToken)
+		os.Setenv("TOKEN_FILE", tokenFile.Name())
+		defer func() {
+			os.Unsetenv("TOKEN")
+			os.Unsetenv("TOKEN_FILE")
+		}()
+
+		token, err := getToken()
+		assert.NoError(t, err)
+		assert.Equal(t, envToken, token, "TOKEN should take precedence over TOKEN_FILE")
+	})
+
+	t.Run("error when neither TOKEN nor TOKEN_FILE is set", func(t *testing.T) {
+		unsetEnvVars()
+
+		token, err := getToken()
+		assert.Error(t, err)
+		assert.Equal(t, "", token)
+		assert.Contains(t, err.Error(), "must set TOKEN or TOKEN_FILE")
+	})
+
+	t.Run("error when TOKEN_FILE points to non-existent file", func(t *testing.T) {
+		unsetEnvVars()
+
+		// Set TOKEN_FILE to a non-existent file
+		os.Setenv("TOKEN_FILE", "/non/existent/file.txt")
+		defer os.Unsetenv("TOKEN_FILE")
+
+		token, err := getToken()
+		assert.Error(t, err)
+		assert.Equal(t, "", token)
+		assert.Contains(t, err.Error(), "no such file or directory")
+	})
+
+	t.Run("handles empty token file", func(t *testing.T) {
+		unsetEnvVars()
+
+		// Create an empty token file
+		tokenFile, err := os.CreateTemp("", "token-test-*.txt")
+		require.NoError(t, err)
+		defer os.Remove(tokenFile.Name())
+		tokenFile.Close()
+
+		// Set TOKEN_FILE env var
+		os.Setenv("TOKEN_FILE", tokenFile.Name())
+		defer os.Unsetenv("TOKEN_FILE")
+
+		token, err := getToken()
+		assert.NoError(t, err)
+		assert.Equal(t, "", token, "Empty file should return empty string")
+	})
+}

beszel/internal/agent/connection_manager.go

@@ -0,0 +1,220 @@
+package agent
+
+import (
+	"beszel/internal/agent/health"
+	"errors"
+	"log/slog"
+	"os"
+	"os/signal"
+	"syscall"
+	"time"
+)
+
+// ConnectionManager manages the connection state and events for the agent.
+// It handles both WebSocket and SSH connections, automatically switching between
+// them based on availability and managing reconnection attempts.
+type ConnectionManager struct {
+	agent         *Agent               // Reference to the parent agent
+	State         ConnectionState      // Current connection state
+	eventChan     chan ConnectionEvent // Channel for connection events
+	wsClient      *WebSocketClient     // WebSocket client for hub communication
+	serverOptions ServerOptions        // Configuration for SSH server
+	wsTicker      *time.Ticker         // Ticker for WebSocket connection attempts
+	isConnecting  bool                 // Prevents multiple simultaneous reconnection attempts
+}
+
+// ConnectionState represents the current connection state of the agent.
+type ConnectionState uint8
+
+// ConnectionEvent represents connection-related events that can occur.
+type ConnectionEvent uint8
+
+// Connection states
+const (
+	Disconnected       ConnectionState = iota // No active connection
+	WebSocketConnected                        // Connected via WebSocket
+	SSHConnected                              // Connected via SSH
+)
+
+// Connection events
+const (
+	WebSocketConnect    ConnectionEvent = iota // WebSocket connection established
+	WebSocketDisconnect                        // WebSocket connection lost
+	SSHConnect                                 // SSH connection established
+	SSHDisconnect                              // SSH connection lost
+)
+
+const wsTickerInterval = 10 * time.Second
+
+// newConnectionManager creates a new connection manager for the given agent.
+func newConnectionManager(agent *Agent) *ConnectionManager {
+	cm := &ConnectionManager{
+		agent: agent,
+		State: Disconnected,
+	}
+	return cm
+}
+
+// startWsTicker starts or resets the WebSocket connection attempt ticker.
+func (c *ConnectionManager) startWsTicker() {
+	if c.wsTicker == nil {
+		c.wsTicker = time.NewTicker(wsTickerInterval)
+	} else {
+		c.wsTicker.Reset(wsTickerInterval)
+	}
+}
+
+// stopWsTicker stops the WebSocket connection attempt ticker.
+func (c *ConnectionManager) stopWsTicker() {
+	if c.wsTicker != nil {
+		c.wsTicker.Stop()
+	}
+}
+
+// Start begins connection attempts and enters the main event loop.
+// It handles connection events, periodic health updates, and graceful shutdown.
+func (c *ConnectionManager) Start(serverOptions ServerOptions) error {
+	if c.eventChan != nil {
+		return errors.New("already started")
+	}
+
+	wsClient, err := newWebSocketClient(c.agent)
+	if err != nil {
+		slog.Warn("Error creating WebSocket client", "err", err)
+	}
+	c.wsClient = wsClient
+
+	c.serverOptions = serverOptions
+	c.eventChan = make(chan ConnectionEvent, 1)
+
+	// signal handling for shutdown
+	sigChan := make(chan os.Signal, 1)
+	signal.Notify(sigChan, syscall.SIGINT, syscall.SIGTERM)
+
+	c.startWsTicker()
+	c.connect()
+
+	// update health status immediately and every 90 seconds
+	_ = health.Update()
+	healthTicker := time.Tick(90 * time.Second)
+
+	for {
+		select {
+		case connectionEvent := <-c.eventChan:
+			c.handleEvent(connectionEvent)
+		case <-c.wsTicker.C:
+			_ = c.startWebSocketConnection()
+		case <-healthTicker:
+			_ = health.Update()
+		case <-sigChan:
+			slog.Info("Shutting down")
+			_ = c.agent.StopServer()
+			c.closeWebSocket()
+			return health.CleanUp()
+		}
+	}
+}
+
+// handleEvent processes connection events and updates the connection state accordingly.
+func (c *ConnectionManager) handleEvent(event ConnectionEvent) {
+	switch event {
+	case WebSocketConnect:
+		c.handleStateChange(WebSocketConnected)
+	case SSHConnect:
+		c.handleStateChange(SSHConnected)
+	case WebSocketDisconnect:
+		if c.State == WebSocketConnected {
+			c.handleStateChange(Disconnected)
+		}
+	case SSHDisconnect:
+		if c.State == SSHConnected {
+			c.handleStateChange(Disconnected)
+		}
+	}
+}
+
+// handleStateChange updates the connection state and performs necessary actions
+// based on the new state, including stopping services and initiating reconnections.
+func (c *ConnectionManager) handleStateChange(newState ConnectionState) {
+	if c.State == newState {
+		return
+	}
+	c.State = newState
+	switch newState {
+	case WebSocketConnected:
+		slog.Info("WebSocket connected", "host", c.wsClient.hubURL.Host)
+		c.stopWsTicker()
+		_ = c.agent.StopServer()
+		c.isConnecting = false
+	case SSHConnected:
+		// stop new ws connection attempts
+		slog.Info("SSH connection established")
+		c.stopWsTicker()
+		c.isConnecting = false
+	case Disconnected:
+		if c.isConnecting {
+			// Already handling reconnection, avoid duplicate attempts
+			return
+		}
+		c.isConnecting = true
+		slog.Warn("Disconnected from hub")
+		// make sure old ws connection is closed
+		c.closeWebSocket()
+		// reconnect
+		go c.connect()
+	}
+}
+
+// connect handles the connection logic with proper delays and priority.
+// It attempts WebSocket connection first, falling back to SSH server if needed.
+func (c *ConnectionManager) connect() {
+	c.isConnecting = true
+	defer func() {
+		c.isConnecting = false
+	}()
+
+	if c.wsClient != nil && time.Since(c.wsClient.lastConnectAttempt) < 5*time.Second {
+		time.Sleep(5 * time.Second)
+	}
+
+	// Try WebSocket first, if it fails, start SSH server
+	err := c.startWebSocketConnection()
+	if err != nil && c.State == Disconnected {
+		c.startSSHServer()
+		c.startWsTicker()
+	}
+}
+
+// startWebSocketConnection attempts to establish a WebSocket connection to the hub.
+func (c *ConnectionManager) startWebSocketConnection() error {
+	if c.State != Disconnected {
+		return errors.New("already connected")
+	}
+	if c.wsClient == nil {
+		return errors.New("WebSocket client not initialized")
+	}
+	if time.Since(c.wsClient.lastConnectAttempt) < 5*time.Second {
+		return errors.New("already connecting")
+	}
+
+	err := c.wsClient.Connect()
+	if err != nil {
+		slog.Warn("WebSocket connection failed", "err", err)
+		c.closeWebSocket()
+	}
+	return err
+}
+
+// startSSHServer starts the SSH server if the agent is currently disconnected.
+func (c *ConnectionManager) startSSHServer() {
+	if c.State == Disconnected {
+		go c.agent.StartServer(c.serverOptions)
+	}
+}
+
+// closeWebSocket closes the WebSocket connection if it exists.
+func (c *ConnectionManager) closeWebSocket() {
+	if c.wsClient != nil {
+		c.wsClient.Close()
+	}
+}

beszel/internal/agent/connection_manager_test.go

@@ -0,0 +1,315 @@
+//go:build testing
+// +build testing
+
+package agent
+
+import (
+	"crypto/ed25519"
+	"fmt"
+	"net"
+	"net/url"
+	"os"
+	"testing"
+	"time"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+	"golang.org/x/crypto/ssh"
+)
+
+func createTestAgent(t *testing.T) *Agent {
+	dataDir := t.TempDir()
+	agent, err := NewAgent(dataDir)
+	require.NoError(t, err)
+	return agent
+}
+
+func createTestServerOptions(t *testing.T) ServerOptions {
+	// Generate test key pair
+	_, privKey, err := ed25519.GenerateKey(nil)
+	require.NoError(t, err)
+	sshPubKey, err := ssh.NewPublicKey(privKey.Public().(ed25519.PublicKey))
+	require.NoError(t, err)
+
+	// Find available port
+	listener, err := net.Listen("tcp", "127.0.0.1:0")
+	require.NoError(t, err)
+	port := listener.Addr().(*net.TCPAddr).Port
+	listener.Close()
+
+	return ServerOptions{
+		Network: "tcp",
+		Addr:    fmt.Sprintf("127.0.0.1:%d", port),
+		Keys:    []ssh.PublicKey{sshPubKey},
+	}
+}
+
+// TestConnectionManager_NewConnectionManager tests connection manager creation
+func TestConnectionManager_NewConnectionManager(t *testing.T) {
+	agent := createTestAgent(t)
+	cm := newConnectionManager(agent)
+
+	assert.NotNil(t, cm, "Connection manager should not be nil")
+	assert.Equal(t, agent, cm.agent, "Agent reference should be set")
+	assert.Equal(t, Disconnected, cm.State, "Initial state should be Disconnected")
+	assert.Nil(t, cm.eventChan, "Event channel should be nil initially")
+	assert.Nil(t, cm.wsClient, "WebSocket client should be nil initially")
+	assert.Nil(t, cm.wsTicker, "WebSocket ticker should be nil initially")
+	assert.False(t, cm.isConnecting, "isConnecting should be false initially")
+}
+
+// TestConnectionManager_StateTransitions tests basic state transitions
+func TestConnectionManager_StateTransitions(t *testing.T) {
+	agent := createTestAgent(t)
+	cm := agent.connectionManager
+	initialState := cm.State
+	cm.wsClient = &WebSocketClient{
+		hubURL: &url.URL{
+			Host: "localhost:8080",
+		},
+	}
+	assert.NotNil(t, cm, "Connection manager should not be nil")
+	assert.Equal(t, Disconnected, initialState, "Initial state should be Disconnected")
+
+	// Test state transitions
+	cm.handleStateChange(WebSocketConnected)
+	assert.Equal(t, WebSocketConnected, cm.State, "State should change to WebSocketConnected")
+
+	cm.handleStateChange(SSHConnected)
+	assert.Equal(t, SSHConnected, cm.State, "State should change to SSHConnected")
+
+	cm.handleStateChange(Disconnected)
+	assert.Equal(t, Disconnected, cm.State, "State should change to Disconnected")
+
+	// Test that same state doesn't trigger changes
+	cm.State = WebSocketConnected
+	cm.handleStateChange(WebSocketConnected)
+	assert.Equal(t, WebSocketConnected, cm.State, "Same state should not trigger change")
+}
+
+// TestConnectionManager_EventHandling tests event handling logic
+func TestConnectionManager_EventHandling(t *testing.T) {
+	agent := createTestAgent(t)
+	cm := agent.connectionManager
+	cm.wsClient = &WebSocketClient{
+		hubURL: &url.URL{
+			Host: "localhost:8080",
+		},
+	}
+
+	testCases := []struct {
+		name          string
+		initialState  ConnectionState
+		event         ConnectionEvent
+		expectedState ConnectionState
+	}{
+		{
+			name:          "WebSocket connect from disconnected",
+			initialState:  Disconnected,
+			event:         WebSocketConnect,
+			expectedState: WebSocketConnected,
+		},
+		{
+			name:          "SSH connect from disconnected",
+			initialState:  Disconnected,
+			event:         SSHConnect,
+			expectedState: SSHConnected,
+		},
+		{
+			name:          "WebSocket disconnect from connected",
+			initialState:  WebSocketConnected,
+			event:         WebSocketDisconnect,
+			expectedState: Disconnected,
+		},
+		{
+			name:          "SSH disconnect from connected",
+			initialState:  SSHConnected,
+			event:         SSHDisconnect,
+			expectedState: Disconnected,
+		},
+		{
+			name:          "WebSocket disconnect from SSH connected (no change)",
+			initialState:  SSHConnected,
+			event:         WebSocketDisconnect,
+			expectedState: SSHConnected,
+		},
+		{
+			name:          "SSH disconnect from WebSocket connected (no change)",
+			initialState:  WebSocketConnected,
+			event:         SSHDisconnect,
+			expectedState: WebSocketConnected,
+		},
+	}
+
+	for _, tc := range testCases {
+		t.Run(tc.name, func(t *testing.T) {
+			cm.State = tc.initialState
+			cm.handleEvent(tc.event)
+			assert.Equal(t, tc.expectedState, cm.State, "State should match expected after event")
+		})
+	}
+}
+
+// TestConnectionManager_TickerManagement tests WebSocket ticker management
+func TestConnectionManager_TickerManagement(t *testing.T) {
+	agent := createTestAgent(t)
+	cm := agent.connectionManager
+
+	// Test starting ticker
+	cm.startWsTicker()
+	assert.NotNil(t, cm.wsTicker, "Ticker should be created")
+
+	// Test stopping ticker (should not panic)
+	assert.NotPanics(t, func() {
+		cm.stopWsTicker()
+	}, "Stopping ticker should not panic")
+
+	// Test stopping nil ticker (should not panic)
+	cm.wsTicker = nil
+	assert.NotPanics(t, func() {
+		cm.stopWsTicker()
+	}, "Stopping nil ticker should not panic")
+
+	// Test restarting ticker
+	cm.startWsTicker()
+	assert.NotNil(t, cm.wsTicker, "Ticker should be recreated")
+
+	// Test resetting existing ticker
+	firstTicker := cm.wsTicker
+	cm.startWsTicker()
+	assert.Equal(t, firstTicker, cm.wsTicker, "Same ticker instance should be reused")
+
+	cm.stopWsTicker()
+}
+
+// TestConnectionManager_WebSocketConnectionFlow tests WebSocket connection logic
+func TestConnectionManager_WebSocketConnectionFlow(t *testing.T) {
+	if testing.Short() {
+		t.Skip("Skipping WebSocket connection test in short mode")
+	}
+
+	agent := createTestAgent(t)
+	cm := agent.connectionManager
+
+	// Test WebSocket connection without proper environment
+	err := cm.startWebSocketConnection()
+	assert.Error(t, err, "WebSocket connection should fail without proper environment")
+	assert.Equal(t, Disconnected, cm.State, "State should remain Disconnected after failed connection")
+
+	// Test with invalid URL
+	os.Setenv("BESZEL_AGENT_HUB_URL", "invalid-url")
+	os.Setenv("BESZEL_AGENT_TOKEN", "test-token")
+	defer func() {
+		os.Unsetenv("BESZEL_AGENT_HUB_URL")
+		os.Unsetenv("BESZEL_AGENT_TOKEN")
+	}()
+
+	// Test with missing token
+	os.Setenv("BESZEL_AGENT_HUB_URL", "http://localhost:8080")
+	os.Unsetenv("BESZEL_AGENT_TOKEN")
+
+	_, err2 := newWebSocketClient(agent)
+	assert.Error(t, err2, "WebSocket client creation should fail without token")
+}
+
+// TestConnectionManager_ReconnectionLogic tests reconnection prevention logic
+func TestConnectionManager_ReconnectionLogic(t *testing.T) {
+	agent := createTestAgent(t)
+	cm := agent.connectionManager
+	cm.eventChan = make(chan ConnectionEvent, 1)
+
+	// Test that isConnecting flag prevents duplicate reconnection attempts
+	// Start from connected state, then simulate disconnect
+	cm.State = WebSocketConnected
+	cm.isConnecting = false
+
+	// First disconnect should trigger reconnection logic
+	cm.handleStateChange(Disconnected)
+	assert.Equal(t, Disconnected, cm.State, "Should change to disconnected")
+	assert.True(t, cm.isConnecting, "Should set isConnecting flag")
+}
+
+// TestConnectionManager_ConnectWithRateLimit tests connection rate limiting
+func TestConnectionManager_ConnectWithRateLimit(t *testing.T) {
+	agent := createTestAgent(t)
+	cm := agent.connectionManager
+
+	// Set up environment for WebSocket client creation
+	os.Setenv("BESZEL_AGENT_HUB_URL", "ws://localhost:8080")
+	os.Setenv("BESZEL_AGENT_TOKEN", "test-token")
+	defer func() {
+		os.Unsetenv("BESZEL_AGENT_HUB_URL")
+		os.Unsetenv("BESZEL_AGENT_TOKEN")
+	}()
+
+	// Create WebSocket client
+	wsClient, err := newWebSocketClient(agent)
+	require.NoError(t, err)
+	cm.wsClient = wsClient
+
+	// Set recent connection attempt
+	cm.wsClient.lastConnectAttempt = time.Now()
+
+	// Test that connection is rate limited
+	err = cm.startWebSocketConnection()
+	assert.Error(t, err, "Should error due to rate limiting")
+	assert.Contains(t, err.Error(), "already connecting", "Error should indicate rate limiting")
+
+	// Test connection after rate limit expires
+	cm.wsClient.lastConnectAttempt = time.Now().Add(-10 * time.Second)
+	err = cm.startWebSocketConnection()
+	// This will fail due to no actual server, but should not be rate limited
+	assert.Error(t, err, "Connection should fail but not due to rate limiting")
+	assert.NotContains(t, err.Error(), "already connecting", "Error should not indicate rate limiting")
+}
+
+// TestConnectionManager_StartWithInvalidConfig tests starting with invalid configuration
+func TestConnectionManager_StartWithInvalidConfig(t *testing.T) {
+	agent := createTestAgent(t)
+	cm := agent.connectionManager
+	serverOptions := createTestServerOptions(t)
+
+	// Test starting when already started
+	cm.eventChan = make(chan ConnectionEvent, 5)
+	err := cm.Start(serverOptions)
+	assert.Error(t, err, "Should error when starting already started connection manager")
+}
+
+// TestConnectionManager_CloseWebSocket tests WebSocket closing
+func TestConnectionManager_CloseWebSocket(t *testing.T) {
+	agent := createTestAgent(t)
+	cm := agent.connectionManager
+
+	// Test closing when no WebSocket client exists
+	assert.NotPanics(t, func() {
+		cm.closeWebSocket()
+	}, "Should not panic when closing nil WebSocket client")
+
+	// Set up environment and create WebSocket client
+	os.Setenv("BESZEL_AGENT_HUB_URL", "ws://localhost:8080")
+	os.Setenv("BESZEL_AGENT_TOKEN", "test-token")
+	defer func() {
+		os.Unsetenv("BESZEL_AGENT_HUB_URL")
+		os.Unsetenv("BESZEL_AGENT_TOKEN")
+	}()
+
+	wsClient, err := newWebSocketClient(agent)
+	require.NoError(t, err)
+	cm.wsClient = wsClient
+
+	// Test closing when WebSocket client exists
+	assert.NotPanics(t, func() {
+		cm.closeWebSocket()
+	}, "Should not panic when closing WebSocket client")
+}
+
+// TestConnectionManager_ConnectFlow tests the connect method
+func TestConnectionManager_ConnectFlow(t *testing.T) {
+	agent := createTestAgent(t)
+	cm := agent.connectionManager
+
+	// Test connect without WebSocket client
+	assert.NotPanics(t, func() {
+		cm.connect()
+	}, "Connect should not panic without WebSocket client")
+}

beszel/internal/agent/data_dir.go

@@ -0,0 +1,117 @@
+package agent
+
+import (
+	"errors"
+	"fmt"
+	"os"
+	"path/filepath"
+	"runtime"
+)
+
+// getDataDir returns the path to the data directory for the agent and an error
+// if the directory is not valid. Attempts to find the optimal data directory if
+// no data directories are provided.
+func getDataDir(dataDirs ...string) (string, error) {
+	if len(dataDirs) > 0 {
+		return testDataDirs(dataDirs)
+	}
+
+	dataDir, _ := GetEnv("DATA_DIR")
+	if dataDir != "" {
+		dataDirs = append(dataDirs, dataDir)
+	}
+
+	if runtime.GOOS == "windows" {
+		dataDirs = append(dataDirs,
+			filepath.Join(os.Getenv("APPDATA"), "beszel-agent"),
+			filepath.Join(os.Getenv("LOCALAPPDATA"), "beszel-agent"),
+		)
+	} else {
+		dataDirs = append(dataDirs, "/var/lib/beszel-agent")
+		if homeDir, err := os.UserHomeDir(); err == nil {
+			dataDirs = append(dataDirs, filepath.Join(homeDir, ".config", "beszel"))
+		}
+	}
+	return testDataDirs(dataDirs)
+}
+
+func testDataDirs(paths []string) (string, error) {
+	// first check if the directory exists and is writable
+	for _, path := range paths {
+		if valid, _ := isValidDataDir(path, false); valid {
+			return path, nil
+		}
+	}
+	// if the directory doesn't exist, try to create it
+	for _, path := range paths {
+		exists, _ := directoryExists(path)
+		if exists {
+			continue
+		}
+
+		if err := os.MkdirAll(path, 0755); err != nil {
+			continue
+		}
+
+		// Verify the created directory is actually writable
+		writable, _ := directoryIsWritable(path)
+		if !writable {
+			continue
+		}
+
+		return path, nil
+	}
+
+	return "", errors.New("data directory not found")
+}
+
+func isValidDataDir(path string, createIfNotExists bool) (bool, error) {
+	exists, err := directoryExists(path)
+	if err != nil {
+		return false, err
+	}
+
+	if !exists {
+		if !createIfNotExists {
+			return false, nil
+		}
+		if err = os.MkdirAll(path, 0755); err != nil {
+			return false, err
+		}
+	}
+
+	// Always check if the directory is writable
+	writable, err := directoryIsWritable(path)
+	if err != nil {
+		return false, err
+	}
+	return writable, nil
+}
+
+// directoryExists checks if a directory exists
+func directoryExists(path string) (bool, error) {
+	// Check if directory exists
+	stat, err := os.Stat(path)
+	if err != nil {
+		if os.IsNotExist(err) {
+			return false, nil
+		}
+		return false, err
+	}
+	if !stat.IsDir() {
+		return false, fmt.Errorf("%s is not a directory", path)
+	}
+	return true, nil
+}
+
+// directoryIsWritable tests if a directory is writable by creating and removing a temporary file
+func directoryIsWritable(path string) (bool, error) {
+	testFile := filepath.Join(path, ".write-test")
+	file, err := os.Create(testFile)
+	if err != nil {
+		return false, err
+	}
+	defer file.Close()
+	defer os.Remove(testFile)
+	return true, nil
+}

beszel/internal/agent/data_dir_test.go

@@ -0,0 +1,263 @@
+//go:build testing
+// +build testing
+
+package agent
+
+import (
+	"os"
+	"path/filepath"
+	"runtime"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func TestGetDataDir(t *testing.T) {
+	// Test with explicit dataDir parameter
+	t.Run("explicit data dir", func(t *testing.T) {
+		tempDir := t.TempDir()
+		result, err := getDataDir(tempDir)
+		require.NoError(t, err)
+		assert.Equal(t, tempDir, result)
+	})
+
+	// Test with explicit non-existent dataDir that can be created
+	t.Run("explicit data dir - create new", func(t *testing.T) {
+		tempDir := t.TempDir()
+		newDir := filepath.Join(tempDir, "new-data-dir")
+		result, err := getDataDir(newDir)
+		require.NoError(t, err)
+		assert.Equal(t, newDir, result)
+
+		// Verify directory was created
+		stat, err := os.Stat(newDir)
+		require.NoError(t, err)
+		assert.True(t, stat.IsDir())
+	})
+
+	// Test with DATA_DIR environment variable
+	t.Run("DATA_DIR environment variable", func(t *testing.T) {
+		tempDir := t.TempDir()
+
+		// Set environment variable
+		oldValue := os.Getenv("DATA_DIR")
+		defer func() {
+			if oldValue == "" {
+				os.Unsetenv("BESZEL_AGENT_DATA_DIR")
+			} else {
+				os.Setenv("BESZEL_AGENT_DATA_DIR", oldValue)
+			}
+		}()
+
+		os.Setenv("BESZEL_AGENT_DATA_DIR", tempDir)
+
+		result, err := getDataDir()
+		require.NoError(t, err)
+		assert.Equal(t, tempDir, result)
+	})
+
+	// Test with invalid explicit dataDir
+	t.Run("invalid explicit data dir", func(t *testing.T) {
+		invalidPath := "/invalid/path/that/cannot/be/created"
+		_, err := getDataDir(invalidPath)
+		assert.Error(t, err)
+	})
+
+	// Test fallback behavior (empty dataDir, no env var)
+	t.Run("fallback to default directories", func(t *testing.T) {
+		// Clear DATA_DIR environment variable
+		oldValue := os.Getenv("DATA_DIR")
+		defer func() {
+			if oldValue == "" {
+				os.Unsetenv("DATA_DIR")
+			} else {
+				os.Setenv("DATA_DIR", oldValue)
+			}
+		}()
+		os.Unsetenv("DATA_DIR")
+
+		// This will try platform-specific defaults, which may or may not work
+		// We're mainly testing that it doesn't panic and returns some result
+		result, err := getDataDir()
+		// We don't assert success/failure here since it depends on system permissions
+		// Just verify we get a string result if no error
+		if err == nil {
+			assert.NotEmpty(t, result)
+		}
+	})
+}
+
+func TestTestDataDirs(t *testing.T) {
+	// Test with existing valid directory
+	t.Run("existing valid directory", func(t *testing.T) {
+		tempDir := t.TempDir()
+		result, err := testDataDirs([]string{tempDir})
+		require.NoError(t, err)
+		assert.Equal(t, tempDir, result)
+	})
+
+	// Test with multiple directories, first one valid
+	t.Run("multiple dirs - first valid", func(t *testing.T) {
+		tempDir := t.TempDir()
+		invalidDir := "/invalid/path"
+		result, err := testDataDirs([]string{tempDir, invalidDir})
+		require.NoError(t, err)
+		assert.Equal(t, tempDir, result)
+	})
+
+	// Test with multiple directories, second one valid
+	t.Run("multiple dirs - second valid", func(t *testing.T) {
+		tempDir := t.TempDir()
+		invalidDir := "/invalid/path"
+		result, err := testDataDirs([]string{invalidDir, tempDir})
+		require.NoError(t, err)
+		assert.Equal(t, tempDir, result)
+	})
+
+	// Test with non-existing directory that can be created
+	t.Run("create new directory", func(t *testing.T) {
+		tempDir := t.TempDir()
+		newDir := filepath.Join(tempDir, "new-dir")
+		result, err := testDataDirs([]string{newDir})
+		require.NoError(t, err)
+		assert.Equal(t, newDir, result)
+
+		// Verify directory was created
+		stat, err := os.Stat(newDir)
+		require.NoError(t, err)
+		assert.True(t, stat.IsDir())
+	})
+
+	// Test with no valid directories
+	t.Run("no valid directories", func(t *testing.T) {
+		invalidPaths := []string{"/invalid/path1", "/invalid/path2"}
+		_, err := testDataDirs(invalidPaths)
+		assert.Error(t, err)
+		assert.Contains(t, err.Error(), "data directory not found")
+	})
+}
+
+func TestIsValidDataDir(t *testing.T) {
+	// Test with existing directory
+	t.Run("existing directory", func(t *testing.T) {
+		tempDir := t.TempDir()
+		valid, err := isValidDataDir(tempDir, false)
+		require.NoError(t, err)
+		assert.True(t, valid)
+	})
+
+	// Test with non-existing directory, createIfNotExists=false
+	t.Run("non-existing dir - no create", func(t *testing.T) {
+		tempDir := t.TempDir()
+		nonExistentDir := filepath.Join(tempDir, "does-not-exist")
+		valid, err := isValidDataDir(nonExistentDir, false)
+		require.NoError(t, err)
+		assert.False(t, valid)
+	})
+
+	// Test with non-existing directory, createIfNotExists=true
+	t.Run("non-existing dir - create", func(t *testing.T) {
+		tempDir := t.TempDir()
+		newDir := filepath.Join(tempDir, "new-dir")
+		valid, err := isValidDataDir(newDir, true)
+		require.NoError(t, err)
+		assert.True(t, valid)
+
+		// Verify directory was created
+		stat, err := os.Stat(newDir)
+		require.NoError(t, err)
+		assert.True(t, stat.IsDir())
+	})
+
+	// Test with file instead of directory
+	t.Run("file instead of directory", func(t *testing.T) {
+		tempDir := t.TempDir()
+		tempFile := filepath.Join(tempDir, "testfile")
+		err := os.WriteFile(tempFile, []byte("test"), 0644)
+		require.NoError(t, err)
+
+		valid, err := isValidDataDir(tempFile, false)
+		assert.Error(t, err)
+		assert.False(t, valid)
+		assert.Contains(t, err.Error(), "is not a directory")
+	})
+}
+
+func TestDirectoryExists(t *testing.T) {
+	// Test with existing directory
+	t.Run("existing directory", func(t *testing.T) {
+		tempDir := t.TempDir()
+		exists, err := directoryExists(tempDir)
+		require.NoError(t, err)
+		assert.True(t, exists)
+	})
+
+	// Test with non-existing directory
+	t.Run("non-existing directory", func(t *testing.T) {
+		tempDir := t.TempDir()
+		nonExistentDir := filepath.Join(tempDir, "does-not-exist")
+		exists, err := directoryExists(nonExistentDir)
+		require.NoError(t, err)
+		assert.False(t, exists)
+	})
+
+	// Test with file instead of directory
+	t.Run("file instead of directory", func(t *testing.T) {
+		tempDir := t.TempDir()
+		tempFile := filepath.Join(tempDir, "testfile")
+		err := os.WriteFile(tempFile, []byte("test"), 0644)
+		require.NoError(t, err)
+
+		exists, err := directoryExists(tempFile)
+		assert.Error(t, err)
+		assert.False(t, exists)
+		assert.Contains(t, err.Error(), "is not a directory")
+	})
+}
+
+func TestDirectoryIsWritable(t *testing.T) {
+	// Test with writable directory
+	t.Run("writable directory", func(t *testing.T) {
+		tempDir := t.TempDir()
+		writable, err := directoryIsWritable(tempDir)
+		require.NoError(t, err)
+		assert.True(t, writable)
+	})
+
+	// Test with non-existing directory
+	t.Run("non-existing directory", func(t *testing.T) {
+		tempDir := t.TempDir()
+		nonExistentDir := filepath.Join(tempDir, "does-not-exist")
+		writable, err := directoryIsWritable(nonExistentDir)
+		assert.Error(t, err)
+		assert.False(t, writable)
+	})
+
+	// Test with non-writable directory (Unix-like systems only)
+	t.Run("non-writable directory", func(t *testing.T) {
+		if runtime.GOOS != "linux" && runtime.GOOS != "darwin" {
+			t.Skip("Skipping non-writable directory test on", runtime.GOOS)
+		}
+
+		tempDir := t.TempDir()
+		readOnlyDir := filepath.Join(tempDir, "readonly")
+
+		// Create the directory
+		err := os.Mkdir(readOnlyDir, 0755)
+		require.NoError(t, err)
+
+		// Make it read-only
+		err = os.Chmod(readOnlyDir, 0444)
+		require.NoError(t, err)
+
+		// Restore permissions after test for cleanup
+		defer func() {
+			os.Chmod(readOnlyDir, 0755)
+		}()
+
+		writable, err := directoryIsWritable(readOnlyDir)
+		assert.Error(t, err)
+		assert.False(t, writable)
+	})
+}

beszel/internal/agent/docker.go

@@ -2,6 +2,7 @@ package agent
 
 import (
 	"beszel/internal/entities/container"
+	"bytes"
 	"context"
 	"encoding/json"
 	"fmt"
@@ -27,6 +28,9 @@ type dockerManager struct {
 	validIds            map[string]struct{}         // Map of valid container ids, used to prune invalid containers from containerStatsMap
 	goodDockerVersion   bool                        // Whether docker version is at least 25.0.0 (one-shot works correctly)
 	isWindows           bool                        // Whether the Docker Engine API is running on Windows
+	buf                 *bytes.Buffer               // Buffer to store and read response bodies
+	decoder             *json.Decoder               // Reusable JSON decoder that reads from buf
+	apiStats            *container.ApiStats         // Reusable API stats object
 }
 
 // userAgentRoundTripper is a custom http.RoundTripper that adds a User-Agent header to all requests
@@ -63,10 +67,9 @@ func (dm *dockerManager) getDockerStats() ([]*container.Stats, error) {
 	if err != nil {
 		return nil, err
 	}
-	defer resp.Body.Close()
 
 	dm.apiContainerList = dm.apiContainerList[:0]
-	if err := json.NewDecoder(resp.Body).Decode(&dm.apiContainerList); err != nil {
+	if err := dm.decode(resp, &dm.apiContainerList); err != nil {
 		return nil, err
 	}
 
@@ -83,7 +86,8 @@ func (dm *dockerManager) getDockerStats() ([]*container.Stats, error) {
 
 	var failedContainers []*container.ApiInfo
 
-	for _, ctr := range dm.apiContainerList {
+	for i := range dm.apiContainerList {
+		ctr := dm.apiContainerList[i]
 		ctr.IdShort = ctr.Id[:12]
 		dm.validIds[ctr.IdShort] = struct{}{}
 		// check if container is less than 1 minute old (possible restart)
@@ -111,7 +115,8 @@ func (dm *dockerManager) getDockerStats() ([]*container.Stats, error) {
 	// retry failed containers separately so we can run them in parallel (docker 24 bug)
 	if len(failedContainers) > 0 {
 		slog.Debug("Retrying failed containers", "count", len(failedContainers))
-		for _, ctr := range failedContainers {
+		for i := range failedContainers {
+			ctr := failedContainers[i]
 			dm.queue()
 			go func() {
 				defer dm.dequeue()
@@ -164,8 +169,13 @@ func (dm *dockerManager) updateContainerStats(ctr *container.ApiInfo) error {
 	stats.NetworkRecv = 0
 
 	// docker host container stats response
-	var res container.ApiStats
+	// res := dm.getApiStats()
-	if err := json.NewDecoder(resp.Body).Decode(&res); err != nil {
+	// defer dm.putApiStats(res)
+	//
+
+	res := dm.apiStats
+	res.Networks = nil
+	if err := dm.decode(resp, res); err != nil {
 		return err
 	}
 
@@ -173,9 +183,14 @@ func (dm *dockerManager) updateContainerStats(ctr *container.ApiInfo) error {
 	var usedMemory uint64
 	var cpuPct float64
 
+	// store current cpu stats
+	prevCpuContainer, prevCpuSystem := stats.CpuContainer, stats.CpuSystem
+	stats.CpuContainer = res.CPUStats.CPUUsage.TotalUsage
+	stats.CpuSystem = res.CPUStats.SystemUsage
+
 	if dm.isWindows {
 		usedMemory = res.MemoryStats.PrivateWorkingSet
-		cpuPct = res.CalculateCpuPercentWindows(stats.PrevCpu[0], stats.PrevRead)
+		cpuPct = res.CalculateCpuPercentWindows(prevCpuContainer, stats.PrevReadTime)
 	} else {
 		// check if container has valid data, otherwise may be in restart loop (#103)
 		if res.MemoryStats.Usage == 0 {
@@ -187,13 +202,12 @@ func (dm *dockerManager) updateContainerStats(ctr *container.ApiInfo) error {
 		}
 		usedMemory = res.MemoryStats.Usage - memCache
 
-		cpuPct = res.CalculateCpuPercentLinux(stats.PrevCpu)
+		cpuPct = res.CalculateCpuPercentLinux(prevCpuContainer, prevCpuSystem)
 	}
 
 	if cpuPct > 100 {
 		return fmt.Errorf("%s cpu pct greater than 100: %+v", name, cpuPct)
 	}
-	stats.PrevCpu = [2]uint64{res.CPUStats.CPUUsage.TotalUsage, res.CPUStats.SystemUsage}
 
 	// network
 	var total_sent, total_recv uint64
@@ -201,21 +215,25 @@ func (dm *dockerManager) updateContainerStats(ctr *container.ApiInfo) error {
 		total_sent += v.TxBytes
 		total_recv += v.RxBytes
 	}
-	var sent_delta, recv_delta float64
+	var sent_delta, recv_delta uint64
-	// prevent first run from sending all prev sent/recv bytes
+	millisecondsElapsed := uint64(time.Since(stats.PrevReadTime).Milliseconds())
-	if initialized {
+	if initialized && millisecondsElapsed > 0 {
-		secondsElapsed := time.Since(stats.PrevRead).Seconds()
+		// get bytes per second
-		sent_delta = float64(total_sent-stats.PrevNet.Sent) / secondsElapsed
+		sent_delta = (total_sent - stats.PrevNet.Sent) * 1000 / millisecondsElapsed
-		recv_delta = float64(total_recv-stats.PrevNet.Recv) / secondsElapsed
+		recv_delta = (total_recv - stats.PrevNet.Recv) * 1000 / millisecondsElapsed
+		// check for unrealistic network values (> 5GB/s)
+		if sent_delta > 5e9 || recv_delta > 5e9 {
+			slog.Warn("Bad network delta", "container", name)
+			sent_delta, recv_delta = 0, 0
+		}
 	}
-	stats.PrevNet.Sent = total_sent
+	stats.PrevNet.Sent, stats.PrevNet.Recv = total_sent, total_recv
-	stats.PrevNet.Recv = total_recv
 
 	stats.Cpu = twoDecimals(cpuPct)
 	stats.Mem = bytesToMegabytes(float64(usedMemory))
-	stats.NetworkSent = bytesToMegabytes(sent_delta)
+	stats.NetworkSent = bytesToMegabytes(float64(sent_delta))
-	stats.NetworkRecv = bytesToMegabytes(recv_delta)
+	stats.NetworkRecv = bytesToMegabytes(float64(recv_delta))
-	stats.PrevRead = res.Read
+	stats.PrevReadTime = res.Read
 
 	return nil
 }
@@ -231,7 +249,6 @@ func (dm *dockerManager) deleteContainerStatsSync(id string) {
 func newDockerManager(a *Agent) *dockerManager {
 	dockerHost, exists := GetEnv("DOCKER_HOST")
 	if exists {
-		slog.Info("DOCKER_HOST", "host", dockerHost)
 		// return nil if set to empty string
 		if dockerHost == "" {
 			return nil
@@ -242,7 +259,6 @@ func newDockerManager(a *Agent) *dockerManager {
 
 	parsedURL, err := url.Parse(dockerHost)
 	if err != nil {
-		slog.Error("Error parsing DOCKER_HOST", "err", err)
 		os.Exit(1)
 	}
 
@@ -290,6 +306,7 @@ func newDockerManager(a *Agent) *dockerManager {
 		containerStatsMap: make(map[string]*container.Stats),
 		sem:               make(chan struct{}, 5),
 		apiContainerList:  []*container.ApiInfo{},
+		apiStats:          &container.ApiStats{},
 	}
 
 	// If using podman, return client
@@ -308,9 +325,8 @@ func newDockerManager(a *Agent) *dockerManager {
 	if err != nil {
 		return manager
 	}
-	defer resp.Body.Close()
 
-	if err := json.NewDecoder(resp.Body).Decode(&versionInfo); err != nil {
+	if err := manager.decode(resp, &versionInfo); err != nil {
 		return manager
 	}
 
@@ -324,6 +340,22 @@ func newDockerManager(a *Agent) *dockerManager {
 	return manager
 }
 
+// Decodes Docker API JSON response using a reusable buffer and decoder. Not thread safe.
+func (dm *dockerManager) decode(resp *http.Response, d any) error {
+	if dm.buf == nil {
+		// initialize buffer with 256kb starting size
+		dm.buf = bytes.NewBuffer(make([]byte, 0, 1024*256))
+		dm.decoder = json.NewDecoder(dm.buf)
+	}
+	defer resp.Body.Close()
+	defer dm.buf.Reset()
+	_, err := dm.buf.ReadFrom(resp.Body)
+	if err != nil {
+		return err
+	}
+	return dm.decoder.Decode(d)
+}
+
 // Test docker / podman sockets and return if one exists
 func getDockerHost() string {
 	scheme := "unix://"

beszel/internal/agent/gpu.go

@@ -18,24 +18,24 @@ import (
 
 const (
 	// Commands
-	nvidiaSmiCmd  = "nvidia-smi"
+	nvidiaSmiCmd  string = "nvidia-smi"
-	rocmSmiCmd    = "rocm-smi"
+	rocmSmiCmd    string = "rocm-smi"
-	tegraStatsCmd = "tegrastats"
+	tegraStatsCmd string = "tegrastats"
 
 	// Polling intervals
-	nvidiaSmiInterval  = "4"    // in seconds
+	nvidiaSmiInterval  string        = "4"    // in seconds
-	tegraStatsInterval = "3700" // in milliseconds
+	tegraStatsInterval string        = "3700" // in milliseconds
-	rocmSmiInterval    = 4300 * time.Millisecond
+	rocmSmiInterval    time.Duration = 4300 * time.Millisecond
 
 	// Command retry and timeout constants
-	retryWaitTime     = 5 * time.Second
+	retryWaitTime     time.Duration = 5 * time.Second
-	maxFailureRetries = 5
+	maxFailureRetries int           = 5
 
-	cmdBufferSize = 10 * 1024
+	cmdBufferSize uint16 = 10 * 1024
 
 	// Unit Conversions
-	mebibytesInAMegabyte = 1.024  // nvidia-smi reports memory in MiB
+	mebibytesInAMegabyte float64 = 1.024  // nvidia-smi reports memory in MiB
-	milliwattsInAWatt    = 1000.0 // tegrastats reports power in mW
+	milliwattsInAWatt    float64 = 1000.0 // tegrastats reports power in mW
 )
 
 // GPUManager manages data collection for GPUs (either Nvidia or AMD)
@@ -243,21 +243,26 @@ func (gm *GPUManager) GetCurrentData() map[string]system.GPUData {
 	// copy / reset the data
 	gpuData := make(map[string]system.GPUData, len(gm.GpuDataMap))
 	for id, gpu := range gm.GpuDataMap {
-		// sum the data
+		gpuAvg := *gpu
-		gpu.Temperature = twoDecimals(gpu.Temperature)
+
-		gpu.MemoryUsed = twoDecimals(gpu.MemoryUsed)
+		gpuAvg.Temperature = twoDecimals(gpu.Temperature)
-		gpu.MemoryTotal = twoDecimals(gpu.MemoryTotal)
+		gpuAvg.MemoryUsed = twoDecimals(gpu.MemoryUsed)
-		gpu.Usage = twoDecimals(gpu.Usage / gpu.Count)
+		gpuAvg.MemoryTotal = twoDecimals(gpu.MemoryTotal)
-		gpu.Power = twoDecimals(gpu.Power / gpu.Count)
+
-		// reset the count
+		// avoid division by zero
-		gpu.Count = 1
+		if gpu.Count > 0 {
-		// dereference to avoid overwriting anything else
+			gpuAvg.Usage = twoDecimals(gpu.Usage / gpu.Count)
-		gpuCopy := *gpu
+			gpuAvg.Power = twoDecimals(gpu.Power / gpu.Count)
+		}
+
+		// reset accumulators in the original
+		gpu.Usage, gpu.Power, gpu.Count = 0, 0, 0
+
 		// append id to the name if there are multiple GPUs with the same name
 		if nameCounts[gpu.Name] > 1 {
-			gpuCopy.Name = fmt.Sprintf("%s %s", gpu.Name, id)
+			gpuAvg.Name = fmt.Sprintf("%s %s", gpu.Name, id)
 		}
-		gpuData[id] = gpuCopy
+		gpuData[id] = gpuAvg
 	}
 	slog.Debug("GPU", "data", gpuData)
 	return gpuData

beszel/internal/agent/gpu_test.go

@@ -279,6 +279,19 @@ func TestParseJetsonData(t *testing.T) {
 				Count:       1,
 			},
 		},
+		{
+			name:  "orin nano",
+			input: "06-18-2025 11:25:24 RAM 3452/7620MB (lfb 25x4MB) SWAP 1518/16384MB (cached 174MB) CPU [1%@1420,2%@1420,0%@1420,2%@1420,2%@729,1%@729] GR3D_FREQ 0% cpu@50.031C soc2@49.031C soc0@50C gpu@49.031C tj@50.25C soc1@50.25C VDD_IN 4824mW/4824mW VDD_CPU_GPU_CV 518mW/518mW VDD_SOC 1475mW/1475mW",
+			wantMetrics: &system.GPUData{
+				Name:        "GPU",
+				MemoryUsed:  3452.0,
+				MemoryTotal: 7620.0,
+				Usage:       0.0,
+				Temperature: 50.25,
+				Power:       0.518,
+				Count:       1,
+			},
+		},
 		{
 			name:  "missing temperature",
 			input: "11-14-2024 22:54:33 RAM 4300/30698MB GR3D_FREQ 45% VDD_GPU_SOC 2171mW",
@@ -318,44 +331,85 @@ func TestParseJetsonData(t *testing.T) {
 }
 
 func TestGetCurrentData(t *testing.T) {
-	gm := &GPUManager{
+	t.Run("calculates averages and resets accumulators", func(t *testing.T) {
-		GpuDataMap: map[string]*system.GPUData{
+		gm := &GPUManager{
-			"0": {
+			GpuDataMap: map[string]*system.GPUData{
-				Name:        "GPU1",
+				"0": {
-				Temperature: 50,
+					Name:        "GPU1",
-				MemoryUsed:  2048,
+					Temperature: 50,
-				MemoryTotal: 4096,
+					MemoryUsed:  2048,
-				Usage:       100, // 100 over 2 counts = 50 avg
+					MemoryTotal: 4096,
-				Power:       200, // 200 over 2 counts = 100 avg
+					Usage:       100, // 100 over 2 counts = 50 avg
-				Count:       2,
+					Power:       200, // 200 over 2 counts = 100 avg
+					Count:       2,
+				},
+				"1": {
+					Name:        "GPU1",
+					Temperature: 60,
+					MemoryUsed:  3072,
+					MemoryTotal: 8192,
+					Usage:       30,
+					Power:       60,
+					Count:       1,
+				},
+				"2": {
+					Name:        "GPU 2",
+					Temperature: 70,
+					MemoryUsed:  4096,
+					MemoryTotal: 8192,
+					Usage:       200,
+					Power:       400,
+					Count:       1,
+				},
 			},
-			"1": {
+		}
-				Name:        "GPU1",
+
-				Temperature: 60,
+		result := gm.GetCurrentData()
-				MemoryUsed:  3072,
+
-				MemoryTotal: 8192,
+		// Verify name disambiguation
-				Usage:       30,
+		assert.Equal(t, "GPU1 0", result["0"].Name)
-				Power:       60,
+		assert.Equal(t, "GPU1 1", result["1"].Name)
-				Count:       1,
+		assert.Equal(t, "GPU 2", result["2"].Name)
+
+		// Check averaged values in the result
+		assert.InDelta(t, 50.0, result["0"].Usage, 0.01)
+		assert.InDelta(t, 100.0, result["0"].Power, 0.01)
+		assert.InDelta(t, 30.0, result["1"].Usage, 0.01)
+		assert.InDelta(t, 60.0, result["1"].Power, 0.01)
+
+		// Verify that accumulators in the original map are reset
+		assert.Equal(t, float64(0), gm.GpuDataMap["0"].Count, "GPU 0 Count should be reset")
+		assert.Equal(t, float64(0), gm.GpuDataMap["0"].Usage, "GPU 0 Usage should be reset")
+		assert.Equal(t, float64(0), gm.GpuDataMap["0"].Power, "GPU 0 Power should be reset")
+		assert.Equal(t, float64(0), gm.GpuDataMap["1"].Count, "GPU 1 Count should be reset")
+		assert.Equal(t, float64(0), gm.GpuDataMap["1"].Usage, "GPU 1 Usage should be reset")
+		assert.Equal(t, float64(0), gm.GpuDataMap["1"].Power, "GPU 1 Power should be reset")
+	})
+
+	t.Run("handles zero count without panicking", func(t *testing.T) {
+		gm := &GPUManager{
+			GpuDataMap: map[string]*system.GPUData{
+				"0": {
+					Name:  "TestGPU",
+					Count: 0,
+					Usage: 0,
+					Power: 0,
+				},
 			},
-		},
+		}
-	}
 
-	result := gm.GetCurrentData()
+		var result map[string]system.GPUData
+		assert.NotPanics(t, func() {
+			result = gm.GetCurrentData()
+		})
 
-	// Verify name disambiguation
+		// Check that usage and power are 0
-	assert.Equal(t, "GPU1 0", result["0"].Name)
+		assert.Equal(t, 0.0, result["0"].Usage)
-	assert.Equal(t, "GPU1 1", result["1"].Name)
+		assert.Equal(t, 0.0, result["0"].Power)
 
-	// Check averaged values
+		// Verify reset count
-	assert.InDelta(t, 50.0, result["0"].Usage, 0.01)
+		assert.Equal(t, float64(0), gm.GpuDataMap["0"].Count)
-	assert.InDelta(t, 100.0, result["0"].Power, 0.01)
+	})
-	assert.InDelta(t, 30.0, result["1"].Usage, 0.01)
-	assert.InDelta(t, 60.0, result["1"].Power, 0.01)
-
-	// Verify reset counts
-	assert.Equal(t, float64(1), gm.GpuDataMap["0"].Count)
-	assert.Equal(t, float64(1), gm.GpuDataMap["1"].Count)
 }
 
 func TestDetectGPUs(t *testing.T) {
@@ -722,6 +776,18 @@ func TestAccumulation(t *testing.T) {
 				assert.InDelta(t, expected.avgUsage, gpu.Usage, 0.01, "Average usage in GetCurrentData should match")
 				assert.InDelta(t, expected.avgPower, gpu.Power, 0.01, "Average power in GetCurrentData should match")
 			}
+
+			// Verify that accumulators in the original map are reset
+			for id := range tt.expectedValues {
+				gpu, exists := gm.GpuDataMap[id]
+				assert.True(t, exists, "GPU with ID %s should still exist after GetCurrentData", id)
+				if !exists {
+					continue
+				}
+				assert.Equal(t, float64(0), gpu.Count, "Count should be reset for GPU ID %s", id)
+				assert.Equal(t, float64(0), gpu.Usage, "Usage should be reset for GPU ID %s", id)
+				assert.Equal(t, float64(0), gpu.Power, "Power should be reset for GPU ID %s", id)
+			}
 		})
 	}
 }

beszel/internal/agent/health.go

@@ -1,18 +0,0 @@
-package agent
-
-import (
-	"net"
-	"time"
-)
-
-// Health checks if the agent's server is running by attempting to connect to it.
-//
-// If an error occurs when attempting to connect to the server, it returns the error.
-func Health(addr string, network string) error {
-	conn, err := net.DialTimeout(network, addr, 4*time.Second)
-	if err != nil {
-		return err
-	}
-	conn.Close()
-	return nil
-}

beszel/internal/agent/health/health.go

@@ -0,0 +1,43 @@
+// Package health provides functions to check and update the health of the agent.
+// It uses a file in the temp directory to store the timestamp of the last connection attempt.
+// If the timestamp is older than 90 seconds, the agent is considered unhealthy.
+// NB: The agent must be started with the Start() method to be considered healthy.
+package health
+
+import (
+	"errors"
+	"log"
+	"os"
+	"path/filepath"
+	"time"
+)
+
+// healthFile is the path to the health file
+var healthFile = filepath.Join(os.TempDir(), "beszel_health")
+
+// Check checks if the agent is connected by checking the modification time of the health file
+func Check() error {
+	fileInfo, err := os.Stat(healthFile)
+	if err != nil {
+		return err
+	}
+	if time.Since(fileInfo.ModTime()) > 91*time.Second {
+		log.Println("over 90 seconds since last connection")
+		return errors.New("unhealthy")
+	}
+	return nil
+}
+
+// Update updates the modification time of the health file
+func Update() error {
+	file, err := os.Create(healthFile)
+	if err != nil {
+		return err
+	}
+	return file.Close()
+}
+
+// CleanUp removes the health file
+func CleanUp() error {
+	return os.Remove(healthFile)
+}

beszel/internal/agent/health/health_test.go

@@ -0,0 +1,67 @@
+//go:build testing
+// +build testing
+
+package health
+
+import (
+	"os"
+	"path/filepath"
+	"testing"
+	"time"
+
+	"testing/synctest"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func TestHealth(t *testing.T) {
+	// Override healthFile to use a temporary directory for this test.
+	originalHealthFile := healthFile
+	tmpDir := t.TempDir()
+	healthFile = filepath.Join(tmpDir, "beszel_health_test")
+	defer func() { healthFile = originalHealthFile }()
+
+	t.Run("check with no health file", func(t *testing.T) {
+		err := Check()
+		require.Error(t, err)
+		assert.True(t, os.IsNotExist(err), "expected a file-not-exist error, but got: %v", err)
+	})
+
+	t.Run("update and check", func(t *testing.T) {
+		err := Update()
+		require.NoError(t, err, "Update() failed")
+
+		err = Check()
+		assert.NoError(t, err, "Check() failed immediately after Update()")
+	})
+
+	// This test uses synctest to simulate time passing.
+	// NOTE: This test requires GOEXPERIMENT=synctest to run.
+	t.Run("check with simulated time", func(t *testing.T) {
+		synctest.Test(t, func(t *testing.T) {
+			// Update the file to set the initial timestamp.
+			require.NoError(t, Update(), "Update() failed inside synctest")
+
+			// Set the mtime to the current fake time to align the file's timestamp with the simulated clock.
+			now := time.Now()
+			require.NoError(t, os.Chtimes(healthFile, now, now), "Chtimes failed")
+
+			// Wait a duration less than the threshold.
+			time.Sleep(89 * time.Second)
+			synctest.Wait()
+
+			// The check should still pass.
+			assert.NoError(t, Check(), "Check() failed after 89s")
+
+			// Wait for the total duration to exceed the threshold.
+			time.Sleep(5 * time.Second)
+			synctest.Wait()
+
+			// The check should now fail as unhealthy.
+			err := Check()
+			require.Error(t, err, "Check() should have failed after 91s")
+			assert.Equal(t, "unhealthy", err.Error(), "Check() returned wrong error")
+		})
+	})
+}

beszel/internal/agent/health_test.go

@@ -1,118 +0,0 @@
-//go:build testing
-// +build testing
-
-package agent_test
-
-import (
-	"fmt"
-	"net"
-	"os"
-	"testing"
-
-	"github.com/stretchr/testify/require"
-
-	"beszel/internal/agent"
-)
-
-// setupTestServer creates a temporary server for testing
-func setupTestServer(t *testing.T) (string, func()) {
-	// Create a temporary socket file for Unix socket testing
-	tempSockFile := os.TempDir() + "/beszel_health_test.sock"
-
-	// Clean up any existing socket file
-	os.Remove(tempSockFile)
-
-	// Create a listener
-	listener, err := net.Listen("unix", tempSockFile)
-	require.NoError(t, err, "Failed to create test listener")
-
-	// Start a simple server in a goroutine
-	go func() {
-		conn, err := listener.Accept()
-		if err != nil {
-			return // Listener closed
-		}
-		defer conn.Close()
-		// Just accept the connection and do nothing
-	}()
-
-	// Return the socket file path and a cleanup function
-	return tempSockFile, func() {
-		listener.Close()
-		os.Remove(tempSockFile)
-	}
-}
-
-// setupTCPTestServer creates a temporary TCP server for testing
-func setupTCPTestServer(t *testing.T) (string, func()) {
-	// Listen on a random available port
-	listener, err := net.Listen("tcp", "127.0.0.1:0")
-	require.NoError(t, err, "Failed to create test listener")
-
-	// Get the port that was assigned
-	addr := listener.Addr().(*net.TCPAddr)
-	port := addr.Port
-
-	// Start a simple server in a goroutine
-	go func() {
-		conn, err := listener.Accept()
-		if err != nil {
-			return // Listener closed
-		}
-		defer conn.Close()
-		// Just accept the connection and do nothing
-	}()
-
-	// Return the address and a cleanup function
-	return fmt.Sprintf("127.0.0.1:%d", port), func() {
-		listener.Close()
-	}
-}
-
-func TestHealth(t *testing.T) {
-	t.Run("server is running (unix socket)", func(t *testing.T) {
-		// Setup a test server
-		sockFile, cleanup := setupTestServer(t)
-		defer cleanup()
-
-		// Run the health check with explicit parameters
-		err := agent.Health(sockFile, "unix")
-		require.NoError(t, err, "Failed to check health")
-	})
-
-	t.Run("server is running (tcp address)", func(t *testing.T) {
-		// Setup a test server
-		addr, cleanup := setupTCPTestServer(t)
-		defer cleanup()
-
-		// Run the health check with explicit parameters
-		err := agent.Health(addr, "tcp")
-		require.NoError(t, err, "Failed to check health")
-	})
-
-	t.Run("server is not running", func(t *testing.T) {
-		// Use an address that's likely not in use
-		addr := "127.0.0.1:65535"
-
-		// Run the health check with explicit parameters
-		err := agent.Health(addr, "tcp")
-		require.Error(t, err, "Health check should return an error when server is not running")
-	})
-
-	t.Run("invalid network", func(t *testing.T) {
-		// Use an invalid network type
-		err := agent.Health("127.0.0.1:8080", "invalid_network")
-		require.Error(t, err, "Health check should return an error with invalid network")
-	})
-
-	t.Run("unix socket not found", func(t *testing.T) {
-		// Use a non-existent unix socket
-		nonExistentSocket := os.TempDir() + "/non_existent_socket.sock"
-
-		// Make sure it really doesn't exist
-		os.Remove(nonExistentSocket)
-
-		err := agent.Health(nonExistentSocket, "unix")
-		require.Error(t, err, "Health check should return an error when socket doesn't exist")
-	})
-}

beszel/internal/agent/lhm/beszel_lhm.cs

@@ -0,0 +1,80 @@
+using System;
+using System.Globalization;
+using LibreHardwareMonitor.Hardware;
+
+class Program
+{
+  static void Main()
+  {
+    var computer = new Computer
+    {
+      IsCpuEnabled = true,
+      IsGpuEnabled = true,
+      IsMemoryEnabled = true,
+      IsMotherboardEnabled = true,
+      IsStorageEnabled = true,
+      // IsPsuEnabled = true,
+      // IsNetworkEnabled = true,
+    };
+    computer.Open();
+
+    var reader = Console.In;
+    var writer = Console.Out;
+
+    string line;
+    while ((line = reader.ReadLine()) != null)
+    {
+      if (line.Trim().Equals("getTemps", StringComparison.OrdinalIgnoreCase))
+      {
+        foreach (var hw in computer.Hardware)
+        {
+          // process main hardware sensors
+          ProcessSensors(hw, writer);
+
+          // process subhardware sensors
+          foreach (var subhardware in hw.SubHardware)
+          {
+            ProcessSensors(subhardware, writer);
+          }
+        }
+        // send empty line to signal end of sensor data
+        writer.WriteLine();
+        writer.Flush();
+      }
+    }
+
+    computer.Close();
+  }
+
+  static void ProcessSensors(IHardware hardware, System.IO.TextWriter writer)
+  {
+    var updated = false;
+    foreach (var sensor in hardware.Sensors)
+    {
+      var validTemp = sensor.SensorType == SensorType.Temperature && sensor.Value.HasValue;
+      if (!validTemp || sensor.Name.Contains("Distance"))
+      {
+        continue;
+      }
+
+      if (!updated)
+      {
+        hardware.Update();
+        updated = true;
+      }
+
+      var name = sensor.Name;
+      // if sensor.Name starts with "Temperature" replace with hardware.Identifier but retain the rest of the name.
+      // usually this is a number like Temperature 3
+      if (sensor.Name.StartsWith("Temperature"))
+      {
+        name = hardware.Identifier.ToString().Replace("/", "_").TrimStart('_') + sensor.Name.Substring(11);
+      }
+
+      // invariant culture assures the value is parsable as a float
+      var value = sensor.Value.Value.ToString("0.##", CultureInfo.InvariantCulture);
+      // write the name and value to the writer
+      writer.WriteLine($"{name}|{value}");
+    }
+  }
+}

beszel/internal/agent/lhm/beszel_lhm.csproj

@@ -0,0 +1,11 @@
+<Project Sdk="Microsoft.NET.Sdk">
+  <PropertyGroup>
+    <OutputType>Exe</OutputType>
+    <TargetFramework>net48</TargetFramework>
+    <Platforms>x64</Platforms>
+  </PropertyGroup>
+
+  <ItemGroup>
+    <PackageReference Include="LibreHardwareMonitorLib" Version="0.9.4" />
+  </ItemGroup>
+</Project>

beszel/internal/agent/network.go

@@ -57,6 +57,7 @@ func (a *Agent) skipNetworkInterface(v psutilNet.IOCountersStat) bool {
 		strings.HasPrefix(v.Name, "docker"),
 		strings.HasPrefix(v.Name, "br-"),
 		strings.HasPrefix(v.Name, "veth"),
+		strings.HasPrefix(v.Name, "bond"),
 		v.BytesRecv == 0,
 		v.BytesSent == 0:
 		return true

beszel/internal/agent/sensors.go

@@ -3,10 +3,13 @@ package agent
 import (
 	"beszel/internal/entities/system"
 	"context"
+	"fmt"
 	"log/slog"
 	"path"
+	"runtime"
 	"strconv"
 	"strings"
+	"unicode/utf8"
 
 	"github.com/shirou/gopsutil/v4/common"
 	"github.com/shirou/gopsutil/v4/sensors"
@@ -30,6 +33,9 @@ func (a *Agent) newSensorConfig() *SensorConfig {
 	return a.newSensorConfigWithEnv(primarySensor, sysSensors, sensorsEnvVal, skipCollection)
 }
 
+// Matches sensors.TemperaturesWithContext to allow for panic recovery (gopsutil/issues/1832)
+type getTempsFn func(ctx context.Context) ([]sensors.TemperatureStat, error)
+
 // newSensorConfigWithEnv creates a SensorConfig with the provided environment variables
 // sensorsSet indicates if the SENSORS environment variable was explicitly set (even to empty string)
 func (a *Agent) newSensorConfigWithEnv(primarySensor, sysSensors, sensorsEnvVal string, skipCollection bool) *SensorConfig {
@@ -78,8 +84,18 @@ func (a *Agent) updateTemperatures(systemStats *system.Stats) {
 	// reset high temp
 	a.systemInfo.DashboardTemp = 0
 
-	// get sensor data
+	temps, err := a.getTempsWithPanicRecovery(getSensorTemps)
-	temps, _ := sensors.TemperaturesWithContext(a.sensorConfig.context)
+	if err != nil {
+		// retry once on panic (gopsutil/issues/1832)
+		temps, err = a.getTempsWithPanicRecovery(getSensorTemps)
+		if err != nil {
+			slog.Warn("Error updating temperatures", "err", err)
+			if len(systemStats.Temperatures) > 0 {
+				systemStats.Temperatures = make(map[string]float64)
+			}
+			return
+		}
+	}
 	slog.Debug("Temperature", "sensors", temps)
 
 	// return if no sensors
@@ -89,6 +105,15 @@ func (a *Agent) updateTemperatures(systemStats *system.Stats) {
 
 	systemStats.Temperatures = make(map[string]float64, len(temps))
 	for i, sensor := range temps {
+		// check for malformed strings on darwin (gopsutil/issues/1832)
+		if runtime.GOOS == "darwin" && !utf8.ValidString(sensor.SensorKey) {
+			continue
+		}
+
+		// scale temperature
+		if sensor.Temperature != 0 && sensor.Temperature < 1 {
+			sensor.Temperature = scaleTemperature(sensor.Temperature)
+		}
 		// skip if temperature is unreasonable
 		if sensor.Temperature <= 0 || sensor.Temperature >= 200 {
 			continue
@@ -103,15 +128,28 @@ func (a *Agent) updateTemperatures(systemStats *system.Stats) {
 			continue
 		}
 		// set dashboard temperature
-		if a.sensorConfig.primarySensor == "" {
+		switch a.sensorConfig.primarySensor {
+		case "":
 			a.systemInfo.DashboardTemp = max(a.systemInfo.DashboardTemp, sensor.Temperature)
-		} else if a.sensorConfig.primarySensor == sensorName {
+		case sensorName:
 			a.systemInfo.DashboardTemp = sensor.Temperature
 		}
 		systemStats.Temperatures[sensorName] = twoDecimals(sensor.Temperature)
 	}
 }
 
+// getTempsWithPanicRecovery wraps sensors.TemperaturesWithContext to recover from panics (gopsutil/issues/1832)
+func (a *Agent) getTempsWithPanicRecovery(getTemps getTempsFn) (temps []sensors.TemperatureStat, err error) {
+	defer func() {
+		if r := recover(); r != nil {
+			err = fmt.Errorf("panic: %v", r)
+		}
+	}()
+	// get sensor data (error ignored intentionally as it may be only with one sensor)
+	temps, _ = getTemps(a.sensorConfig.context)
+	return
+}
+
 // isValidSensor checks if a sensor is valid based on the sensor name and the sensor config
 func isValidSensor(sensorName string, config *SensorConfig) bool {
 	// if no sensors configured, everything is valid
@@ -141,3 +179,19 @@ func isValidSensor(sensorName string, config *SensorConfig) bool {
 
 	return config.isBlacklist
 }
+
+// scaleTemperature scales temperatures in fractional values to reasonable Celsius values
+func scaleTemperature(temp float64) float64 {
+	if temp > 1 {
+		return temp
+	}
+	scaled100 := temp * 100
+	scaled1000 := temp * 1000
+
+	if scaled100 >= 15 && scaled100 <= 95 {
+		return scaled100
+	} else if scaled1000 >= 15 && scaled1000 <= 95 {
+		return scaled1000
+	}
+	return scaled100
+}

beszel/internal/agent/sensors_default.go

@@ -0,0 +1,9 @@
+//go:build !windows
+
+package agent
+
+import (
+	"github.com/shirou/gopsutil/v4/sensors"
+)
+
+var getSensorTemps = sensors.TemperaturesWithContext

beszel/internal/agent/sensors_test.go

@@ -4,11 +4,14 @@
 package agent
 
 import (
+	"beszel/internal/entities/system"
 	"context"
+	"fmt"
 	"os"
 	"testing"
 
 	"github.com/shirou/gopsutil/v4/common"
+	"github.com/shirou/gopsutil/v4/sensors"
 	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"
 )
@@ -372,3 +375,179 @@ func TestNewSensorConfig(t *testing.T) {
 	require.True(t, ok, "EnvMap should contain HostSysEnvKey")
 	assert.Equal(t, "/test/path", sysPath)
 }
+
+func TestScaleTemperature(t *testing.T) {
+	tests := []struct {
+		name     string
+		input    float64
+		expected float64
+		desc     string
+	}{
+		// Normal temperatures (no scaling needed)
+		{"normal_cpu_temp", 45.0, 45.0, "Normal CPU temperature"},
+		{"normal_room_temp", 25.0, 25.0, "Normal room temperature"},
+		{"high_cpu_temp", 85.0, 85.0, "High CPU temperature"},
+		// Zero temperature
+		{"zero_temp", 0.0, 0.0, "Zero temperature"},
+		// Fractional values that should use 100x scaling
+		{"fractional_45c", 0.45, 45.0, "0.45 should become 45°C (100x)"},
+		{"fractional_25c", 0.25, 25.0, "0.25 should become 25°C (100x)"},
+		{"fractional_60c", 0.60, 60.0, "0.60 should become 60°C (100x)"},
+		{"fractional_75c", 0.75, 75.0, "0.75 should become 75°C (100x)"},
+		{"fractional_30c", 0.30, 30.0, "0.30 should become 30°C (100x)"},
+		// Fractional values that should use 1000x scaling
+		{"millifractional_45c", 0.045, 45.0, "0.045 should become 45°C (1000x)"},
+		{"millifractional_25c", 0.025, 25.0, "0.025 should become 25°C (1000x)"},
+		{"millifractional_60c", 0.060, 60.0, "0.060 should become 60°C (1000x)"},
+		{"millifractional_75c", 0.075, 75.0, "0.075 should become 75°C (1000x)"},
+		{"millifractional_35c", 0.035, 35.0, "0.035 should become 35°C (1000x)"},
+		// Edge cases - values outside reasonable range
+		{"very_low_fractional", 0.01, 1.0, "0.01 should default to 100x scaling (1°C)"},
+		{"very_high_fractional", 0.99, 99.0, "0.99 should default to 100x scaling (99°C)"},
+		{"extremely_low", 0.001, 0.1, "0.001 should default to 100x scaling (0.1°C)"},
+		// Boundary cases around the reasonable range (15-95°C)
+		{"boundary_low_100x", 0.15, 15.0, "0.15 should use 100x scaling (15°C)"},
+		{"boundary_high_100x", 0.95, 95.0, "0.95 should use 100x scaling (95°C)"},
+		{"boundary_low_1000x", 0.015, 15.0, "0.015 should use 1000x scaling (15°C)"},
+		{"boundary_high_1000x", 0.095, 95.0, "0.095 should use 1000x scaling (95°C)"},
+		// Values just outside reasonable range
+		{"just_below_range_100x", 0.14, 14.0, "0.14 should default to 100x (14°C)"},
+		{"just_above_range_100x", 0.96, 96.0, "0.96 should default to 100x (96°C)"},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			result := scaleTemperature(tt.input)
+			assert.InDelta(t, tt.expected, result, 0.001,
+				"scaleTemperature(%v) = %v, expected %v (%s)",
+				tt.input, result, tt.expected, tt.desc)
+		})
+	}
+}
+
+func TestScaleTemperatureLogic(t *testing.T) {
+	// Test the logic flow for ambiguous cases
+	t.Run("prefers_100x_when_both_valid", func(t *testing.T) {
+		// 0.5 could be 50°C (100x) or 500°C (1000x)
+		// Should prefer 100x since it's tried first and is in range
+		result := scaleTemperature(0.5)
+		expected := 50.0
+		assert.InDelta(t, expected, result, 0.001,
+			"scaleTemperature(0.5) = %v, expected %v (should prefer 100x scaling)",
+			result, expected)
+	})
+
+	t.Run("uses_1000x_when_100x_too_low", func(t *testing.T) {
+		// 0.05 -> 5°C (100x, too low) or 50°C (1000x, in range)
+		// Should use 1000x since 100x is below reasonable range
+		result := scaleTemperature(0.05)
+		expected := 50.0
+		assert.InDelta(t, expected, result, 0.001,
+			"scaleTemperature(0.05) = %v, expected %v (should use 1000x scaling)",
+			result, expected)
+	})
+
+	t.Run("defaults_to_100x_when_both_invalid", func(t *testing.T) {
+		// 0.005 -> 0.5°C (100x, too low) or 5°C (1000x, too low)
+		// Should default to 100x scaling
+		result := scaleTemperature(0.005)
+		expected := 0.5
+		assert.InDelta(t, expected, result, 0.001,
+			"scaleTemperature(0.005) = %v, expected %v (should default to 100x)",
+			result, expected)
+	})
+}
+
+func TestGetTempsWithPanicRecovery(t *testing.T) {
+	agent := &Agent{
+		systemInfo: system.Info{},
+		sensorConfig: &SensorConfig{
+			context: context.Background(),
+		},
+	}
+
+	tests := []struct {
+		name        string
+		getTempsFn  getTempsFn
+		expectError bool
+		errorMsg    string
+	}{
+		{
+			name: "successful_function_call",
+			getTempsFn: func(ctx context.Context) ([]sensors.TemperatureStat, error) {
+				return []sensors.TemperatureStat{
+					{SensorKey: "test_sensor", Temperature: 45.0},
+				}, nil
+			},
+			expectError: false,
+		},
+		{
+			name: "function_returns_error",
+			getTempsFn: func(ctx context.Context) ([]sensors.TemperatureStat, error) {
+				return []sensors.TemperatureStat{
+					{SensorKey: "test_sensor", Temperature: 45.0},
+				}, fmt.Errorf("sensor error")
+			},
+			expectError: false, // getTempsWithPanicRecovery ignores errors from the function
+		},
+		{
+			name: "function_panics_with_string",
+			getTempsFn: func(ctx context.Context) ([]sensors.TemperatureStat, error) {
+				panic("test panic")
+			},
+			expectError: true,
+			errorMsg:    "panic: test panic",
+		},
+		{
+			name: "function_panics_with_error",
+			getTempsFn: func(ctx context.Context) ([]sensors.TemperatureStat, error) {
+				panic(fmt.Errorf("panic error"))
+			},
+			expectError: true,
+			errorMsg:    "panic:",
+		},
+		{
+			name: "function_panics_with_index_out_of_bounds",
+			getTempsFn: func(ctx context.Context) ([]sensors.TemperatureStat, error) {
+				slice := []int{1, 2, 3}
+				_ = slice[10] // out of bounds panic
+				return nil, nil
+			},
+			expectError: true,
+			errorMsg:    "panic:",
+		},
+		{
+			name: "function_panics_with_any_conversion",
+			getTempsFn: func(ctx context.Context) ([]sensors.TemperatureStat, error) {
+				var i any = "string"
+				_ = i.(int) // type assertion panic
+				return nil, nil
+			},
+			expectError: true,
+			errorMsg:    "panic:",
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			var temps []sensors.TemperatureStat
+			var err error
+
+			// The function should not panic, regardless of what the injected function does
+			assert.NotPanics(t, func() {
+				temps, err = agent.getTempsWithPanicRecovery(tt.getTempsFn)
+			}, "getTempsWithPanicRecovery should not panic")
+
+			if tt.expectError {
+				assert.Error(t, err, "Expected an error to be returned")
+				if tt.errorMsg != "" {
+					assert.Contains(t, err.Error(), tt.errorMsg,
+						"Error message should contain expected text")
+				}
+				assert.Nil(t, temps, "Temps should be nil when panic occurs")
+			} else {
+				assert.NoError(t, err, "Should not return error for successful calls")
+			}
+		})
+	}
+}

beszel/internal/agent/sensors_windows.go

@@ -0,0 +1,281 @@
+//go:build windows
+
+//go:generate dotnet build -c Release lhm/beszel_lhm.csproj
+
+package agent
+
+import (
+	"bufio"
+	"context"
+	"embed"
+	"errors"
+	"fmt"
+	"io"
+	"log/slog"
+	"os"
+	"os/exec"
+	"path"
+	"path/filepath"
+	"strconv"
+	"strings"
+	"sync"
+	"time"
+
+	"github.com/shirou/gopsutil/v4/sensors"
+)
+
+// Note: This is always called from Agent.gatherStats() which holds Agent.Lock(),
+// so no internal concurrency protection is needed.
+
+// lhmProcess is a wrapper around the LHM .NET process.
+type lhmProcess struct {
+	cmd                  *exec.Cmd
+	stdin                io.WriteCloser
+	stdout               io.ReadCloser
+	scanner              *bufio.Scanner
+	isRunning            bool
+	stoppedNoSensors     bool
+	consecutiveNoSensors uint8
+	execPath             string
+	tempDir              string
+}
+
+//go:embed all:lhm/bin/Release/net48
+var lhmFs embed.FS
+
+var (
+	beszelLhm     *lhmProcess
+	beszelLhmOnce sync.Once
+)
+
+var errNoSensors = errors.New("no sensors found (try running as admin)")
+
+// newlhmProcess copies the embedded LHM executable to a temporary directory and starts it.
+func newlhmProcess() (*lhmProcess, error) {
+	destDir := filepath.Join(os.TempDir(), "beszel")
+	execPath := filepath.Join(destDir, "beszel_lhm.exe")
+
+	if err := os.MkdirAll(destDir, 0755); err != nil {
+		return nil, fmt.Errorf("failed to create temp directory: %w", err)
+	}
+
+	// Only copy if executable doesn't exist
+	if _, err := os.Stat(execPath); os.IsNotExist(err) {
+		if err := copyEmbeddedDir(lhmFs, "lhm/bin/Release/net48", destDir); err != nil {
+			return nil, fmt.Errorf("failed to copy embedded directory: %w", err)
+		}
+	}
+
+	lhm := &lhmProcess{
+		execPath: execPath,
+		tempDir:  destDir,
+	}
+
+	if err := lhm.startProcess(); err != nil {
+		return nil, fmt.Errorf("failed to start process: %w", err)
+	}
+
+	return lhm, nil
+}
+
+// startProcess starts the external LHM process
+func (lhm *lhmProcess) startProcess() error {
+	// Clean up any existing process
+	lhm.cleanupProcess()
+
+	cmd := exec.Command(lhm.execPath)
+	stdin, err := cmd.StdinPipe()
+	if err != nil {
+		return err
+	}
+
+	stdout, err := cmd.StdoutPipe()
+	if err != nil {
+		stdin.Close()
+		return err
+	}
+
+	if err := cmd.Start(); err != nil {
+		stdin.Close()
+		stdout.Close()
+		return err
+	}
+
+	// Update process state
+	lhm.cmd = cmd
+	lhm.stdin = stdin
+	lhm.stdout = stdout
+	lhm.scanner = bufio.NewScanner(stdout)
+	lhm.isRunning = true
+
+	// Give process a moment to initialize
+	time.Sleep(100 * time.Millisecond)
+
+	return nil
+}
+
+// cleanupProcess terminates the process and closes resources but preserves files
+func (lhm *lhmProcess) cleanupProcess() {
+	lhm.isRunning = false
+
+	if lhm.cmd != nil && lhm.cmd.Process != nil {
+		lhm.cmd.Process.Kill()
+		lhm.cmd.Wait()
+	}
+
+	if lhm.stdin != nil {
+		lhm.stdin.Close()
+		lhm.stdin = nil
+	}
+	if lhm.stdout != nil {
+		lhm.stdout.Close()
+		lhm.stdout = nil
+	}
+
+	lhm.cmd = nil
+	lhm.scanner = nil
+	lhm.stoppedNoSensors = false
+	lhm.consecutiveNoSensors = 0
+}
+
+func (lhm *lhmProcess) getTemps(ctx context.Context) (temps []sensors.TemperatureStat, err error) {
+	if lhm.stoppedNoSensors {
+		// Fall back to gopsutil if we can't get sensors from LHM
+		return sensors.TemperaturesWithContext(ctx)
+	}
+
+	// Start process if it's not running
+	if !lhm.isRunning || lhm.stdin == nil || lhm.scanner == nil {
+		err := lhm.startProcess()
+		if err != nil {
+			return temps, err
+		}
+	}
+
+	// Send command to process
+	_, err = fmt.Fprintln(lhm.stdin, "getTemps")
+	if err != nil {
+		lhm.isRunning = false
+		return temps, fmt.Errorf("failed to send command: %w", err)
+	}
+
+	// Read all sensor lines until we hit an empty line or EOF
+	for lhm.scanner.Scan() {
+		line := strings.TrimSpace(lhm.scanner.Text())
+		if line == "" {
+			break
+		}
+
+		parts := strings.Split(line, "|")
+		if len(parts) != 2 {
+			slog.Debug("Invalid sensor format", "line", line)
+			continue
+		}
+
+		name := strings.TrimSpace(parts[0])
+		valueStr := strings.TrimSpace(parts[1])
+
+		value, err := strconv.ParseFloat(valueStr, 64)
+		if err != nil {
+			slog.Debug("Failed to parse sensor", "err", err, "line", line)
+			continue
+		}
+
+		if name == "" || value <= 0 || value > 150 {
+			slog.Debug("Invalid sensor", "name", name, "val", value, "line", line)
+			continue
+		}
+
+		temps = append(temps, sensors.TemperatureStat{
+			SensorKey:   name,
+			Temperature: value,
+		})
+	}
+
+	if err := lhm.scanner.Err(); err != nil {
+		lhm.isRunning = false
+		return temps, err
+	}
+
+	// Handle no sensors case
+	if len(temps) == 0 {
+		lhm.consecutiveNoSensors++
+		if lhm.consecutiveNoSensors >= 3 {
+			lhm.stoppedNoSensors = true
+			slog.Warn(errNoSensors.Error())
+			lhm.cleanup()
+		}
+		return sensors.TemperaturesWithContext(ctx)
+	}
+
+	lhm.consecutiveNoSensors = 0
+
+	return temps, nil
+}
+
+// getSensorTemps attempts to pull sensor temperatures from the embedded LHM process.
+// NB: LibreHardwareMonitorLib requires admin privileges to access all available sensors.
+func getSensorTemps(ctx context.Context) (temps []sensors.TemperatureStat, err error) {
+	defer func() {
+		if err != nil {
+			slog.Debug("Error reading sensors", "err", err)
+		}
+	}()
+
+	// Initialize process once
+	beszelLhmOnce.Do(func() {
+		beszelLhm, err = newlhmProcess()
+	})
+
+	if err != nil {
+		return temps, fmt.Errorf("failed to initialize lhm: %w", err)
+	}
+
+	if beszelLhm == nil {
+		return temps, fmt.Errorf("lhm not available")
+	}
+
+	return beszelLhm.getTemps(ctx)
+}
+
+// cleanup terminates the process and closes resources
+func (lhm *lhmProcess) cleanup() {
+	lhm.cleanupProcess()
+	if lhm.tempDir != "" {
+		os.RemoveAll(lhm.tempDir)
+	}
+}
+
+// copyEmbeddedDir copies the embedded directory to the destination path
+func copyEmbeddedDir(fs embed.FS, srcPath, destPath string) error {
+	entries, err := fs.ReadDir(srcPath)
+	if err != nil {
+		return err
+	}
+	if err := os.MkdirAll(destPath, 0755); err != nil {
+		return err
+	}
+
+	for _, entry := range entries {
+		srcEntryPath := path.Join(srcPath, entry.Name())
+		destEntryPath := filepath.Join(destPath, entry.Name())
+
+		if entry.IsDir() {
+			if err := copyEmbeddedDir(fs, srcEntryPath, destEntryPath); err != nil {
+				return err
+			}
+			continue
+		}
+
+		data, err := fs.ReadFile(srcEntryPath)
+		if err != nil {
+			return err
+		}
+
+		if err := os.WriteFile(destEntryPath, data, 0755); err != nil {
+			return err
+		}
+	}
+
+	return nil
+}

beszel/internal/agent/server.go

@@ -1,25 +1,43 @@
 package agent
 
 import (
+	"beszel"
+	"beszel/internal/common"
+	"beszel/internal/entities/system"
 	"encoding/json"
+	"errors"
 	"fmt"
+	"io"
 	"log/slog"
 	"net"
 	"os"
 	"strings"
+	"time"
 
-	sshServer "github.com/gliderlabs/ssh"
+	"github.com/blang/semver"
-	"golang.org/x/crypto/ssh"
+	"github.com/fxamacker/cbor/v2"
+	"github.com/gliderlabs/ssh"
+	gossh "golang.org/x/crypto/ssh"
 )
 
+// ServerOptions contains configuration options for starting the SSH server.
 type ServerOptions struct {
-	Addr    string
+	Addr    string            // Network address to listen on (e.g., ":45876" or "/path/to/socket")
-	Network string
+	Network string            // Network type ("tcp" or "unix")
-	Keys    []ssh.PublicKey
+	Keys    []gossh.PublicKey // SSH public keys for authentication
 }
 
+// hubVersions caches hub versions by session ID to avoid repeated parsing.
+var hubVersions map[string]semver.Version
+
+// StartServer starts the SSH server with the provided options.
+// It configures the server with secure defaults, sets up authentication,
+// and begins listening for connections. Returns an error if the server
+// is already running or if there's an issue starting the server.
 func (a *Agent) StartServer(opts ServerOptions) error {
-	sshServer.Handle(a.handleSession)
+	if a.server != nil {
+		return errors.New("server already started")
+	}
 
 	slog.Info("Starting SSH server", "addr", opts.Addr, "network", opts.Network)
 
@@ -37,33 +55,109 @@ func (a *Agent) StartServer(opts ServerOptions) error {
 	}
 	defer ln.Close()
 
-	// Start SSH server on the listener
+	// base config (limit to allowed algorithms)
-	return sshServer.Serve(ln, nil, sshServer.NoPty(),
+	config := &gossh.ServerConfig{
-		sshServer.PublicKeyAuth(func(ctx sshServer.Context, key sshServer.PublicKey) bool {
+		ServerVersion: fmt.Sprintf("SSH-2.0-%s_%s", beszel.AppName, beszel.Version),
+	}
+	config.KeyExchanges = common.DefaultKeyExchanges
+	config.MACs = common.DefaultMACs
+	config.Ciphers = common.DefaultCiphers
+
+	// set default handler
+	ssh.Handle(a.handleSession)
+
+	a.server = &ssh.Server{
+		ServerConfigCallback: func(ctx ssh.Context) *gossh.ServerConfig {
+			return config
+		},
+		// check public key(s)
+		PublicKeyHandler: func(ctx ssh.Context, key ssh.PublicKey) bool {
+			remoteAddr := ctx.RemoteAddr()
 			for _, pubKey := range opts.Keys {
-				if sshServer.KeysEqual(key, pubKey) {
+				if ssh.KeysEqual(key, pubKey) {
+					slog.Info("SSH connected", "addr", remoteAddr)
 					return true
 				}
 			}
+			slog.Warn("Invalid SSH key", "addr", remoteAddr)
 			return false
-		}),
+		},
-	)
+		// disable pty
+		PtyCallback: func(ctx ssh.Context, pty ssh.Pty) bool {
+			return false
+		},
+		// close idle connections after 70 seconds
+		IdleTimeout: 70 * time.Second,
+	}
+
+	// Start SSH server on the listener
+	return a.server.Serve(ln)
 }
 
-func (a *Agent) handleSession(s sshServer.Session) {
+// getHubVersion retrieves and caches the hub version for a given session.
-	slog.Debug("New session", "client", s.RemoteAddr())
+// It extracts the version from the SSH client version string and caches
-	stats := a.gatherStats(s.Context().SessionID())
+// it to avoid repeated parsing. Returns a zero version if parsing fails.
-	if err := json.NewEncoder(s).Encode(stats); err != nil {
+func (a *Agent) getHubVersion(sessionId string, sessionCtx ssh.Context) semver.Version {
+	if hubVersions == nil {
+		hubVersions = make(map[string]semver.Version, 1)
+	}
+	hubVersion, ok := hubVersions[sessionId]
+	if ok {
+		return hubVersion
+	}
+	// Extract hub version from SSH client version
+	clientVersion := sessionCtx.Value(ssh.ContextKeyClientVersion)
+	if versionStr, ok := clientVersion.(string); ok {
+		hubVersion, _ = extractHubVersion(versionStr)
+	}
+	hubVersions[sessionId] = hubVersion
+	return hubVersion
+}
+
+// handleSession handles an incoming SSH session by gathering system statistics
+// and sending them to the hub. It signals connection events, determines the
+// appropriate encoding format based on hub version, and exits with appropriate
+// status codes.
+func (a *Agent) handleSession(s ssh.Session) {
+	a.connectionManager.eventChan <- SSHConnect
+
+	sessionCtx := s.Context()
+	sessionID := sessionCtx.SessionID()
+
+	hubVersion := a.getHubVersion(sessionID, sessionCtx)
+
+	stats := a.gatherStats(sessionID)
+
+	err := a.writeToSession(s, stats, hubVersion)
+	if err != nil {
 		slog.Error("Error encoding stats", "err", err, "stats", stats)
 		s.Exit(1)
+	} else {
+		s.Exit(0)
 	}
-	s.Exit(0)
+}
+
+// writeToSession encodes and writes system statistics to the session.
+// It chooses between CBOR and JSON encoding based on the hub version,
+// using CBOR for newer versions and JSON for legacy compatibility.
+func (a *Agent) writeToSession(w io.Writer, stats *system.CombinedData, hubVersion semver.Version) error {
+	if hubVersion.GTE(beszel.MinVersionCbor) {
+		return cbor.NewEncoder(w).Encode(stats)
+	}
+	return json.NewEncoder(w).Encode(stats)
+}
+
+// extractHubVersion extracts the beszel version from SSH client version string.
+// Expected format: "SSH-2.0-beszel_X.Y.Z" or "beszel_X.Y.Z"
+func extractHubVersion(versionString string) (semver.Version, error) {
+	_, after, _ := strings.Cut(versionString, "_")
+	return semver.Parse(after)
 }
 
 // ParseKeys parses a string containing SSH public keys in authorized_keys format.
 // It returns a slice of ssh.PublicKey and an error if any key fails to parse.
-func ParseKeys(input string) ([]ssh.PublicKey, error) {
+func ParseKeys(input string) ([]gossh.PublicKey, error) {
-	var parsedKeys []ssh.PublicKey
+	var parsedKeys []gossh.PublicKey
 	for line := range strings.Lines(input) {
 		line = strings.TrimSpace(line)
 		// Skip empty lines or comments
@@ -71,7 +165,7 @@ func ParseKeys(input string) ([]ssh.PublicKey, error) {
 			continue
 		}
 		// Parse the key
-		parsedKey, _, _, _, err := ssh.ParseAuthorizedKey([]byte(line))
+		parsedKey, _, _, _, err := gossh.ParseAuthorizedKey([]byte(line))
 		if err != nil {
 			return nil, fmt.Errorf("failed to parse key: %s, error: %w", line, err)
 		}
@@ -80,7 +174,9 @@ func ParseKeys(input string) ([]ssh.PublicKey, error) {
 	return parsedKeys, nil
 }
 
-// GetAddress gets the address to listen on or connect to from environment variables or default value.
+// GetAddress determines the network address to listen on from various sources.
+// It checks the provided address, then environment variables (LISTEN, PORT),
+// and finally defaults to ":45876".
 func GetAddress(addr string) string {
 	if addr == "" {
 		addr, _ = GetEnv("LISTEN")
@@ -99,7 +195,9 @@ func GetAddress(addr string) string {
 	return addr
 }
 
-// GetNetwork returns the network type to use based on the address
+// GetNetwork determines the network type based on the address format.
+// It checks the NETWORK environment variable first, then infers from
+// the address format: addresses starting with "/" are "unix", others are "tcp".
 func GetNetwork(addr string) string {
 	if network, ok := GetEnv("NETWORK"); ok && network != "" {
 		return network
@@ -109,3 +207,17 @@ func GetNetwork(addr string) string {
 	}
 	return "tcp"
 }
+
+// StopServer stops the SSH server if it's running.
+// It returns an error if the server is not running or if there's an error stopping it.
+func (a *Agent) StopServer() error {
+	if a.server == nil {
+		return errors.New("SSH server not running")
+	}
+
+	slog.Info("Stopping SSH server")
+	_ = a.server.Close()
+	a.server = nil
+	a.connectionManager.eventChan <- SSHDisconnect
+	return nil
+}

beszel/internal/agent/server_test.go

@@ -1,34 +1,43 @@
 package agent
 
 import (
+	"beszel/internal/entities/container"
+	"beszel/internal/entities/system"
+	"context"
 	"crypto/ed25519"
+	"encoding/json"
 	"fmt"
+	"net"
 	"os"
 	"path/filepath"
 	"strings"
+	"sync"
 	"testing"
 	"time"
 
+	"github.com/blang/semver"
+	"github.com/fxamacker/cbor/v2"
+	"github.com/gliderlabs/ssh"
 	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"
-	"golang.org/x/crypto/ssh"
+	gossh "golang.org/x/crypto/ssh"
 )
 
 func TestStartServer(t *testing.T) {
 	// Generate a test key pair
 	pubKey, privKey, err := ed25519.GenerateKey(nil)
 	require.NoError(t, err)
-	signer, err := ssh.NewSignerFromKey(privKey)
+	signer, err := gossh.NewSignerFromKey(privKey)
 	require.NoError(t, err)
-	sshPubKey, err := ssh.NewPublicKey(pubKey)
+	sshPubKey, err := gossh.NewPublicKey(pubKey)
 	require.NoError(t, err)
 
 	// Generate a different key pair for bad key test
 	badPubKey, badPrivKey, err := ed25519.GenerateKey(nil)
 	require.NoError(t, err)
-	badSigner, err := ssh.NewSignerFromKey(badPrivKey)
+	badSigner, err := gossh.NewSignerFromKey(badPrivKey)
 	require.NoError(t, err)
-	sshBadPubKey, err := ssh.NewPublicKey(badPubKey)
+	sshBadPubKey, err := gossh.NewPublicKey(badPubKey)
 	require.NoError(t, err)
 
 	socketFile := filepath.Join(t.TempDir(), "beszel-test.sock")
@@ -46,7 +55,7 @@ func TestStartServer(t *testing.T) {
 			config: ServerOptions{
 				Network: "tcp",
 				Addr:    ":45987",
-				Keys:    []ssh.PublicKey{sshPubKey},
+				Keys:    []gossh.PublicKey{sshPubKey},
 			},
 		},
 		{
@@ -54,7 +63,7 @@ func TestStartServer(t *testing.T) {
 			config: ServerOptions{
 				Network: "tcp4",
 				Addr:    "127.0.0.1:45988",
-				Keys:    []ssh.PublicKey{sshPubKey},
+				Keys:    []gossh.PublicKey{sshPubKey},
 			},
 		},
 		{
@@ -62,7 +71,7 @@ func TestStartServer(t *testing.T) {
 			config: ServerOptions{
 				Network: "tcp6",
 				Addr:    "[::1]:45989",
-				Keys:    []ssh.PublicKey{sshPubKey},
+				Keys:    []gossh.PublicKey{sshPubKey},
 			},
 		},
 		{
@@ -70,7 +79,7 @@ func TestStartServer(t *testing.T) {
 			config: ServerOptions{
 				Network: "unix",
 				Addr:    socketFile,
-				Keys:    []ssh.PublicKey{sshPubKey},
+				Keys:    []gossh.PublicKey{sshPubKey},
 			},
 			setup: func() error {
 				// Create a socket file that should be removed
@@ -89,7 +98,7 @@ func TestStartServer(t *testing.T) {
 			config: ServerOptions{
 				Network: "tcp",
 				Addr:    ":45987",
-				Keys:    []ssh.PublicKey{sshBadPubKey},
+				Keys:    []gossh.PublicKey{sshBadPubKey},
 			},
 			wantErr:     true,
 			errContains: "ssh: handshake failed",
@@ -99,7 +108,7 @@ func TestStartServer(t *testing.T) {
 			config: ServerOptions{
 				Network: "tcp",
 				Addr:    ":45987",
-				Keys:    []ssh.PublicKey{sshPubKey},
+				Keys:    []gossh.PublicKey{sshPubKey},
 			},
 		},
 	}
@@ -115,7 +124,8 @@ func TestStartServer(t *testing.T) {
 				defer tt.cleanup()
 			}
 
-			agent := NewAgent()
+			agent, err := NewAgent("")
+			require.NoError(t, err)
 
 			// Start server in a goroutine since it blocks
 			errChan := make(chan error, 1)
@@ -127,8 +137,7 @@ func TestStartServer(t *testing.T) {
 			time.Sleep(100 * time.Millisecond)
 
 			// Try to connect to verify server is running
-			var client *ssh.Client
+			var client *gossh.Client
-			var err error
 
 			// Choose the appropriate signer based on the test case
 			testSigner := signer
@@ -136,23 +145,23 @@ func TestStartServer(t *testing.T) {
 				testSigner = badSigner
 			}
 
-			sshClientConfig := &ssh.ClientConfig{
+			sshClientConfig := &gossh.ClientConfig{
 				User: "a",
-				Auth: []ssh.AuthMethod{
+				Auth: []gossh.AuthMethod{
-					ssh.PublicKeys(testSigner),
+					gossh.PublicKeys(testSigner),
 				},
-				HostKeyCallback: ssh.InsecureIgnoreHostKey(),
+				HostKeyCallback: gossh.InsecureIgnoreHostKey(),
 				Timeout:         4 * time.Second,
 			}
 
 			switch tt.config.Network {
 			case "unix":
-				client, err = ssh.Dial("unix", tt.config.Addr, sshClientConfig)
+				client, err = gossh.Dial("unix", tt.config.Addr, sshClientConfig)
 			default:
 				if !strings.Contains(tt.config.Addr, ":") {
 					tt.config.Addr = ":" + tt.config.Addr
 				}
-				client, err = ssh.Dial("tcp", tt.config.Addr, sshClientConfig)
+				client, err = gossh.Dial("tcp", tt.config.Addr, sshClientConfig)
 			}
 
 			if tt.wantErr {
@@ -287,3 +296,310 @@ func TestParseInvalidKey(t *testing.T) {
 		t.Fatalf("Expected error message to contain '%s', got: %v", expectedErrMsg, err)
 	}
 }
+
+/////////////////////////////////////////////////////////////////
+//////////////////// Hub Version Tests //////////////////////////
+/////////////////////////////////////////////////////////////////
+
+func TestExtractHubVersion(t *testing.T) {
+	tests := []struct {
+		name            string
+		clientVersion   string
+		expectedVersion string
+		expectError     bool
+	}{
+		{
+			name:            "valid beszel client version with underscore",
+			clientVersion:   "SSH-2.0-beszel_0.11.1",
+			expectedVersion: "0.11.1",
+			expectError:     false,
+		},
+		{
+			name:            "valid beszel client version with beta",
+			clientVersion:   "SSH-2.0-beszel_1.0.0-beta",
+			expectedVersion: "1.0.0-beta",
+			expectError:     false,
+		},
+		{
+			name:            "valid beszel client version with rc",
+			clientVersion:   "SSH-2.0-beszel_0.12.0-rc1",
+			expectedVersion: "0.12.0-rc1",
+			expectError:     false,
+		},
+		{
+			name:            "different SSH client",
+			clientVersion:   "SSH-2.0-OpenSSH_8.0",
+			expectedVersion: "8.0",
+			expectError:     true,
+		},
+		{
+			name:          "malformed version string without underscore",
+			clientVersion: "SSH-2.0-beszel",
+			expectError:   true,
+		},
+		{
+			name:          "empty version string",
+			clientVersion: "",
+			expectError:   true,
+		},
+		{
+			name:            "version string with underscore but no version",
+			clientVersion:   "beszel_",
+			expectedVersion: "",
+			expectError:     true,
+		},
+		{
+			name:            "version with patch and build metadata",
+			clientVersion:   "SSH-2.0-beszel_1.2.3+build.123",
+			expectedVersion: "1.2.3+build.123",
+			expectError:     false,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			result, err := extractHubVersion(tt.clientVersion)
+
+			if tt.expectError {
+				assert.Error(t, err)
+				return
+			}
+
+			require.NoError(t, err)
+			assert.Equal(t, tt.expectedVersion, result.String())
+		})
+	}
+}
+
+/////////////////////////////////////////////////////////////////
+/////////////// Hub Version Detection Tests ////////////////////
+/////////////////////////////////////////////////////////////////
+
+func TestGetHubVersion(t *testing.T) {
+	agent, err := NewAgent("")
+	require.NoError(t, err)
+
+	// Mock SSH context that implements the ssh.Context interface
+	mockCtx := &mockSSHContext{
+		sessionID:     "test-session-123",
+		clientVersion: "SSH-2.0-beszel_0.12.0",
+	}
+
+	// Test first call - should extract and cache version
+	version := agent.getHubVersion("test-session-123", mockCtx)
+	assert.Equal(t, "0.12.0", version.String())
+
+	// Test second call - should return cached version
+	mockCtx.clientVersion = "SSH-2.0-beszel_0.11.0" // Change version but should still return cached
+	version = agent.getHubVersion("test-session-123", mockCtx)
+	assert.Equal(t, "0.12.0", version.String()) // Should still be cached version
+
+	// Test different session - should extract new version
+	version = agent.getHubVersion("different-session", mockCtx)
+	assert.Equal(t, "0.11.0", version.String())
+
+	// Test with invalid version string (non-beszel client)
+	mockCtx.clientVersion = "SSH-2.0-OpenSSH_8.0"
+	version = agent.getHubVersion("invalid-session", mockCtx)
+	assert.Equal(t, "0.0.0", version.String()) // Should be empty version for non-beszel clients
+
+	// Test with no client version
+	mockCtx.clientVersion = ""
+	version = agent.getHubVersion("no-version-session", mockCtx)
+	assert.True(t, version.EQ(semver.Version{})) // Should be empty version
+}
+
+// mockSSHContext implements ssh.Context for testing
+type mockSSHContext struct {
+	context.Context
+	sync.Mutex
+	sessionID     string
+	clientVersion string
+}
+
+func (m *mockSSHContext) SessionID() string {
+	return m.sessionID
+}
+
+func (m *mockSSHContext) ClientVersion() string {
+	return m.clientVersion
+}
+
+func (m *mockSSHContext) ServerVersion() string {
+	return "SSH-2.0-beszel_test"
+}
+
+func (m *mockSSHContext) Value(key interface{}) interface{} {
+	if key == ssh.ContextKeyClientVersion {
+		return m.clientVersion
+	}
+	return nil
+}
+
+func (m *mockSSHContext) User() string                    { return "test-user" }
+func (m *mockSSHContext) RemoteAddr() net.Addr            { return nil }
+func (m *mockSSHContext) LocalAddr() net.Addr             { return nil }
+func (m *mockSSHContext) Permissions() *ssh.Permissions   { return nil }
+func (m *mockSSHContext) SetValue(key, value interface{}) {}
+
+/////////////////////////////////////////////////////////////////
+/////////////// CBOR vs JSON Encoding Tests ////////////////////
+/////////////////////////////////////////////////////////////////
+
+// TestWriteToSessionEncoding tests that writeToSession actually encodes data in the correct format
+func TestWriteToSessionEncoding(t *testing.T) {
+	tests := []struct {
+		name             string
+		hubVersion       string
+		expectedUsesCbor bool
+	}{
+		{
+			name:             "old hub version should use JSON",
+			hubVersion:       "0.11.1",
+			expectedUsesCbor: false,
+		},
+		{
+			name:             "non-beta release should use CBOR",
+			hubVersion:       "0.12.0",
+			expectedUsesCbor: true,
+		},
+		{
+			name:             "even newer hub version should use CBOR",
+			hubVersion:       "0.16.4",
+			expectedUsesCbor: true,
+		},
+		{
+			name:             "beta version below release threshold should use JSON",
+			hubVersion:       "0.12.0-beta0",
+			expectedUsesCbor: false,
+		},
+		// {
+		// 	name:             "matching beta version should use CBOR",
+		// 	hubVersion:       "0.12.0-beta2",
+		// 	expectedUsesCbor: true,
+		// },
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			// Reset the global hubVersions map to ensure clean state for each test
+			hubVersions = nil
+
+			agent, err := NewAgent("")
+			require.NoError(t, err)
+
+			// Parse the test version
+			version, err := semver.Parse(tt.hubVersion)
+			require.NoError(t, err)
+
+			// Create test data to encode
+			testData := createTestCombinedData()
+
+			var buf strings.Builder
+			err = agent.writeToSession(&buf, testData, version)
+			require.NoError(t, err)
+
+			encodedData := buf.String()
+			require.NotEmpty(t, encodedData)
+
+			// Verify the encoding format by attempting to decode
+			if tt.expectedUsesCbor {
+				var decodedCbor system.CombinedData
+				err = cbor.Unmarshal([]byte(encodedData), &decodedCbor)
+				assert.NoError(t, err, "Should be valid CBOR data")
+
+				var decodedJson system.CombinedData
+				err = json.Unmarshal([]byte(encodedData), &decodedJson)
+				assert.Error(t, err, "Should not be valid JSON data")
+
+				assert.Equal(t, testData.Info.Hostname, decodedCbor.Info.Hostname)
+				assert.Equal(t, testData.Stats.Cpu, decodedCbor.Stats.Cpu)
+			} else {
+				// Should be JSON - try to decode as JSON
+				var decodedJson system.CombinedData
+				err = json.Unmarshal([]byte(encodedData), &decodedJson)
+				assert.NoError(t, err, "Should be valid JSON data")
+
+				var decodedCbor system.CombinedData
+				err = cbor.Unmarshal([]byte(encodedData), &decodedCbor)
+				assert.Error(t, err, "Should not be valid CBOR data")
+
+				// Verify the decoded JSON data matches our test data
+				assert.Equal(t, testData.Info.Hostname, decodedJson.Info.Hostname)
+				assert.Equal(t, testData.Stats.Cpu, decodedJson.Stats.Cpu)
+
+				// Verify it looks like JSON (starts with '{' and contains readable field names)
+				assert.True(t, strings.HasPrefix(encodedData, "{"), "JSON should start with '{'")
+				assert.Contains(t, encodedData, `"info"`, "JSON should contain readable field names")
+				assert.Contains(t, encodedData, `"stats"`, "JSON should contain readable field names")
+			}
+		})
+	}
+}
+
+// Helper function to create test data for encoding tests
+func createTestCombinedData() *system.CombinedData {
+	return &system.CombinedData{
+		Stats: system.Stats{
+			Cpu:       25.5,
+			Mem:       8589934592, // 8GB
+			MemUsed:   4294967296, // 4GB
+			MemPct:    50.0,
+			DiskTotal: 1099511627776, // 1TB
+			DiskUsed:  549755813888,  // 512GB
+			DiskPct:   50.0,
+		},
+		Info: system.Info{
+			Hostname:     "test-host",
+			Cores:        8,
+			CpuModel:     "Test CPU Model",
+			Uptime:       3600,
+			AgentVersion: "0.12.0",
+			Os:           system.Linux,
+		},
+		Containers: []*container.Stats{
+			{
+				Name: "test-container",
+				Cpu:  10.5,
+				Mem:  1073741824, // 1GB
+			},
+		},
+	}
+}
+
+func TestHubVersionCaching(t *testing.T) {
+	// Reset the global hubVersions map to ensure clean state
+	hubVersions = nil
+
+	agent, err := NewAgent("")
+	require.NoError(t, err)
+
+	ctx1 := &mockSSHContext{
+		sessionID:     "session1",
+		clientVersion: "SSH-2.0-beszel_0.12.0",
+	}
+	ctx2 := &mockSSHContext{
+		sessionID:     "session2",
+		clientVersion: "SSH-2.0-beszel_0.11.0",
+	}
+
+	// First calls should cache the versions
+	v1 := agent.getHubVersion("session1", ctx1)
+	v2 := agent.getHubVersion("session2", ctx2)
+
+	assert.Equal(t, "0.12.0", v1.String())
+	assert.Equal(t, "0.11.0", v2.String())
+
+	// Verify caching by changing context but keeping same session ID
+	ctx1.clientVersion = "SSH-2.0-beszel_0.10.0"
+	v1Cached := agent.getHubVersion("session1", ctx1)
+	assert.Equal(t, "0.12.0", v1Cached.String()) // Should still be cached version
+
+	// New session should get new version
+	ctx3 := &mockSSHContext{
+		sessionID:     "session3",
+		clientVersion: "SSH-2.0-beszel_0.13.0",
+	}
+	v3 := agent.getHubVersion("session3", ctx3)
+	assert.Equal(t, "0.13.0", v3.String())
+}

beszel/internal/agent/system.go

@@ -2,6 +2,7 @@ package agent
 
 import (
 	"beszel"
+	"beszel/internal/agent/battery"
 	"beszel/internal/entities/system"
 	"bufio"
 	"fmt"
@@ -14,6 +15,7 @@ import (
 	"github.com/shirou/gopsutil/v4/cpu"
 	"github.com/shirou/gopsutil/v4/disk"
 	"github.com/shirou/gopsutil/v4/host"
+	"github.com/shirou/gopsutil/v4/load"
 	"github.com/shirou/gopsutil/v4/mem"
 	psutilNet "github.com/shirou/gopsutil/v4/net"
 )
@@ -58,10 +60,10 @@ func (a *Agent) initializeSystemInfo() {
 	}
 
 	// zfs
-	if _, err := getARCSize(); err == nil {
+	if _, err := getARCSize(); err != nil {
-		a.zfs = true
-	} else {
 		slog.Debug("Not monitoring ZFS ARC", "err", err)
+	} else {
+		a.zfs = true
 	}
 }
 
@@ -69,6 +71,11 @@ func (a *Agent) initializeSystemInfo() {
 func (a *Agent) getSystemStats() system.Stats {
 	systemStats := system.Stats{}
 
+	// battery
+	if battery.HasReadableBattery() {
+		systemStats.Battery[0], systemStats.Battery[1], _ = battery.GetBatteryStats()
+	}
+
 	// cpu percent
 	cpuPct, err := cpu.Percent(0, false)
 	if err != nil {
@@ -77,6 +84,16 @@ func (a *Agent) getSystemStats() system.Stats {
 		systemStats.Cpu = twoDecimals(cpuPct[0])
 	}
 
+	// load average
+	if avgstat, err := load.Avg(); err == nil {
+		systemStats.LoadAvg[0] = avgstat.Load1
+		systemStats.LoadAvg[1] = avgstat.Load5
+		systemStats.LoadAvg[2] = avgstat.Load15
+		slog.Debug("Load average", "5m", avgstat.Load5, "15m", avgstat.Load15)
+	} else {
+		slog.Error("Error getting load average", "err", err)
+	}
+
 	// memory
 	if v, err := mem.VirtualMemory(); err == nil {
 		// swap
@@ -163,24 +180,27 @@ func (a *Agent) getSystemStats() system.Stats {
 		a.initializeNetIoStats()
 	}
 	if netIO, err := psutilNet.IOCounters(true); err == nil {
-		secondsElapsed := time.Since(a.netIoStats.Time).Seconds()
+		msElapsed := uint64(time.Since(a.netIoStats.Time).Milliseconds())
 		a.netIoStats.Time = time.Now()
-		bytesSent := uint64(0)
+		totalBytesSent := uint64(0)
-		bytesRecv := uint64(0)
+		totalBytesRecv := uint64(0)
 		// sum all bytes sent and received
 		for _, v := range netIO {
 			// skip if not in valid network interfaces list
 			if _, exists := a.netInterfaces[v.Name]; !exists {
 				continue
 			}
-			bytesSent += v.BytesSent
+			totalBytesSent += v.BytesSent
-			bytesRecv += v.BytesRecv
+			totalBytesRecv += v.BytesRecv
 		}
 		// add to systemStats
-		sentPerSecond := float64(bytesSent-a.netIoStats.BytesSent) / secondsElapsed
+		var bytesSentPerSecond, bytesRecvPerSecond uint64
-		recvPerSecond := float64(bytesRecv-a.netIoStats.BytesRecv) / secondsElapsed
+		if msElapsed > 0 {
-		networkSentPs := bytesToMegabytes(sentPerSecond)
+			bytesSentPerSecond = (totalBytesSent - a.netIoStats.BytesSent) * 1000 / msElapsed
-		networkRecvPs := bytesToMegabytes(recvPerSecond)
+			bytesRecvPerSecond = (totalBytesRecv - a.netIoStats.BytesRecv) * 1000 / msElapsed
+		}
+		networkSentPs := bytesToMegabytes(float64(bytesSentPerSecond))
+		networkRecvPs := bytesToMegabytes(float64(bytesRecvPerSecond))
 		// add check for issue (#150) where sent is a massive number
 		if networkSentPs > 10_000 || networkRecvPs > 10_000 {
 			slog.Warn("Invalid net stats. Resetting.", "sent", networkSentPs, "recv", networkRecvPs)
@@ -195,9 +215,10 @@ func (a *Agent) getSystemStats() system.Stats {
 		} else {
 			systemStats.NetworkSent = networkSentPs
 			systemStats.NetworkRecv = networkRecvPs
+			systemStats.Bandwidth[0], systemStats.Bandwidth[1] = bytesSentPerSecond, bytesRecvPerSecond
 			// update netIoStats
-			a.netIoStats.BytesSent = bytesSent
+			a.netIoStats.BytesSent = totalBytesSent
-			a.netIoStats.BytesRecv = bytesRecv
+			a.netIoStats.BytesRecv = totalBytesRecv
 		}
 	}
 
@@ -240,10 +261,17 @@ func (a *Agent) getSystemStats() system.Stats {
 
 	// update base system info
 	a.systemInfo.Cpu = systemStats.Cpu
+	a.systemInfo.LoadAvg = systemStats.LoadAvg
+	// TODO: remove these in future release in favor of load avg array
+	a.systemInfo.LoadAvg1 = systemStats.LoadAvg[0]
+	a.systemInfo.LoadAvg5 = systemStats.LoadAvg[1]
+	a.systemInfo.LoadAvg15 = systemStats.LoadAvg[2]
 	a.systemInfo.MemPct = systemStats.MemPct
 	a.systemInfo.DiskPct = systemStats.DiskPct
 	a.systemInfo.Uptime, _ = host.Uptime()
+	// TODO: in future release, remove MB bandwidth values in favor of bytes
 	a.systemInfo.Bandwidth = twoDecimals(systemStats.NetworkSent + systemStats.NetworkRecv)
+	a.systemInfo.BandwidthBytes = systemStats.Bandwidth[0] + systemStats.Bandwidth[1]
 	slog.Debug("sysinfo", "data", a.systemInfo)
 
 	return systemStats

beszel/internal/agent/update.go

@@ -1,56 +1,150 @@
 package agent
 
 import (
-	"beszel"
+	"beszel/internal/ghupdate"
 	"fmt"
+	"log"
 	"os"
+	"os/exec"
 	"strings"
-
-	"github.com/blang/semver"
-	"github.com/rhysd/go-github-selfupdate/selfupdate"
 )
 
-// Update updates beszel-agent to the latest version
+// restarter knows how to restart the beszel-agent service.
-func Update() {
+type restarter interface {
-	var latest *selfupdate.Release
+	Restart() error
-	var found bool
+}
-	var err error
+
-	currentVersion := semver.MustParse(beszel.Version)
+type systemdRestarter struct{ cmd string }
-	fmt.Println("beszel-agent", currentVersion)
+
-	fmt.Println("Checking for updates...")
+func (s *systemdRestarter) Restart() error {
-	updater, _ := selfupdate.NewUpdater(selfupdate.Config{
+	// Only restart if the service is active
-		Filters: []string{"beszel-agent"},
+	if err := exec.Command(s.cmd, "is-active", "beszel-agent.service").Run(); err != nil {
-	})
+		return nil
-	latest, found, err = updater.DetectLatest("henrygd/beszel")
+	}
-
+	ghupdate.ColorPrint(ghupdate.ColorYellow, "Restarting beszel-agent.service via systemd…")
-	if err != nil {
+	return exec.Command(s.cmd, "restart", "beszel-agent.service").Run()
-		fmt.Println("Error checking for updates:", err)
+}
-		os.Exit(1)
+
-	}
+type openRCRestarter struct{ cmd string }
-
+
-	if !found {
+func (o *openRCRestarter) Restart() error {
-		fmt.Println("No updates found")
+	if err := exec.Command(o.cmd, "status", "beszel-agent").Run(); err != nil {
-		os.Exit(0)
+		return nil
-	}
+	}
-
+	ghupdate.ColorPrint(ghupdate.ColorYellow, "Restarting beszel-agent via OpenRC…")
-	fmt.Println("Latest version:", latest.Version)
+	return exec.Command(o.cmd, "restart", "beszel-agent").Run()
-
+}
-	if latest.Version.LTE(currentVersion) {
+
-		fmt.Println("You are up to date")
+type openWRTRestarter struct{ cmd string }
-		return
+
-	}
+func (w *openWRTRestarter) Restart() error {
-
+	if err := exec.Command(w.cmd, "running", "beszel-agent").Run(); err != nil {
-	var binaryPath string
+		return nil
-	fmt.Printf("Updating from %s to %s...\n", currentVersion, latest.Version)
+	}
-	binaryPath, err = os.Executable()
+	ghupdate.ColorPrint(ghupdate.ColorYellow, "Restarting beszel-agent via procd…")
-	if err != nil {
+	return exec.Command(w.cmd, "restart", "beszel-agent").Run()
-		fmt.Println("Error getting binary path:", err)
+}
-		os.Exit(1)
+
-	}
+func detectRestarter() restarter {
-	err = selfupdate.UpdateTo(latest.AssetURL, binaryPath)
+	if path, err := exec.LookPath("systemctl"); err == nil {
-	if err != nil {
+		return &systemdRestarter{cmd: path}
-		fmt.Println("Please try rerunning with sudo. Error:", err)
+	}
-		os.Exit(1)
+	if path, err := exec.LookPath("rc-service"); err == nil {
-	}
+		return &openRCRestarter{cmd: path}
-	fmt.Printf("Successfully updated to %s\n\n%s\n", latest.Version, strings.TrimSpace(latest.ReleaseNotes))
+	}
+	if path, err := exec.LookPath("service"); err == nil {
+		return &openWRTRestarter{cmd: path}
+	}
+	return nil
+}
+
+// Update checks GitHub for a newer release of beszel-agent, applies it,
+// fixes SELinux context if needed, and restarts the service.
+func Update(useMirror bool) error {
+	exePath, _ := os.Executable()
+
+	dataDir, err := getDataDir()
+	if err != nil {
+		dataDir = os.TempDir()
+	}
+	updated, err := ghupdate.Update(ghupdate.Config{
+		ArchiveExecutable: "beszel-agent",
+		DataDir:           dataDir,
+		UseMirror:         useMirror,
+	})
+	if err != nil {
+		log.Fatal(err)
+	}
+	if !updated {
+		return nil
+	}
+
+	// make sure the file is executable
+	if err := os.Chmod(exePath, 0755); err != nil {
+		ghupdate.ColorPrintf(ghupdate.ColorYellow, "Warning: failed to set executable permissions: %v", err)
+	}
+	// set ownership to beszel:beszel if possible
+	if chownPath, err := exec.LookPath("chown"); err == nil {
+		if err := exec.Command(chownPath, "beszel:beszel", exePath).Run(); err != nil {
+			ghupdate.ColorPrintf(ghupdate.ColorYellow, "Warning: failed to set file ownership: %v", err)
+		}
+	}
+
+	// 6) Fix SELinux context if necessary
+	if err := handleSELinuxContext(exePath); err != nil {
+		ghupdate.ColorPrintf(ghupdate.ColorYellow, "Warning: SELinux context handling: %v", err)
+	}
+
+	// 7) Restart service if running under a recognised init system
+	if r := detectRestarter(); r != nil {
+		if err := r.Restart(); err != nil {
+			ghupdate.ColorPrintf(ghupdate.ColorYellow, "Warning: failed to restart service: %v", err)
+			ghupdate.ColorPrint(ghupdate.ColorYellow, "Please restart the service manually.")
+		} else {
+			ghupdate.ColorPrint(ghupdate.ColorGreen, "Service restarted successfully")
+		}
+	} else {
+		ghupdate.ColorPrint(ghupdate.ColorYellow, "No supported init system detected; please restart manually if needed.")
+	}
+
+	return nil
+}
+
+// handleSELinuxContext restores or applies the correct SELinux label to the binary.
+func handleSELinuxContext(path string) error {
+	out, err := exec.Command("getenforce").Output()
+	if err != nil {
+		// SELinux not enabled or getenforce not available
+		return nil
+	}
+	state := strings.TrimSpace(string(out))
+	if state == "Disabled" {
+		return nil
+	}
+
+	ghupdate.ColorPrint(ghupdate.ColorYellow, "SELinux is enabled; applying context…")
+	var errs []string
+
+	// Try persistent context via semanage+restorecon
+	if semanagePath, err := exec.LookPath("semanage"); err == nil {
+		if err := exec.Command(semanagePath, "fcontext", "-a", "-t", "bin_t", path).Run(); err != nil {
+			errs = append(errs, "semanage fcontext failed: "+err.Error())
+		} else if restoreconPath, err := exec.LookPath("restorecon"); err == nil {
+			if err := exec.Command(restoreconPath, "-v", path).Run(); err != nil {
+				errs = append(errs, "restorecon failed: "+err.Error())
+			}
+		}
+	}
+
+	// Fallback to temporary context via chcon
+	if chconPath, err := exec.LookPath("chcon"); err == nil {
+		if err := exec.Command(chconPath, "-t", "bin_t", path).Run(); err != nil {
+			errs = append(errs, "chcon failed: "+err.Error())
+		}
+	}
+
+	if len(errs) > 0 {
+		return fmt.Errorf("SELinux context errors: %s", strings.Join(errs, "; "))
+	}
+	return nil
 }

beszel/internal/alerts/alerts.go

@@ -10,13 +10,17 @@ import (
 
 	"github.com/nicholas-fedor/shoutrrr"
 	"github.com/pocketbase/dbx"
-	"github.com/pocketbase/pocketbase/apis"
 	"github.com/pocketbase/pocketbase/core"
 	"github.com/pocketbase/pocketbase/tools/mailer"
 )
 
+type hubLike interface {
+	core.App
+	MakeLink(parts ...string) string
+}
+
 type AlertManager struct {
-	app           core.App
+	hub           hubLike
 	alertQueue    chan alertTask
 	stopChan      chan struct{}
 	pendingAlerts sync.Map
@@ -42,6 +46,7 @@ type SystemAlertStats struct {
 	NetSent      float64            `json:"ns"`
 	NetRecv      float64            `json:"nr"`
 	Temperatures map[string]float32 `json:"t"`
+	LoadAvg      [3]float64         `json:"la"`
 }
 
 type SystemAlertData struct {
@@ -79,19 +84,27 @@ var supportsTitle = map[string]struct{}{
 }
 
 // NewAlertManager creates a new AlertManager instance.
-func NewAlertManager(app core.App) *AlertManager {
+func NewAlertManager(app hubLike) *AlertManager {
 	am := &AlertManager{
-		app:        app,
+		hub:        app,
 		alertQueue: make(chan alertTask),
 		stopChan:   make(chan struct{}),
 	}
+	am.bindEvents()
 	go am.startWorker()
 	return am
 }
 
+// Bind events to the alerts collection lifecycle
+func (am *AlertManager) bindEvents() {
+	am.hub.OnRecordAfterUpdateSuccess("alerts").BindFunc(updateHistoryOnAlertUpdate)
+	am.hub.OnRecordAfterDeleteSuccess("alerts").BindFunc(resolveHistoryOnAlertDelete)
+}
+
+// SendAlert sends an alert to the user
 func (am *AlertManager) SendAlert(data AlertMessageData) error {
 	// get user settings
-	record, err := am.app.FindFirstRecordByFilter(
+	record, err := am.hub.FindFirstRecordByFilter(
 		"user_settings", "user={:user}",
 		dbx.Params{"user": data.UserID},
 	)
@@ -104,12 +117,12 @@ func (am *AlertManager) SendAlert(data AlertMessageData) error {
 		Webhooks: []string{},
 	}
 	if err := record.UnmarshalJSONField("settings", &userAlertSettings); err != nil {
-		am.app.Logger().Error("Failed to unmarshal user settings", "err", err.Error())
+		am.hub.Logger().Error("Failed to unmarshal user settings", "err", err)
 	}
 	// send alerts via webhooks
 	for _, webhook := range userAlertSettings.Webhooks {
 		if err := am.SendShoutrrrAlert(webhook, data.Title, data.Message, data.Link, data.LinkText); err != nil {
-			am.app.Logger().Error("Failed to send shoutrrr alert", "err", err.Error())
+			am.hub.Logger().Error("Failed to send shoutrrr alert", "err", err)
 		}
 	}
 	// send alerts via email
@@ -125,15 +138,15 @@ func (am *AlertManager) SendAlert(data AlertMessageData) error {
 		Subject: data.Title,
 		Text:    data.Message + fmt.Sprintf("\n\n%s", data.Link),
 		From: mail.Address{
-			Address: am.app.Settings().Meta.SenderAddress,
+			Address: am.hub.Settings().Meta.SenderAddress,
-			Name:    am.app.Settings().Meta.SenderName,
+			Name:    am.hub.Settings().Meta.SenderName,
 		},
 	}
-	err = am.app.NewMailClient().Send(&message)
+	err = am.hub.NewMailClient().Send(&message)
 	if err != nil {
 		return err
 	}
-	am.app.Logger().Info("Sent email alert", "to", message.To, "subj", message.Subject)
+	am.hub.Logger().Info("Sent email alert", "to", message.To, "subj", message.Subject)
 	return nil
 }
 
@@ -183,25 +196,23 @@ func (am *AlertManager) SendShoutrrrAlert(notificationUrl, title, message, link,
 	err = shoutrrr.Send(parsedURL.String(), message)
 
 	if err == nil {
-		am.app.Logger().Info("Sent shoutrrr alert", "title", title)
+		am.hub.Logger().Info("Sent shoutrrr alert", "title", title)
 	} else {
-		am.app.Logger().Error("Error sending shoutrrr alert", "err", err.Error())
+		am.hub.Logger().Error("Error sending shoutrrr alert", "err", err)
 		return err
 	}
 	return nil
 }
 
 func (am *AlertManager) SendTestNotification(e *core.RequestEvent) error {
-	info, _ := e.RequestInfo()
+	var data struct {
-	if info.Auth == nil {
+		URL string `json:"url"`
-		return apis.NewForbiddenError("Forbidden", nil)
 	}
-	url := e.Request.URL.Query().Get("url")
+	err := e.BindBody(&data)
-	// log.Println("url", url)
+	if err != nil || data.URL == "" {
-	if url == "" {
+		return e.BadRequestError("URL is required", err)
-		return e.JSON(200, map[string]string{"err": "URL is required"})
 	}
-	err := am.SendShoutrrrAlert(url, "Test Alert", "This is a notification from Beszel.", am.app.Settings().Meta.AppURL, "View Beszel")
+	err = am.SendShoutrrrAlert(data.URL, "Test Alert", "This is a notification from Beszel.", am.hub.Settings().Meta.AppURL, "View Beszel")
 	if err != nil {
 		return e.JSON(200, map[string]string{"err": err.Error()})
 	}

beszel/internal/alerts/alerts_api.go

@@ -0,0 +1,119 @@
+package alerts
+
+import (
+	"database/sql"
+	"errors"
+	"net/http"
+
+	"github.com/pocketbase/dbx"
+	"github.com/pocketbase/pocketbase/core"
+)
+
+// UpsertUserAlerts handles API request to create or update alerts for a user
+// across multiple systems (POST /api/beszel/user-alerts)
+func UpsertUserAlerts(e *core.RequestEvent) error {
+	userID := e.Auth.Id
+
+	reqData := struct {
+		Min       uint8    `json:"min"`
+		Value     float64  `json:"value"`
+		Name      string   `json:"name"`
+		Systems   []string `json:"systems"`
+		Overwrite bool     `json:"overwrite"`
+	}{}
+	err := e.BindBody(&reqData)
+	if err != nil || userID == "" || reqData.Name == "" || len(reqData.Systems) == 0 {
+		return e.BadRequestError("Bad data", err)
+	}
+
+	alertsCollection, err := e.App.FindCachedCollectionByNameOrId("alerts")
+	if err != nil {
+		return err
+	}
+
+	err = e.App.RunInTransaction(func(txApp core.App) error {
+		for _, systemId := range reqData.Systems {
+			// find existing matching alert
+			alertRecord, err := txApp.FindFirstRecordByFilter(alertsCollection,
+				"system={:system} && name={:name} && user={:user}",
+				dbx.Params{"system": systemId, "name": reqData.Name, "user": userID})
+
+			if err != nil && !errors.Is(err, sql.ErrNoRows) {
+				return err
+			}
+
+			// skip if alert already exists and overwrite is not set
+			if !reqData.Overwrite && alertRecord != nil {
+				continue
+			}
+
+			// create new alert if it doesn't exist
+			if alertRecord == nil {
+				alertRecord = core.NewRecord(alertsCollection)
+				alertRecord.Set("user", userID)
+				alertRecord.Set("system", systemId)
+				alertRecord.Set("name", reqData.Name)
+			}
+
+			alertRecord.Set("value", reqData.Value)
+			alertRecord.Set("min", reqData.Min)
+
+			if err := txApp.SaveNoValidate(alertRecord); err != nil {
+				return err
+			}
+		}
+		return nil
+	})
+
+	if err != nil {
+		return err
+	}
+
+	return e.JSON(http.StatusOK, map[string]any{"success": true})
+}
+
+// DeleteUserAlerts handles API request to delete alerts for a user across multiple systems
+// (DELETE /api/beszel/user-alerts)
+func DeleteUserAlerts(e *core.RequestEvent) error {
+	userID := e.Auth.Id
+
+	reqData := struct {
+		AlertName string   `json:"name"`
+		Systems   []string `json:"systems"`
+	}{}
+	err := e.BindBody(&reqData)
+	if err != nil || userID == "" || reqData.AlertName == "" || len(reqData.Systems) == 0 {
+		return e.BadRequestError("Bad data", err)
+	}
+
+	var numDeleted uint16
+
+	err = e.App.RunInTransaction(func(txApp core.App) error {
+		for _, systemId := range reqData.Systems {
+			// Find existing alert to delete
+			alertRecord, err := txApp.FindFirstRecordByFilter("alerts",
+				"system={:system} && name={:name} && user={:user}",
+				dbx.Params{"system": systemId, "name": reqData.AlertName, "user": userID})
+
+			if err != nil {
+				if errors.Is(err, sql.ErrNoRows) {
+					// alert doesn't exist, continue to next system
+					continue
+				}
+				return err
+			}
+
+			if err := txApp.Delete(alertRecord); err != nil {
+				return err
+			}
+			numDeleted++
+		}
+		return nil
+	})
+
+	if err != nil {
+		return err
+	}
+
+	return e.JSON(http.StatusOK, map[string]any{"success": true, "count": numDeleted})
+}

beszel/internal/alerts/alerts_history.go

@@ -0,0 +1,85 @@
+package alerts
+
+import (
+	"time"
+
+	"github.com/pocketbase/dbx"
+	"github.com/pocketbase/pocketbase/core"
+)
+
+// On triggered alert record delete, set matching alert history record to resolved
+func resolveHistoryOnAlertDelete(e *core.RecordEvent) error {
+	if !e.Record.GetBool("triggered") {
+		return e.Next()
+	}
+	_ = resolveAlertHistoryRecord(e.App, e.Record.Id)
+	return e.Next()
+}
+
+// On alert record update, update alert history record
+func updateHistoryOnAlertUpdate(e *core.RecordEvent) error {
+	original := e.Record.Original()
+	new := e.Record
+
+	originalTriggered := original.GetBool("triggered")
+	newTriggered := new.GetBool("triggered")
+
+	// no need to update alert history if triggered state has not changed
+	if originalTriggered == newTriggered {
+		return e.Next()
+	}
+
+	// if new state is triggered, create new alert history record
+	if newTriggered {
+		_, _ = createAlertHistoryRecord(e.App, new)
+		return e.Next()
+	}
+
+	// if new state is not triggered, check for matching alert history record and set it to resolved
+	_ = resolveAlertHistoryRecord(e.App, new.Id)
+	return e.Next()
+}
+
+// resolveAlertHistoryRecord sets the resolved field to the current time
+func resolveAlertHistoryRecord(app core.App, alertRecordID string) error {
+	alertHistoryRecords, err := app.FindRecordsByFilter(
+		"alerts_history",
+		"alert_id={:alert_id} && resolved=null",
+		"-created",
+		1,
+		0,
+		dbx.Params{"alert_id": alertRecordID},
+	)
+	if err != nil {
+		return err
+	}
+	if len(alertHistoryRecords) == 0 {
+		return nil
+	}
+	alertHistoryRecord := alertHistoryRecords[0] // there should be only one record
+	alertHistoryRecord.Set("resolved", time.Now().UTC())
+	err = app.Save(alertHistoryRecord)
+	if err != nil {
+		app.Logger().Error("Failed to resolve alert history", "err", err)
+	}
+	return err
+}
+
+// createAlertHistoryRecord creates a new alert history record
+func createAlertHistoryRecord(app core.App, alertRecord *core.Record) (alertHistoryRecord *core.Record, err error) {
+	alertHistoryCollection, err := app.FindCachedCollectionByNameOrId("alerts_history")
+	if err != nil {
+		return nil, err
+	}
+	alertHistoryRecord = core.NewRecord(alertHistoryCollection)
+	alertHistoryRecord.Set("alert_id", alertRecord.Id)
+	alertHistoryRecord.Set("user", alertRecord.GetString("user"))
+	alertHistoryRecord.Set("system", alertRecord.GetString("system"))
+	alertHistoryRecord.Set("name", alertRecord.GetString("name"))
+	alertHistoryRecord.Set("value", alertRecord.GetFloat("value"))
+	err = app.Save(alertHistoryRecord)
+	if err != nil {
+		app.Logger().Error("Failed to save alert history", "err", err)
+	}
+	return alertHistoryRecord, err
+}

beszel/internal/alerts/alerts_status.go

@@ -2,7 +2,6 @@ package alerts
 
 import (
 	"fmt"
-	"net/url"
 	"strings"
 	"time"
 
@@ -87,7 +86,7 @@ func (am *AlertManager) HandleStatusAlerts(newStatus string, systemRecord *core.
 
 // getSystemStatusAlerts retrieves all "Status" alert records for a given system ID.
 func (am *AlertManager) getSystemStatusAlerts(systemID string) ([]*core.Record, error) {
-	alertRecords, err := am.app.FindAllRecords("alerts", dbx.HashExp{
+	alertRecords, err := am.hub.FindAllRecords("alerts", dbx.HashExp{
 		"system": systemID,
 		"name":   "Status",
 	})
@@ -130,13 +129,21 @@ func (am *AlertManager) handleSystemUp(systemName string, alertRecords []*core.R
 		}
 		// No alert scheduled for this record, send "up" alert
 		if err := am.sendStatusAlert("up", systemName, alertRecord); err != nil {
-			am.app.Logger().Error("Failed to send alert", "err", err.Error())
+			am.hub.Logger().Error("Failed to send alert", "err", err)
 		}
 	}
 }
 
 // sendStatusAlert sends a status alert ("up" or "down") to the users associated with the alert records.
 func (am *AlertManager) sendStatusAlert(alertStatus string, systemName string, alertRecord *core.Record) error {
+	switch alertStatus {
+	case "up":
+		alertRecord.Set("triggered", false)
+	case "down":
+		alertRecord.Set("triggered", true)
+	}
+	am.hub.Save(alertRecord)
+
 	var emoji string
 	if alertStatus == "up" {
 		emoji = "\u2705" // Green checkmark emoji
@@ -147,19 +154,19 @@ func (am *AlertManager) sendStatusAlert(alertStatus string, systemName string, a
 	title := fmt.Sprintf("Connection to %s is %s %v", systemName, alertStatus, emoji)
 	message := strings.TrimSuffix(title, emoji)
 
-	if errs := am.app.ExpandRecord(alertRecord, []string{"user"}, nil); len(errs) > 0 {
+	// if errs := am.hub.ExpandRecord(alertRecord, []string{"user"}, nil); len(errs) > 0 {
-		return errs["user"]
+	// 	return errs["user"]
-	}
+	// }
-	user := alertRecord.ExpandedOne("user")
+	// user := alertRecord.ExpandedOne("user")
-	if user == nil {
+	// if user == nil {
-		return nil
+	// 	return nil
-	}
+	// }
 
 	return am.SendAlert(AlertMessageData{
-		UserID:   user.Id,
+		UserID:   alertRecord.GetString("user"),
 		Title:    title,
 		Message:  message,
-		Link:     am.app.Settings().Meta.AppURL + "/system/" + url.PathEscape(systemName),
+		Link:     am.hub.MakeLink("system", systemName),
 		LinkText: "View " + systemName,
 	})
 }

beszel/internal/alerts/alerts_system.go

@@ -2,12 +2,11 @@ package alerts
 
 import (
 	"beszel/internal/entities/system"
+	"encoding/json"
 	"fmt"
-	"net/url"
 	"strings"
 	"time"
 
-	"github.com/goccy/go-json"
 	"github.com/pocketbase/dbx"
 	"github.com/pocketbase/pocketbase/core"
 	"github.com/pocketbase/pocketbase/tools/types"
@@ -15,8 +14,8 @@ import (
 )
 
 func (am *AlertManager) HandleSystemAlerts(systemRecord *core.Record, data *system.CombinedData) error {
-	alertRecords, err := am.app.FindAllRecords("alerts",
+	alertRecords, err := am.hub.FindAllRecords("alerts",
-		dbx.NewExp("system={:system}", dbx.Params{"system": systemRecord.Id}),
+		dbx.NewExp("system={:system} AND name!='Status'", dbx.Params{"system": systemRecord.Id}),
 	)
 	if err != nil || len(alertRecords) == 0 {
 		// log.Println("no alerts found for system")
@@ -55,6 +54,15 @@ func (am *AlertManager) HandleSystemAlerts(systemRecord *core.Record, data *syst
 			}
 			val = data.Info.DashboardTemp
 			unit = "°C"
+		case "LoadAvg1":
+			val = data.Info.LoadAvg[0]
+			unit = ""
+		case "LoadAvg5":
+			val = data.Info.LoadAvg[1]
+			unit = ""
+		case "LoadAvg15":
+			val = data.Info.LoadAvg[2]
+			unit = ""
 		}
 
 		triggered := alertRecord.GetBool("triggered")
@@ -101,7 +109,7 @@ func (am *AlertManager) HandleSystemAlerts(systemRecord *core.Record, data *syst
 		Created types.DateTime `db:"created"`
 	}{}
 
-	err = am.app.DB().
+	err = am.hub.DB().
 		Select("stats", "created").
 		From("system_stats").
 		Where(dbx.NewExp(
@@ -191,6 +199,12 @@ func (am *AlertManager) HandleSystemAlerts(systemRecord *core.Record, data *syst
 					}
 					alert.mapSums[key] += temp
 				}
+			case "LoadAvg1":
+				alert.val += stats.LoadAvg[0]
+			case "LoadAvg5":
+				alert.val += stats.LoadAvg[1]
+			case "LoadAvg15":
+				alert.val += stats.LoadAvg[2]
 			default:
 				continue
 			}
@@ -248,6 +262,10 @@ func (am *AlertManager) sendSystemAlert(alert SystemAlertData) {
 	if alert.name == "Disk" {
 		alert.name += " usage"
 	}
+	// format LoadAvg5 and LoadAvg15
+	if after, ok := strings.CutPrefix(alert.name, "LoadAvg"); ok {
+		alert.name = after + "m Load"
+	}
 
 	// make title alert name lowercase if not CPU
 	titleAlertName := alert.name
@@ -271,22 +289,15 @@ func (am *AlertManager) sendSystemAlert(alert SystemAlertData) {
 	body := fmt.Sprintf("%s averaged %.2f%s for the previous %v %s.", alert.descriptor, alert.val, alert.unit, alert.min, minutesLabel)
 
 	alert.alertRecord.Set("triggered", alert.triggered)
-	if err := am.app.Save(alert.alertRecord); err != nil {
+	if err := am.hub.Save(alert.alertRecord); err != nil {
-		// app.Logger().Error("failed to save alert record", "err", err.Error())
+		// app.Logger().Error("failed to save alert record", "err", err)
 		return
 	}
-	// expand the user relation and send the alert
+	am.SendAlert(AlertMessageData{
-	if errs := am.app.ExpandRecord(alert.alertRecord, []string{"user"}, nil); len(errs) > 0 {
+		UserID:   alert.alertRecord.GetString("user"),
-		// app.Logger().Error("failed to expand user relation", "errs", errs)
+		Title:    subject,
-		return
+		Message:  body,
-	}
+		Link:     am.hub.MakeLink("system", systemName),
-	if user := alert.alertRecord.ExpandedOne("user"); user != nil {
+		LinkText: "View " + systemName,
-		am.SendAlert(AlertMessageData{
+	})
-			UserID:   user.Id,
-			Title:    subject,
-			Message:  body,
-			Link:     am.app.Settings().Meta.AppURL + "/system/" + url.PathEscape(systemName),
-			LinkText: "View " + systemName,
-		})
-	}
 }

beszel/internal/alerts/alerts_test.go

@@ -0,0 +1,368 @@
+//go:build testing
+// +build testing
+
+package alerts_test
+
+import (
+	"bytes"
+	"encoding/json"
+	"io"
+	"net/http"
+	"strings"
+	"testing"
+
+	beszelTests "beszel/internal/tests"
+
+	"github.com/pocketbase/dbx"
+	"github.com/pocketbase/pocketbase/core"
+	pbTests "github.com/pocketbase/pocketbase/tests"
+	"github.com/stretchr/testify/assert"
+)
+
+// marshal to json and return an io.Reader (for use in ApiScenario.Body)
+func jsonReader(v any) io.Reader {
+	data, err := json.Marshal(v)
+	if err != nil {
+		panic(err)
+	}
+	return bytes.NewReader(data)
+}
+
+func TestUserAlertsApi(t *testing.T) {
+	hub, _ := beszelTests.NewTestHub(t.TempDir())
+	defer hub.Cleanup()
+
+	hub.StartHub()
+
+	user1, _ := beszelTests.CreateUser(hub, "alertstest@example.com", "password")
+	user1Token, _ := user1.NewAuthToken()
+
+	user2, _ := beszelTests.CreateUser(hub, "alertstest2@example.com", "password")
+	user2Token, _ := user2.NewAuthToken()
+
+	system1, _ := beszelTests.CreateRecord(hub, "systems", map[string]any{
+		"name":  "system1",
+		"users": []string{user1.Id},
+		"host":  "127.0.0.1",
+	})
+
+	system2, _ := beszelTests.CreateRecord(hub, "systems", map[string]any{
+		"name":  "system2",
+		"users": []string{user1.Id, user2.Id},
+		"host":  "127.0.0.2",
+	})
+
+	userRecords, _ := hub.CountRecords("users")
+	assert.EqualValues(t, 2, userRecords, "all users should be created")
+
+	systemRecords, _ := hub.CountRecords("systems")
+	assert.EqualValues(t, 2, systemRecords, "all systems should be created")
+
+	testAppFactory := func(t testing.TB) *pbTests.TestApp {
+		return hub.TestApp
+	}
+
+	scenarios := []beszelTests.ApiScenario{
+		{
+			Name:            "GET not implemented - returns index",
+			Method:          http.MethodGet,
+			URL:             "/api/beszel/user-alerts",
+			ExpectedStatus:  200,
+			ExpectedContent: []string{"<html ", "globalThis.BESZEL"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:            "POST no auth",
+			Method:          http.MethodPost,
+			URL:             "/api/beszel/user-alerts",
+			ExpectedStatus:  401,
+			ExpectedContent: []string{"requires valid"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "POST no body",
+			Method: http.MethodPost,
+			URL:    "/api/beszel/user-alerts",
+			Headers: map[string]string{
+				"Authorization": user1Token,
+			},
+			ExpectedStatus:  400,
+			ExpectedContent: []string{"Bad data"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "POST bad data",
+			Method: http.MethodPost,
+			URL:    "/api/beszel/user-alerts",
+			Headers: map[string]string{
+				"Authorization": user1Token,
+			},
+			ExpectedStatus:  400,
+			ExpectedContent: []string{"Bad data"},
+			TestAppFactory:  testAppFactory,
+			Body: jsonReader(map[string]any{
+				"invalidField": "this should cause validation error",
+				"threshold":    "not a number",
+			}),
+		},
+		{
+			Name:   "POST malformed JSON",
+			Method: http.MethodPost,
+			URL:    "/api/beszel/user-alerts",
+			Headers: map[string]string{
+				"Authorization": user1Token,
+			},
+			ExpectedStatus:  400,
+			ExpectedContent: []string{"Bad data"},
+			TestAppFactory:  testAppFactory,
+			Body:            strings.NewReader(`{"alertType": "cpu", "threshold": 80, "enabled": true,}`),
+		},
+		{
+			Name:   "POST valid alert data multiple systems",
+			Method: http.MethodPost,
+			URL:    "/api/beszel/user-alerts",
+			Headers: map[string]string{
+				"Authorization": user1Token,
+			},
+			ExpectedStatus:  200,
+			ExpectedContent: []string{"\"success\":true"},
+			TestAppFactory:  testAppFactory,
+			Body: jsonReader(map[string]any{
+				"name":      "CPU",
+				"value":     69,
+				"min":       9,
+				"systems":   []string{system1.Id, system2.Id},
+				"overwrite": false,
+			}),
+			AfterTestFunc: func(t testing.TB, app *pbTests.TestApp, res *http.Response) {
+				// check total alerts
+				alerts, _ := app.CountRecords("alerts")
+				assert.EqualValues(t, 2, alerts, "should have 2 alerts")
+				// check alert has correct values
+				matchingAlerts, _ := app.CountRecords("alerts", dbx.HashExp{"name": "CPU", "user": user1.Id, "system": system1.Id, "value": 69, "min": 9})
+				assert.EqualValues(t, 1, matchingAlerts, "should have 1 alert")
+			},
+		},
+		{
+			Name:   "POST valid alert data single system",
+			Method: http.MethodPost,
+			URL:    "/api/beszel/user-alerts",
+			Headers: map[string]string{
+				"Authorization": user1Token,
+			},
+			ExpectedStatus:  200,
+			ExpectedContent: []string{"\"success\":true"},
+			TestAppFactory:  testAppFactory,
+			Body: jsonReader(map[string]any{
+				"name":    "Memory",
+				"systems": []string{system1.Id},
+				"value":   90,
+				"min":     10,
+			}),
+			AfterTestFunc: func(t testing.TB, app *pbTests.TestApp, res *http.Response) {
+				user1Alerts, _ := app.CountRecords("alerts", dbx.HashExp{"user": user1.Id})
+				assert.EqualValues(t, 3, user1Alerts, "should have 3 alerts")
+			},
+		},
+		{
+			Name:   "Overwrite: false, should not overwrite existing alert",
+			Method: http.MethodPost,
+			URL:    "/api/beszel/user-alerts",
+			Headers: map[string]string{
+				"Authorization": user1Token,
+			},
+			ExpectedStatus:  200,
+			ExpectedContent: []string{"\"success\":true"},
+			TestAppFactory:  testAppFactory,
+			Body: jsonReader(map[string]any{
+				"name":      "CPU",
+				"value":     45,
+				"min":       5,
+				"systems":   []string{system1.Id},
+				"overwrite": false,
+			}),
+			BeforeTestFunc: func(t testing.TB, app *pbTests.TestApp, e *core.ServeEvent) {
+				beszelTests.ClearCollection(t, app, "alerts")
+				beszelTests.CreateRecord(app, "alerts", map[string]any{
+					"name":   "CPU",
+					"system": system1.Id,
+					"user":   user1.Id,
+					"value":  80,
+					"min":    10,
+				})
+			},
+			AfterTestFunc: func(t testing.TB, app *pbTests.TestApp, res *http.Response) {
+				alerts, _ := app.CountRecords("alerts")
+				assert.EqualValues(t, 1, alerts, "should have 1 alert")
+				alert, _ := app.FindFirstRecordByFilter("alerts", "name = 'CPU' && user = {:user}", dbx.Params{"user": user1.Id})
+				assert.EqualValues(t, 80, alert.Get("value"), "should have 80 as value")
+			},
+		},
+		{
+			Name:   "Overwrite: true, should overwrite existing alert",
+			Method: http.MethodPost,
+			URL:    "/api/beszel/user-alerts",
+			Headers: map[string]string{
+				"Authorization": user2Token,
+			},
+			ExpectedStatus:  200,
+			ExpectedContent: []string{"\"success\":true"},
+			TestAppFactory:  testAppFactory,
+			Body: jsonReader(map[string]any{
+				"name":      "CPU",
+				"value":     45,
+				"min":       5,
+				"systems":   []string{system2.Id},
+				"overwrite": true,
+			}),
+			BeforeTestFunc: func(t testing.TB, app *pbTests.TestApp, e *core.ServeEvent) {
+				beszelTests.ClearCollection(t, app, "alerts")
+				beszelTests.CreateRecord(app, "alerts", map[string]any{
+					"name":   "CPU",
+					"system": system2.Id,
+					"user":   user2.Id,
+					"value":  80,
+					"min":    10,
+				})
+			},
+			AfterTestFunc: func(t testing.TB, app *pbTests.TestApp, res *http.Response) {
+				alerts, _ := app.CountRecords("alerts")
+				assert.EqualValues(t, 1, alerts, "should have 1 alert")
+				alert, _ := app.FindFirstRecordByFilter("alerts", "name = 'CPU' && user = {:user}", dbx.Params{"user": user2.Id})
+				assert.EqualValues(t, 45, alert.Get("value"), "should have 45 as value")
+			},
+		},
+		{
+			Name:            "DELETE no auth",
+			Method:          http.MethodDelete,
+			URL:             "/api/beszel/user-alerts",
+			ExpectedStatus:  401,
+			ExpectedContent: []string{"requires valid"},
+			TestAppFactory:  testAppFactory,
+			Body: jsonReader(map[string]any{
+				"name":    "CPU",
+				"systems": []string{system1.Id},
+			}),
+			BeforeTestFunc: func(t testing.TB, app *pbTests.TestApp, e *core.ServeEvent) {
+				beszelTests.ClearCollection(t, app, "alerts")
+				beszelTests.CreateRecord(app, "alerts", map[string]any{
+					"name":   "CPU",
+					"system": system1.Id,
+					"user":   user1.Id,
+					"value":  80,
+					"min":    10,
+				})
+			},
+			AfterTestFunc: func(t testing.TB, app *pbTests.TestApp, res *http.Response) {
+				alerts, _ := app.CountRecords("alerts")
+				assert.EqualValues(t, 1, alerts, "should have 1 alert")
+			},
+		},
+		{
+			Name:   "DELETE alert",
+			Method: http.MethodDelete,
+			URL:    "/api/beszel/user-alerts",
+			Headers: map[string]string{
+				"Authorization": user1Token,
+			},
+			ExpectedStatus:  200,
+			ExpectedContent: []string{"\"count\":1", "\"success\":true"},
+			TestAppFactory:  testAppFactory,
+			Body: jsonReader(map[string]any{
+				"name":    "CPU",
+				"systems": []string{system1.Id},
+			}),
+			BeforeTestFunc: func(t testing.TB, app *pbTests.TestApp, e *core.ServeEvent) {
+				beszelTests.ClearCollection(t, app, "alerts")
+				beszelTests.CreateRecord(app, "alerts", map[string]any{
+					"name":   "CPU",
+					"system": system1.Id,
+					"user":   user1.Id,
+					"value":  80,
+					"min":    10,
+				})
+			},
+			AfterTestFunc: func(t testing.TB, app *pbTests.TestApp, res *http.Response) {
+				alerts, _ := app.CountRecords("alerts")
+				assert.Zero(t, alerts, "should have 0 alerts")
+			},
+		},
+		{
+			Name:   "DELETE alert multiple systems",
+			Method: http.MethodDelete,
+			URL:    "/api/beszel/user-alerts",
+			Headers: map[string]string{
+				"Authorization": user1Token,
+			},
+			ExpectedStatus:  200,
+			ExpectedContent: []string{"\"count\":2", "\"success\":true"},
+			TestAppFactory:  testAppFactory,
+			Body: jsonReader(map[string]any{
+				"name":    "Memory",
+				"systems": []string{system1.Id, system2.Id},
+			}),
+			BeforeTestFunc: func(t testing.TB, app *pbTests.TestApp, e *core.ServeEvent) {
+				beszelTests.ClearCollection(t, app, "alerts")
+				for _, systemId := range []string{system1.Id, system2.Id} {
+					_, err := beszelTests.CreateRecord(app, "alerts", map[string]any{
+						"name":   "Memory",
+						"system": systemId,
+						"user":   user1.Id,
+						"value":  90,
+						"min":    10,
+					})
+					assert.NoError(t, err, "should create alert")
+				}
+				alerts, _ := app.CountRecords("alerts")
+				assert.EqualValues(t, 2, alerts, "should have 2 alerts")
+			},
+			AfterTestFunc: func(t testing.TB, app *pbTests.TestApp, res *http.Response) {
+				alerts, _ := app.CountRecords("alerts")
+				assert.Zero(t, alerts, "should have 0 alerts")
+			},
+		},
+		{
+			Name:   "User 2 should not be able to delete alert of user 1",
+			Method: http.MethodDelete,
+			URL:    "/api/beszel/user-alerts",
+			Headers: map[string]string{
+				"Authorization": user2Token,
+			},
+			ExpectedStatus:  200,
+			ExpectedContent: []string{"\"count\":1", "\"success\":true"},
+			TestAppFactory:  testAppFactory,
+			Body: jsonReader(map[string]any{
+				"name":    "CPU",
+				"systems": []string{system2.Id},
+			}),
+			BeforeTestFunc: func(t testing.TB, app *pbTests.TestApp, e *core.ServeEvent) {
+				beszelTests.ClearCollection(t, app, "alerts")
+				for _, user := range []string{user1.Id, user2.Id} {
+					beszelTests.CreateRecord(app, "alerts", map[string]any{
+						"name":   "CPU",
+						"system": system2.Id,
+						"user":   user,
+						"value":  80,
+						"min":    10,
+					})
+				}
+				alerts, _ := app.CountRecords("alerts")
+				assert.EqualValues(t, 2, alerts, "should have 2 alerts")
+				user1AlertCount, _ := app.CountRecords("alerts", dbx.HashExp{"user": user1.Id})
+				assert.EqualValues(t, 1, user1AlertCount, "should have 1 alert")
+				user2AlertCount, _ := app.CountRecords("alerts", dbx.HashExp{"user": user2.Id})
+				assert.EqualValues(t, 1, user2AlertCount, "should have 1 alert")
+			},
+			AfterTestFunc: func(t testing.TB, app *pbTests.TestApp, res *http.Response) {
+				user1AlertCount, _ := app.CountRecords("alerts", dbx.HashExp{"user": user1.Id})
+				assert.EqualValues(t, 1, user1AlertCount, "should have 1 alert")
+				user2AlertCount, _ := app.CountRecords("alerts", dbx.HashExp{"user": user2.Id})
+				assert.Zero(t, user2AlertCount, "should have 0 alerts")
+			},
+		},
+	}
+
+	for _, scenario := range scenarios {
+		scenario.Test(t)
+	}
+}

beszel/internal/common/common-ssh.go

@@ -0,0 +1,10 @@
+package common
+
+var (
+	// Allowed ssh key exchanges
+	DefaultKeyExchanges = []string{"curve25519-sha256"}
+	// Allowed ssh macs
+	DefaultMACs = []string{"hmac-sha2-256-etm@openssh.com"}
+	// Allowed ssh ciphers
+	DefaultCiphers = []string{"chacha20-poly1305@openssh.com"}
+)

beszel/internal/common/common-ws.go

@@ -0,0 +1,32 @@
+package common
+
+type WebSocketAction = uint8
+
+// Not implemented yet
+// type AgentError = uint8
+
+const (
+	// Request system data from agent
+	GetData WebSocketAction = iota
+	// Check the fingerprint of the agent
+	CheckFingerprint
+)
+
+// HubRequest defines the structure for requests sent from hub to agent.
+type HubRequest[T any] struct {
+	Action WebSocketAction `cbor:"0,keyasint"`
+	Data   T               `cbor:"1,keyasint,omitempty,omitzero"`
+	// Error  AgentError      `cbor:"error,omitempty,omitzero"`
+}
+
+type FingerprintRequest struct {
+	Signature   []byte `cbor:"0,keyasint"`
+	NeedSysInfo bool   `cbor:"1,keyasint"` // For universal token system creation
+}
+
+type FingerprintResponse struct {
+	Fingerprint string `cbor:"0,keyasint"`
+	// Optional system info for universal token system creation
+	Hostname string `cbor:"1,keyasint,omitempty,omitzero"`
+	Port     string `cbor:"2,keyasint,omitempty,omitzero"`
+}

beszel/internal/entities/container/container.go

@@ -34,10 +34,16 @@ type ApiStats struct {
 	MemoryStats MemoryStats `json:"memory_stats"`
 }
 
-func (s *ApiStats) CalculateCpuPercentLinux(prevCpuUsage [2]uint64) float64 {
+func (s *ApiStats) CalculateCpuPercentLinux(prevCpuContainer uint64, prevCpuSystem uint64) float64 {
-	cpuDelta := s.CPUStats.CPUUsage.TotalUsage - prevCpuUsage[0]
+	cpuDelta := s.CPUStats.CPUUsage.TotalUsage - prevCpuContainer
-	systemDelta := s.CPUStats.SystemUsage - prevCpuUsage[1]
+	systemDelta := s.CPUStats.SystemUsage - prevCpuSystem
-	return float64(cpuDelta) / float64(systemDelta) * 100
+
+	// Avoid division by zero and handle first run case
+	if systemDelta == 0 || prevCpuContainer == 0 {
+		return 0.0
+	}
+
+	return float64(cpuDelta) / float64(systemDelta) * 100.0
 }
 
 // from: https://github.com/docker/cli/blob/master/cli/command/container/stats_helpers.go#L185
@@ -99,12 +105,14 @@ type prevNetStats struct {
 
 // Docker container stats
 type Stats struct {
-	Name        string       `json:"n"`
+	Name        string  `json:"n" cbor:"0,keyasint"`
-	Cpu         float64      `json:"c"`
+	Cpu         float64 `json:"c" cbor:"1,keyasint"`
-	Mem         float64      `json:"m"`
+	Mem         float64 `json:"m" cbor:"2,keyasint"`
-	NetworkSent float64      `json:"ns"`
+	NetworkSent float64 `json:"ns" cbor:"3,keyasint"`
-	NetworkRecv float64      `json:"nr"`
+	NetworkRecv float64 `json:"nr" cbor:"4,keyasint"`
-	PrevCpu     [2]uint64    `json:"-"`
+	// PrevCpu     [2]uint64    `json:"-"`
-	PrevNet     prevNetStats `json:"-"`
+	CpuSystem    uint64       `json:"-"`
-	PrevRead    time.Time    `json:"-"`
+	CpuContainer uint64       `json:"-"`
+	PrevNet      prevNetStats `json:"-"`
+	PrevReadTime time.Time    `json:"-"`
 }

beszel/internal/entities/system/system.go

@@ -8,38 +8,47 @@ import (
 )
 
 type Stats struct {
-	Cpu            float64             `json:"cpu"`
+	Cpu            float64             `json:"cpu" cbor:"0,keyasint"`
-	MaxCpu         float64             `json:"cpum,omitempty"`
+	MaxCpu         float64             `json:"cpum,omitempty" cbor:"1,keyasint,omitempty"`
-	Mem            float64             `json:"m"`
+	Mem            float64             `json:"m" cbor:"2,keyasint"`
-	MemUsed        float64             `json:"mu"`
+	MemUsed        float64             `json:"mu" cbor:"3,keyasint"`
-	MemPct         float64             `json:"mp"`
+	MemPct         float64             `json:"mp" cbor:"4,keyasint"`
-	MemBuffCache   float64             `json:"mb"`
+	MemBuffCache   float64             `json:"mb" cbor:"5,keyasint"`
-	MemZfsArc      float64             `json:"mz,omitempty"` // ZFS ARC memory
+	MemZfsArc      float64             `json:"mz,omitempty" cbor:"6,keyasint,omitempty"` // ZFS ARC memory
-	Swap           float64             `json:"s,omitempty"`
+	Swap           float64             `json:"s,omitempty" cbor:"7,keyasint,omitempty"`
-	SwapUsed       float64             `json:"su,omitempty"`
+	SwapUsed       float64             `json:"su,omitempty" cbor:"8,keyasint,omitempty"`
-	DiskTotal      float64             `json:"d"`
+	DiskTotal      float64             `json:"d" cbor:"9,keyasint"`
-	DiskUsed       float64             `json:"du"`
+	DiskUsed       float64             `json:"du" cbor:"10,keyasint"`
-	DiskPct        float64             `json:"dp"`
+	DiskPct        float64             `json:"dp" cbor:"11,keyasint"`
-	DiskReadPs     float64             `json:"dr"`
+	DiskReadPs     float64             `json:"dr" cbor:"12,keyasint"`
-	DiskWritePs    float64             `json:"dw"`
+	DiskWritePs    float64             `json:"dw" cbor:"13,keyasint"`
-	MaxDiskReadPs  float64             `json:"drm,omitempty"`
+	MaxDiskReadPs  float64             `json:"drm,omitempty" cbor:"14,keyasint,omitempty"`
-	MaxDiskWritePs float64             `json:"dwm,omitempty"`
+	MaxDiskWritePs float64             `json:"dwm,omitempty" cbor:"15,keyasint,omitempty"`
-	NetworkSent    float64             `json:"ns"`
+	NetworkSent    float64             `json:"ns" cbor:"16,keyasint"`
-	NetworkRecv    float64             `json:"nr"`
+	NetworkRecv    float64             `json:"nr" cbor:"17,keyasint"`
-	MaxNetworkSent float64             `json:"nsm,omitempty"`
+	MaxNetworkSent float64             `json:"nsm,omitempty" cbor:"18,keyasint,omitempty"`
-	MaxNetworkRecv float64             `json:"nrm,omitempty"`
+	MaxNetworkRecv float64             `json:"nrm,omitempty" cbor:"19,keyasint,omitempty"`
-	Temperatures   map[string]float64  `json:"t,omitempty"`
+	Temperatures   map[string]float64  `json:"t,omitempty" cbor:"20,keyasint,omitempty"`
-	ExtraFs        map[string]*FsStats `json:"efs,omitempty"`
+	ExtraFs        map[string]*FsStats `json:"efs,omitempty" cbor:"21,keyasint,omitempty"`
-	GPUData        map[string]GPUData  `json:"g,omitempty"`
+	GPUData        map[string]GPUData  `json:"g,omitempty" cbor:"22,keyasint,omitempty"`
+	LoadAvg1       float64             `json:"l1,omitempty" cbor:"23,keyasint,omitempty"`
+	LoadAvg5       float64             `json:"l5,omitempty" cbor:"24,keyasint,omitempty"`
+	LoadAvg15      float64             `json:"l15,omitempty" cbor:"25,keyasint,omitempty"`
+	Bandwidth      [2]uint64           `json:"b,omitzero" cbor:"26,keyasint,omitzero"`  // [sent bytes, recv bytes]
+	MaxBandwidth   [2]uint64           `json:"bm,omitzero" cbor:"27,keyasint,omitzero"` // [sent bytes, recv bytes]
+	// TODO: remove other load fields in future release in favor of load avg array
+	LoadAvg [3]float64 `json:"la,omitempty" cbor:"28,keyasint"`
+	Battery [2]uint8   `json:"bat,omitzero" cbor:"29,keyasint,omitzero"` // [percent, charge state, current]
+	MaxMem  float64    `json:"mm,omitempty" cbor:"30,keyasint,omitempty"`
 }
 
 type GPUData struct {
-	Name        string  `json:"n"`
+	Name        string  `json:"n" cbor:"0,keyasint"`
 	Temperature float64 `json:"-"`
-	MemoryUsed  float64 `json:"mu,omitempty"`
+	MemoryUsed  float64 `json:"mu,omitempty" cbor:"1,keyasint,omitempty"`
-	MemoryTotal float64 `json:"mt,omitempty"`
+	MemoryTotal float64 `json:"mt,omitempty" cbor:"2,keyasint,omitempty"`
-	Usage       float64 `json:"u"`
+	Usage       float64 `json:"u" cbor:"3,keyasint"`
-	Power       float64 `json:"p,omitempty"`
+	Power       float64 `json:"p,omitempty" cbor:"4,keyasint,omitempty"`
 	Count       float64 `json:"-"`
 }
 
@@ -47,14 +56,14 @@ type FsStats struct {
 	Time           time.Time `json:"-"`
 	Root           bool      `json:"-"`
 	Mountpoint     string    `json:"-"`
-	DiskTotal      float64   `json:"d"`
+	DiskTotal      float64   `json:"d" cbor:"0,keyasint"`
-	DiskUsed       float64   `json:"du"`
+	DiskUsed       float64   `json:"du" cbor:"1,keyasint"`
 	TotalRead      uint64    `json:"-"`
 	TotalWrite     uint64    `json:"-"`
-	DiskReadPs     float64   `json:"r"`
+	DiskReadPs     float64   `json:"r" cbor:"2,keyasint"`
-	DiskWritePs    float64   `json:"w"`
+	DiskWritePs    float64   `json:"w" cbor:"3,keyasint"`
-	MaxDiskReadPS  float64   `json:"rm,omitempty"`
+	MaxDiskReadPS  float64   `json:"rm,omitempty" cbor:"4,keyasint,omitempty"`
-	MaxDiskWritePS float64   `json:"wm,omitempty"`
+	MaxDiskWritePS float64   `json:"wm,omitempty" cbor:"5,keyasint,omitempty"`
 }
 
 type NetIoStats struct {
@@ -64,7 +73,7 @@ type NetIoStats struct {
 	Name      string
 }
 
-type Os uint8
+type Os = uint8
 
 const (
 	Linux Os = iota
@@ -74,26 +83,32 @@ const (
 )
 
 type Info struct {
-	Hostname      string  `json:"h"`
+	Hostname       string  `json:"h" cbor:"0,keyasint"`
-	KernelVersion string  `json:"k,omitempty"`
+	KernelVersion  string  `json:"k,omitempty" cbor:"1,keyasint,omitempty"`
-	Cores         int     `json:"c"`
+	Cores          int     `json:"c" cbor:"2,keyasint"`
-	Threads       int     `json:"t,omitempty"`
+	Threads        int     `json:"t,omitempty" cbor:"3,keyasint,omitempty"`
-	CpuModel      string  `json:"m"`
+	CpuModel       string  `json:"m" cbor:"4,keyasint"`
-	Uptime        uint64  `json:"u"`
+	Uptime         uint64  `json:"u" cbor:"5,keyasint"`
-	Cpu           float64 `json:"cpu"`
+	Cpu            float64 `json:"cpu" cbor:"6,keyasint"`
-	MemPct        float64 `json:"mp"`
+	MemPct         float64 `json:"mp" cbor:"7,keyasint"`
-	DiskPct       float64 `json:"dp"`
+	DiskPct        float64 `json:"dp" cbor:"8,keyasint"`
-	Bandwidth     float64 `json:"b"`
+	Bandwidth      float64 `json:"b" cbor:"9,keyasint"`
-	AgentVersion  string  `json:"v"`
+	AgentVersion   string  `json:"v" cbor:"10,keyasint"`
-	Podman        bool    `json:"p,omitempty"`
+	Podman         bool    `json:"p,omitempty" cbor:"11,keyasint,omitempty"`
-	GpuPct        float64 `json:"g,omitempty"`
+	GpuPct         float64 `json:"g,omitempty" cbor:"12,keyasint,omitempty"`
-	DashboardTemp float64 `json:"dt,omitempty"`
+	DashboardTemp  float64 `json:"dt,omitempty" cbor:"13,keyasint,omitempty"`
-	Os            Os      `json:"os"`
+	Os             Os      `json:"os" cbor:"14,keyasint"`
+	LoadAvg1       float64 `json:"l1,omitempty" cbor:"15,keyasint,omitempty"`
+	LoadAvg5       float64 `json:"l5,omitempty" cbor:"16,keyasint,omitempty"`
+	LoadAvg15      float64 `json:"l15,omitempty" cbor:"17,keyasint,omitempty"`
+	BandwidthBytes uint64  `json:"bb" cbor:"18,keyasint"`
+	// TODO: remove load fields in future release in favor of load avg array
+	LoadAvg [3]float64 `json:"la,omitempty" cbor:"19,keyasint"`
 }
 
 // Final data structure to return to the hub
 type CombinedData struct {
-	Stats      Stats              `json:"stats"`
+	Stats      Stats              `json:"stats" cbor:"0,keyasint"`
-	Info       Info               `json:"info"`
+	Info       Info               `json:"info" cbor:"1,keyasint"`
-	Containers []*container.Stats `json:"container"`
+	Containers []*container.Stats `json:"container" cbor:"2,keyasint"`
 }

beszel/internal/ghupdate/extract.go

@@ -0,0 +1,140 @@
+package ghupdate
+
+import (
+	"archive/tar"
+	"archive/zip"
+	"compress/gzip"
+	"fmt"
+	"io"
+	"os"
+	"path/filepath"
+	"strings"
+)
+
+// extract extracts an archive file to the destination directory.
+// Supports .zip and .tar.gz files based on the file extension.
+func extract(srcPath, destDir string) error {
+	if strings.HasSuffix(srcPath, ".tar.gz") {
+		return extractTarGz(srcPath, destDir)
+	}
+	// Default to zip extraction
+	return extractZip(srcPath, destDir)
+}
+
+// extractTarGz extracts a tar.gz archive to the destination directory.
+func extractTarGz(srcPath, destDir string) error {
+	src, err := os.Open(srcPath)
+	if err != nil {
+		return err
+	}
+	defer src.Close()
+
+	gz, err := gzip.NewReader(src)
+	if err != nil {
+		return err
+	}
+	defer gz.Close()
+
+	tr := tar.NewReader(gz)
+
+	for {
+		header, err := tr.Next()
+		if err == io.EOF {
+			break
+		}
+		if err != nil {
+			return err
+		}
+
+		if header.Typeflag == tar.TypeDir {
+			if err := os.MkdirAll(filepath.Join(destDir, header.Name), 0755); err != nil {
+				return err
+			}
+			continue
+		}
+
+		if err := os.MkdirAll(filepath.Dir(filepath.Join(destDir, header.Name)), 0755); err != nil {
+			return err
+		}
+
+		outFile, err := os.Create(filepath.Join(destDir, header.Name))
+		if err != nil {
+			return err
+		}
+
+		if _, err := io.Copy(outFile, tr); err != nil {
+			outFile.Close()
+			return err
+		}
+		outFile.Close()
+	}
+
+	return nil
+}
+
+// extractZip extracts the zip archive at "src" to "dest".
+//
+// Note that only dirs and regular files will be extracted.
+// Symbolic links, named pipes, sockets, or any other irregular files
+// are skipped because they come with too many edge cases and ambiguities.
+func extractZip(src, dest string) error {
+	zr, err := zip.OpenReader(src)
+	if err != nil {
+		return err
+	}
+	defer zr.Close()
+
+	// normalize dest path to check later for Zip Slip
+	dest = filepath.Clean(dest) + string(os.PathSeparator)
+
+	for _, f := range zr.File {
+		err := extractFile(f, dest)
+		if err != nil {
+			return err
+		}
+	}
+
+	return nil
+}
+
+// extractFile extracts the provided zipFile into "basePath/zipFileName" path,
+// creating all the necessary path directories.
+func extractFile(zipFile *zip.File, basePath string) error {
+	path := filepath.Join(basePath, zipFile.Name)
+
+	// check for Zip Slip
+	if !strings.HasPrefix(path, basePath) {
+		return fmt.Errorf("invalid file path: %s", path)
+	}
+
+	r, err := zipFile.Open()
+	if err != nil {
+		return err
+	}
+	defer r.Close()
+
+	// allow only dirs or regular files
+	if zipFile.FileInfo().IsDir() {
+		if err := os.MkdirAll(path, os.ModePerm); err != nil {
+			return err
+		}
+	} else if zipFile.FileInfo().Mode().IsRegular() {
+		// ensure that the file path directories are created
+		if err := os.MkdirAll(filepath.Dir(path), os.ModePerm); err != nil {
+			return err
+		}
+
+		f, err := os.OpenFile(path, os.O_WRONLY|os.O_CREATE|os.O_TRUNC, zipFile.Mode())
+		if err != nil {
+			return err
+		}
+		defer f.Close()
+
+		_, err = io.Copy(f, r)
+		if err != nil {
+			return err
+		}
+	}
+
+	return nil
+}

beszel/internal/ghupdate/ghupdate.go

@@ -0,0 +1,348 @@
+// Package ghupdate implements a new command to self update the current
+// executable with the latest GitHub release. This is based on PocketBase's
+// ghupdate package with modifications.
+package ghupdate
+
+import (
+	"beszel"
+	"context"
+	"encoding/json"
+	"fmt"
+	"io"
+	"log/slog"
+	"net/http"
+	"os"
+	"path/filepath"
+	"runtime"
+	"strings"
+
+	"github.com/blang/semver"
+)
+
+// Minimal color functions using ANSI escape codes
+const (
+	colorReset  = "\033[0m"
+	ColorYellow = "\033[33m"
+	ColorGreen  = "\033[32m"
+	colorCyan   = "\033[36m"
+	colorGray   = "\033[90m"
+)
+
+func ColorPrint(color, text string) {
+	fmt.Println(color + text + colorReset)
+}
+
+func ColorPrintf(color, format string, args ...interface{}) {
+	fmt.Printf(color+format+colorReset+"\n", args...)
+}
+
+// HttpClient is a base HTTP client interface (usually used for test purposes).
+type HttpClient interface {
+	Do(req *http.Request) (*http.Response, error)
+}
+
+// Config defines the config options of the ghupdate plugin.
+//
+// NB! This plugin is considered experimental and its config options may change in the future.
+type Config struct {
+	// Owner specifies the account owner of the repository (default to "pocketbase").
+	Owner string
+
+	// Repo specifies the name of the repository (default to "pocketbase").
+	Repo string
+
+	// ArchiveExecutable specifies the name of the executable file in the release archive
+	// (default to "pocketbase"; an additional ".exe" check is also performed as a fallback).
+	ArchiveExecutable string
+
+	// Optional context to use when fetching and downloading the latest release.
+	Context context.Context
+
+	// The HTTP client to use when fetching and downloading the latest release.
+	// Defaults to `http.DefaultClient`.
+	HttpClient HttpClient
+
+	// The data directory to use when fetching and downloading the latest release.
+	DataDir string
+
+	// UseMirror specifies whether to use the beszel.dev mirror instead of GitHub API.
+	// When false (default), always uses api.github.com. When true, uses gh.beszel.dev.
+	UseMirror bool
+}
+
+type updater struct {
+	config         Config
+	currentVersion string
+}
+
+func Update(config Config) (updated bool, err error) {
+	p := &updater{
+		currentVersion: beszel.Version,
+		config:         config,
+	}
+
+	return p.update()
+}
+
+func (p *updater) update() (updated bool, err error) {
+	ColorPrint(ColorYellow, "Fetching release information...")
+
+	if p.config.DataDir == "" {
+		p.config.DataDir = os.TempDir()
+	}
+
+	if p.config.Owner == "" {
+		p.config.Owner = "henrygd"
+	}
+
+	if p.config.Repo == "" {
+		p.config.Repo = "beszel"
+	}
+
+	if p.config.Context == nil {
+		p.config.Context = context.Background()
+	}
+
+	if p.config.HttpClient == nil {
+		p.config.HttpClient = http.DefaultClient
+	}
+
+	var latest *release
+	var useMirror bool
+
+	// Determine the API endpoint based on UseMirror flag
+	apiURL := fmt.Sprintf("https://api.github.com/repos/%s/%s/releases/latest", p.config.Owner, p.config.Repo)
+	if p.config.UseMirror {
+		useMirror = true
+		apiURL = fmt.Sprintf("https://gh.beszel.dev/repos/%s/%s/releases/latest?api=true", p.config.Owner, p.config.Repo)
+		ColorPrint(ColorYellow, "Using mirror for update.")
+	}
+
+	latest, err = fetchLatestRelease(
+		p.config.Context,
+		p.config.HttpClient,
+		apiURL,
+	)
+	if err != nil {
+		return false, err
+	}
+
+	currentVersion := semver.MustParse(strings.TrimPrefix(p.currentVersion, "v"))
+	newVersion := semver.MustParse(strings.TrimPrefix(latest.Tag, "v"))
+
+	if newVersion.LTE(currentVersion) {
+		ColorPrintf(ColorGreen, "You already have the latest version %s.", p.currentVersion)
+		return false, nil
+	}
+
+	suffix := archiveSuffix(p.config.ArchiveExecutable, runtime.GOOS, runtime.GOARCH)
+	asset, err := latest.findAssetBySuffix(suffix)
+	if err != nil {
+		return false, err
+	}
+
+	releaseDir := filepath.Join(p.config.DataDir, ".beszel_update")
+	defer os.RemoveAll(releaseDir)
+
+	ColorPrintf(ColorYellow, "Downloading %s...", asset.Name)
+
+	// download the release asset
+	assetPath := filepath.Join(releaseDir, asset.Name)
+	if err := downloadFile(p.config.Context, p.config.HttpClient, asset.DownloadUrl, assetPath, useMirror); err != nil {
+		return false, err
+	}
+
+	ColorPrintf(ColorYellow, "Extracting %s...", asset.Name)
+
+	extractDir := filepath.Join(releaseDir, "extracted_"+asset.Name)
+	defer os.RemoveAll(extractDir)
+
+	// Extract the archive (automatically detects format)
+	if err := extract(assetPath, extractDir); err != nil {
+		return false, err
+	}
+
+	ColorPrint(ColorYellow, "Replacing the executable...")
+
+	oldExec, err := os.Executable()
+	if err != nil {
+		return false, err
+	}
+	renamedOldExec := oldExec + ".old"
+	defer os.Remove(renamedOldExec)
+
+	newExec := filepath.Join(extractDir, p.config.ArchiveExecutable)
+	if _, err := os.Stat(newExec); err != nil {
+		// try again with an .exe extension
+		newExec = newExec + ".exe"
+		if _, fallbackErr := os.Stat(newExec); fallbackErr != nil {
+			return false, fmt.Errorf("the executable in the extracted path is missing or it is inaccessible: %v, %v", err, fallbackErr)
+		}
+	}
+
+	// rename the current executable
+	if err := os.Rename(oldExec, renamedOldExec); err != nil {
+		return false, fmt.Errorf("failed to rename the current executable: %w", err)
+	}
+
+	tryToRevertExecChanges := func() {
+		if revertErr := os.Rename(renamedOldExec, oldExec); revertErr != nil {
+			slog.Debug(
+				"Failed to revert executable",
+				slog.String("old", renamedOldExec),
+				slog.String("new", oldExec),
+				slog.String("error", revertErr.Error()),
+			)
+		}
+	}
+
+	// replace with the extracted binary
+	if err := os.Rename(newExec, oldExec); err != nil {
+		// If rename fails due to cross-device link, try copying instead
+		if isCrossDeviceError(err) {
+			if err := copyFile(newExec, oldExec); err != nil {
+				tryToRevertExecChanges()
+				return false, fmt.Errorf("failed replacing the executable: %w", err)
+			}
+		} else {
+			tryToRevertExecChanges()
+			return false, fmt.Errorf("failed replacing the executable: %w", err)
+		}
+	}
+
+	ColorPrint(colorGray, "---")
+	ColorPrint(ColorGreen, "Update completed successfully!")
+
+	// print the release notes
+	if latest.Body != "" {
+		fmt.Print("\n")
+		releaseNotes := strings.TrimSpace(strings.Replace(latest.Body, "> _To update the prebuilt executable you can run `./"+p.config.ArchiveExecutable+" update`._", "", 1))
+		ColorPrint(colorCyan, releaseNotes)
+		fmt.Print("\n")
+	}
+
+	return true, nil
+}
+
+func fetchLatestRelease(
+	ctx context.Context,
+	client HttpClient,
+	url string,
+) (*release, error) {
+	req, err := http.NewRequestWithContext(ctx, "GET", url, nil)
+	if err != nil {
+		return nil, err
+	}
+
+	res, err := client.Do(req)
+	if err != nil {
+		return nil, err
+	}
+	defer res.Body.Close()
+
+	rawBody, err := io.ReadAll(res.Body)
+	if err != nil {
+		return nil, err
+	}
+
+	// http.Client doesn't treat non 2xx responses as error
+	if res.StatusCode >= 400 {
+		return nil, fmt.Errorf(
+			"(%d) failed to fetch latest releases:\n%s",
+			res.StatusCode,
+			string(rawBody),
+		)
+	}
+
+	result := &release{}
+	if err := json.Unmarshal(rawBody, result); err != nil {
+		return nil, err
+	}
+
+	return result, nil
+}
+
+func downloadFile(
+	ctx context.Context,
+	client HttpClient,
+	url string,
+	destPath string,
+	useMirror bool,
+) error {
+	if useMirror {
+		url = strings.Replace(url, "github.com", "gh.beszel.dev", 1)
+	}
+	req, err := http.NewRequestWithContext(ctx, "GET", url, nil)
+	if err != nil {
+		return err
+	}
+
+	res, err := client.Do(req)
+	if err != nil {
+		return err
+	}
+	defer res.Body.Close()
+
+	// http.Client doesn't treat non 2xx responses as error
+	if res.StatusCode >= 400 {
+		return fmt.Errorf("(%d) failed to send download file request", res.StatusCode)
+	}
+
+	// ensure that the dest parent dir(s) exist
+	if err := os.MkdirAll(filepath.Dir(destPath), os.ModePerm); err != nil {
+		return err
+	}
+
+	dest, err := os.Create(destPath)
+	if err != nil {
+		return err
+	}
+	defer dest.Close()
+
+	if _, err := io.Copy(dest, res.Body); err != nil {
+		return err
+	}
+
+	return nil
+}
+
+// isCrossDeviceError checks if the error is due to a cross-device link
+func isCrossDeviceError(err error) bool {
+	return err != nil && (strings.Contains(err.Error(), "cross-device") ||
+		strings.Contains(err.Error(), "EXDEV"))
+}
+
+// copyFile copies a file from src to dst, preserving permissions
+func copyFile(src, dst string) error {
+	sourceFile, err := os.Open(src)
+	if err != nil {
+		return err
+	}
+	defer sourceFile.Close()
+
+	destFile, err := os.Create(dst)
+	if err != nil {
+		return err
+	}
+	defer destFile.Close()
+
+	// Copy the file contents
+	if _, err := io.Copy(destFile, sourceFile); err != nil {
+		return err
+	}
+
+	// Preserve the original file permissions
+	sourceInfo, err := sourceFile.Stat()
+	if err != nil {
+		return err
+	}
+
+	return destFile.Chmod(sourceInfo.Mode())
+}
+
+func archiveSuffix(binaryName, goos, goarch string) string {
+	if goos == "windows" {
+		return fmt.Sprintf("%s_%s_%s.zip", binaryName, goos, goarch)
+	}
+	return fmt.Sprintf("%s_%s_%s.tar.gz", binaryName, goos, goarch)
+}

beszel/internal/ghupdate/ghupdate_test.go

@@ -0,0 +1,45 @@
+package ghupdate
+
+import (
+	"path/filepath"
+	"testing"
+)
+
+func TestReleaseFindAssetBySuffix(t *testing.T) {
+	r := release{
+		Assets: []*releaseAsset{
+			{Name: "test1.zip", Id: 1},
+			{Name: "test2.zip", Id: 2},
+			{Name: "test22.zip", Id: 22},
+			{Name: "test3.zip", Id: 3},
+		},
+	}
+
+	asset, err := r.findAssetBySuffix("2.zip")
+	if err != nil {
+		t.Fatalf("Expected nil, got err: %v", err)
+	}
+
+	if asset.Id != 2 {
+		t.Fatalf("Expected asset with id %d, got %v", 2, asset)
+	}
+}
+
+func TestExtractFailure(t *testing.T) {
+	testDir := t.TempDir()
+
+	// Test with missing zip file
+	missingZipPath := filepath.Join(testDir, "missing_test.zip")
+	extractedPath := filepath.Join(testDir, "zip_extract")
+
+	if err := extract(missingZipPath, extractedPath); err == nil {
+		t.Fatal("Expected Extract to fail due to missing zip file")
+	}
+
+	// Test with missing tar.gz file
+	missingTarPath := filepath.Join(testDir, "missing_test.tar.gz")
+
+	if err := extract(missingTarPath, extractedPath); err == nil {
+		t.Fatal("Expected Extract to fail due to missing tar.gz file")
+	}
+}

beszel/internal/ghupdate/release.go

@@ -0,0 +1,36 @@
+package ghupdate
+
+import (
+	"errors"
+	"strings"
+)
+
+type releaseAsset struct {
+	Name        string `json:"name"`
+	DownloadUrl string `json:"browser_download_url"`
+	Id          int    `json:"id"`
+	Size        int    `json:"size"`
+}
+
+type release struct {
+	Name      string          `json:"name"`
+	Tag       string          `json:"tag_name"`
+	Published string          `json:"published_at"`
+	Url       string          `json:"html_url"`
+	Body      string          `json:"body"`
+	Assets    []*releaseAsset `json:"assets"`
+	Id        int             `json:"id"`
+}
+
+// findAssetBySuffix returns the first available asset containing the specified suffix.
+func (r *release) findAssetBySuffix(suffix string) (*releaseAsset, error) {
+	if suffix != "" {
+		for _, asset := range r.Assets {
+			if strings.HasSuffix(asset.Name, suffix) {
+				return asset, nil
+			}
+		}
+	}
+
+	return nil, errors.New("missing asset containing " + suffix)
+}

beszel/internal/hub/agent_connect.go

@@ -0,0 +1,320 @@
+package hub
+
+import (
+	"beszel/internal/common"
+	"beszel/internal/hub/expirymap"
+	"beszel/internal/hub/ws"
+	"errors"
+	"net"
+	"net/http"
+	"strings"
+	"sync"
+	"time"
+
+	"github.com/blang/semver"
+	"github.com/lxzan/gws"
+	"github.com/pocketbase/dbx"
+	"github.com/pocketbase/pocketbase/core"
+)
+
+// agentConnectRequest holds information related to an agent's connection attempt.
+type agentConnectRequest struct {
+	hub         *Hub
+	req         *http.Request
+	res         http.ResponseWriter
+	token       string
+	agentSemVer semver.Version
+	// isUniversalToken is true if the token is a universal token.
+	isUniversalToken bool
+	// userId is the user ID associated with the universal token.
+	userId string
+}
+
+// universalTokenMap stores active universal tokens and their associated user IDs.
+var universalTokenMap tokenMap
+
+type tokenMap struct {
+	store *expirymap.ExpiryMap[string]
+	once  sync.Once
+}
+
+// getMap returns the expirymap, creating it if necessary.
+func (tm *tokenMap) GetMap() *expirymap.ExpiryMap[string] {
+	tm.once.Do(func() {
+		tm.store = expirymap.New[string](time.Hour)
+	})
+	return tm.store
+}
+
+// handleAgentConnect is the HTTP handler for an agent's connection request.
+func (h *Hub) handleAgentConnect(e *core.RequestEvent) error {
+	agentRequest := agentConnectRequest{req: e.Request, res: e.Response, hub: h}
+	_ = agentRequest.agentConnect()
+	return nil
+}
+
+// agentConnect validates agent credentials and upgrades the connection to a WebSocket.
+func (acr *agentConnectRequest) agentConnect() (err error) {
+	var agentVersion string
+
+	acr.token, agentVersion, err = acr.validateAgentHeaders(acr.req.Header)
+	if err != nil {
+		return acr.sendResponseError(acr.res, http.StatusBadRequest, "")
+	}
+
+	// Check if token is an active universal token
+	acr.userId, acr.isUniversalToken = universalTokenMap.GetMap().GetOk(acr.token)
+
+	// Find matching fingerprint records for this token
+	fpRecords := getFingerprintRecordsByToken(acr.token, acr.hub)
+	if len(fpRecords) == 0 && !acr.isUniversalToken {
+		// Invalid token - no records found and not a universal token
+		return acr.sendResponseError(acr.res, http.StatusUnauthorized, "Invalid token")
+	}
+
+	// Validate agent version
+	acr.agentSemVer, err = semver.Parse(agentVersion)
+	if err != nil {
+		return acr.sendResponseError(acr.res, http.StatusUnauthorized, "Invalid agent version")
+	}
+
+	// Upgrade connection to WebSocket
+	conn, err := ws.GetUpgrader().Upgrade(acr.res, acr.req)
+	if err != nil {
+		return acr.sendResponseError(acr.res, http.StatusInternalServerError, "WebSocket upgrade failed")
+	}
+
+	go acr.verifyWsConn(conn, fpRecords)
+
+	return nil
+}
+
+// verifyWsConn verifies the WebSocket connection using the agent's fingerprint and
+// SSH key signature, then adds the system to the system manager.
+func (acr *agentConnectRequest) verifyWsConn(conn *gws.Conn, fpRecords []ws.FingerprintRecord) (err error) {
+	wsConn := ws.NewWsConnection(conn)
+
+	// must set wsConn in connection store before the read loop
+	conn.Session().Store("wsConn", wsConn)
+
+	// make sure connection is closed if there is an error
+	defer func() {
+		if err != nil {
+			wsConn.Close([]byte(err.Error()))
+		}
+	}()
+
+	go conn.ReadLoop()
+
+	signer, err := acr.hub.GetSSHKey("")
+	if err != nil {
+		return err
+	}
+
+	agentFingerprint, err := wsConn.GetFingerprint(acr.token, signer, acr.isUniversalToken)
+	if err != nil {
+		return err
+	}
+
+	// Find or create the appropriate system for this token and fingerprint
+	fpRecord, err := acr.findOrCreateSystemForToken(fpRecords, agentFingerprint)
+	if err != nil {
+		return err
+	}
+
+	return acr.hub.sm.AddWebSocketSystem(fpRecord.SystemId, acr.agentSemVer, wsConn)
+}
+
+// validateAgentHeaders extracts and validates the token and agent version from HTTP headers.
+func (acr *agentConnectRequest) validateAgentHeaders(headers http.Header) (string, string, error) {
+	token := headers.Get("X-Token")
+	agentVersion := headers.Get("X-Beszel")
+
+	if agentVersion == "" || token == "" || len(token) > 64 {
+		return "", "", errors.New("")
+	}
+	return token, agentVersion, nil
+}
+
+// sendResponseError writes an HTTP error response.
+func (acr *agentConnectRequest) sendResponseError(res http.ResponseWriter, code int, message string) error {
+	res.WriteHeader(code)
+	if message != "" {
+		res.Write([]byte(message))
+	}
+	return nil
+}
+
+// getFingerprintRecordsByToken retrieves all fingerprint records associated with a given token.
+func getFingerprintRecordsByToken(token string, h *Hub) []ws.FingerprintRecord {
+	var records []ws.FingerprintRecord
+	// All will populate empty slice even on error
+	_ = h.DB().NewQuery("SELECT id, system, fingerprint, token FROM fingerprints WHERE token = {:token}").
+		Bind(dbx.Params{
+			"token": token,
+		}).
+		All(&records)
+	return records
+}
+
+// findOrCreateSystemForToken finds an existing system matching the token and fingerprint,
+// or creates a new one for a universal token.
+func (acr *agentConnectRequest) findOrCreateSystemForToken(fpRecords []ws.FingerprintRecord, agentFingerprint common.FingerprintResponse) (ws.FingerprintRecord, error) {
+	// No records - only valid for active universal tokens
+	if len(fpRecords) == 0 {
+		return acr.handleNoRecords(agentFingerprint)
+	}
+
+	// Single record - handle as regular token
+	if len(fpRecords) == 1 && !acr.isUniversalToken {
+		return acr.handleSingleRecord(fpRecords[0], agentFingerprint)
+	}
+
+	// Multiple records or universal token - look for matching fingerprint
+	return acr.handleMultipleRecordsOrUniversalToken(fpRecords, agentFingerprint)
+}
+
+// handleNoRecords handles the case where no fingerprint records are found for a token.
+// A new system is created if the token is a valid universal token.
+func (acr *agentConnectRequest) handleNoRecords(agentFingerprint common.FingerprintResponse) (ws.FingerprintRecord, error) {
+	var fpRecord ws.FingerprintRecord
+
+	if !acr.isUniversalToken || acr.userId == "" {
+		return fpRecord, errors.New("no matching fingerprints")
+	}
+
+	return acr.createNewSystemForUniversalToken(agentFingerprint)
+}
+
+// handleSingleRecord handles the case with a single fingerprint record. It validates
+// the agent's fingerprint against the stored one, or sets it on first connect.
+func (acr *agentConnectRequest) handleSingleRecord(fpRecord ws.FingerprintRecord, agentFingerprint common.FingerprintResponse) (ws.FingerprintRecord, error) {
+	// If no current fingerprint, update with new fingerprint (first time connecting)
+	if fpRecord.Fingerprint == "" {
+		if err := acr.hub.SetFingerprint(&fpRecord, agentFingerprint.Fingerprint); err != nil {
+			return fpRecord, err
+		}
+		// Update the record with the fingerprint that was set
+		fpRecord.Fingerprint = agentFingerprint.Fingerprint
+		return fpRecord, nil
+	}
+
+	// Abort if fingerprint exists but doesn't match (different machine)
+	if fpRecord.Fingerprint != agentFingerprint.Fingerprint {
+		return fpRecord, errors.New("fingerprint mismatch")
+	}
+
+	return fpRecord, nil
+}
+
+// handleMultipleRecordsOrUniversalToken finds a matching fingerprint from multiple records.
+// If no match is found and the token is a universal token, a new system is created.
+func (acr *agentConnectRequest) handleMultipleRecordsOrUniversalToken(fpRecords []ws.FingerprintRecord, agentFingerprint common.FingerprintResponse) (ws.FingerprintRecord, error) {
+	// Return existing record with matching fingerprint if found
+	for i := range fpRecords {
+		if fpRecords[i].Fingerprint == agentFingerprint.Fingerprint {
+			return fpRecords[i], nil
+		}
+	}
+
+	// No matching fingerprint record found, but it's
+	// an active universal token so create a new system
+	if acr.isUniversalToken {
+		return acr.createNewSystemForUniversalToken(agentFingerprint)
+	}
+
+	return ws.FingerprintRecord{}, errors.New("fingerprint mismatch")
+}
+
+// createNewSystemForUniversalToken creates a new system and fingerprint record for a universal token.
+func (acr *agentConnectRequest) createNewSystemForUniversalToken(agentFingerprint common.FingerprintResponse) (ws.FingerprintRecord, error) {
+	var fpRecord ws.FingerprintRecord
+	if !acr.isUniversalToken || acr.userId == "" {
+		return fpRecord, errors.New("invalid token")
+	}
+
+	fpRecord.Token = acr.token
+
+	systemId, err := acr.createSystem(agentFingerprint)
+	if err != nil {
+		return fpRecord, err
+	}
+	fpRecord.SystemId = systemId
+
+	// Set the fingerprint for the new system
+	if err := acr.hub.SetFingerprint(&fpRecord, agentFingerprint.Fingerprint); err != nil {
+		return fpRecord, err
+	}
+
+	// Update the record with the fingerprint that was set
+	fpRecord.Fingerprint = agentFingerprint.Fingerprint
+
+	return fpRecord, nil
+}
+
+// createSystem creates a new system record in the database using details from the agent.
+func (acr *agentConnectRequest) createSystem(agentFingerprint common.FingerprintResponse) (recordId string, err error) {
+	systemsCollection, err := acr.hub.FindCachedCollectionByNameOrId("systems")
+	if err != nil {
+		return "", err
+	}
+	remoteAddr := getRealIP(acr.req)
+	// separate port from address
+	if agentFingerprint.Hostname == "" {
+		agentFingerprint.Hostname = remoteAddr
+	}
+	if agentFingerprint.Port == "" {
+		agentFingerprint.Port = "45876"
+	}
+	// create new record
+	systemRecord := core.NewRecord(systemsCollection)
+	systemRecord.Set("name", agentFingerprint.Hostname)
+	systemRecord.Set("host", remoteAddr)
+	systemRecord.Set("port", agentFingerprint.Port)
+	systemRecord.Set("users", []string{acr.userId})
+
+	return systemRecord.Id, acr.hub.Save(systemRecord)
+}
+
+// SetFingerprint creates or updates a fingerprint record in the database.
+func (h *Hub) SetFingerprint(fpRecord *ws.FingerprintRecord, fingerprint string) (err error) {
+	// // can't use raw query here because it doesn't trigger SSE
+	var record *core.Record
+	switch fpRecord.Id {
+	case "":
+		// create new record for universal token
+		collection, _ := h.FindCachedCollectionByNameOrId("fingerprints")
+		record = core.NewRecord(collection)
+		record.Set("system", fpRecord.SystemId)
+	default:
+		record, err = h.FindRecordById("fingerprints", fpRecord.Id)
+	}
+	if err != nil {
+		return err
+	}
+	record.Set("token", fpRecord.Token)
+	record.Set("fingerprint", fingerprint)
+	return h.SaveNoValidate(record)
+}
+
+// getRealIP extracts the client's real IP address from request headers,
+// checking common proxy headers before falling back to the remote address.
+func getRealIP(r *http.Request) string {
+	if ip := r.Header.Get("CF-Connecting-IP"); ip != "" {
+		return ip
+	}
+	if ip := r.Header.Get("X-Forwarded-For"); ip != "" {
+		// X-Forwarded-For can contain a comma-separated list: "client_ip, proxy1, proxy2"
+		// Take the first one
+		ips := strings.Split(ip, ",")
+		if len(ips) > 0 {
+			return strings.TrimSpace(ips[0])
+		}
+	}
+	// Fallback to RemoteAddr
+	ip, _, err := net.SplitHostPort(r.RemoteAddr)
+	if err != nil {
+		return r.RemoteAddr
+	}
+	return ip
+}

beszel/internal/hub/config/config.go

@@ -1,4 +1,5 @@
-package hub
+// Package config provides functions for syncing systems with the config.yml file
+package config
 
 import (
 	"beszel/internal/entities/system"
@@ -7,26 +8,27 @@ import (
 	"os"
 	"path/filepath"
 
+	"github.com/google/uuid"
 	"github.com/pocketbase/dbx"
-	"github.com/pocketbase/pocketbase/apis"
 	"github.com/pocketbase/pocketbase/core"
 	"github.com/spf13/cast"
 	"gopkg.in/yaml.v3"
 )
 
-type Config struct {
+type config struct {
-	Systems []SystemConfig `yaml:"systems"`
+	Systems []systemConfig `yaml:"systems"`
 }
 
-type SystemConfig struct {
+type systemConfig struct {
 	Name  string   `yaml:"name"`
 	Host  string   `yaml:"host"`
 	Port  uint16   `yaml:"port,omitempty"`
+	Token string   `yaml:"token,omitempty"`
 	Users []string `yaml:"users"`
 }
 
 // Syncs systems with the config.yml file
-func syncSystemsWithConfig(e *core.ServeEvent) error {
+func SyncSystems(e *core.ServeEvent) error {
 	h := e.App
 	configPath := filepath.Join(h.DataDir(), "config.yml")
 	configData, err := os.ReadFile(configPath)
@@ -34,7 +36,7 @@ func syncSystemsWithConfig(e *core.ServeEvent) error {
 		return nil
 	}
 
-	var config Config
+	var config config
 	err = yaml.Unmarshal(configData, &config)
 	if err != nil {
 		return fmt.Errorf("failed to parse config.yml: %v", err)
@@ -107,6 +109,14 @@ func syncSystemsWithConfig(e *core.ServeEvent) error {
 			if err := h.Save(existingSystem); err != nil {
 				return err
 			}
+
+			// Only update token if one is specified in config, otherwise preserve existing token
+			if sysConfig.Token != "" {
+				if err := updateFingerprintToken(h, existingSystem.Id, sysConfig.Token); err != nil {
+					return err
+				}
+			}
+
 			delete(existingSystemsMap, key)
 		} else {
 			// Create new system
@@ -124,10 +134,21 @@ func syncSystemsWithConfig(e *core.ServeEvent) error {
 			if err := h.Save(newSystem); err != nil {
 				return fmt.Errorf("failed to create new system: %v", err)
 			}
+
+			// For new systems, generate token if not provided
+			token := sysConfig.Token
+			if token == "" {
+				token = uuid.New().String()
+			}
+
+			// Create fingerprint record for new system
+			if err := createFingerprintRecord(h, newSystem.Id, token); err != nil {
+				return err
+			}
 		}
 	}
 
-	// Delete systems not in config
+	// Delete systems not in config (and their fingerprint records will cascade delete)
 	for _, system := range existingSystemsMap {
 		if err := h.Delete(system); err != nil {
 			return err
@@ -139,7 +160,7 @@ func syncSystemsWithConfig(e *core.ServeEvent) error {
 }
 
 // Generates content for the config.yml file as a YAML string
-func (h *Hub) generateConfigYAML() (string, error) {
+func generateYAML(h core.App) (string, error) {
 	// Fetch all systems from the database
 	systems, err := h.FindRecordsByFilter("systems", "id != ''", "name", -1, 0)
 	if err != nil {
@@ -147,8 +168,8 @@ func (h *Hub) generateConfigYAML() (string, error) {
 	}
 
 	// Create a Config struct to hold the data
-	config := Config{
+	config := config{
-		Systems: make([]SystemConfig, 0, len(systems)),
+		Systems: make([]systemConfig, 0, len(systems)),
 	}
 
 	// Fetch all users at once
@@ -156,11 +177,29 @@ func (h *Hub) generateConfigYAML() (string, error) {
 	for _, system := range systems {
 		allUserIDs = append(allUserIDs, system.GetStringSlice("users")...)
 	}
-	userEmailMap, err := h.getUserEmailMap(allUserIDs)
+	userEmailMap, err := getUserEmailMap(h, allUserIDs)
 	if err != nil {
 		return "", err
 	}
 
+	// Fetch all fingerprint records to get tokens
+	type fingerprintData struct {
+		ID     string `db:"id"`
+		System string `db:"system"`
+		Token  string `db:"token"`
+	}
+	var fingerprints []fingerprintData
+	err = h.DB().NewQuery("SELECT id, system, token FROM fingerprints").All(&fingerprints)
+	if err != nil {
+		return "", err
+	}
+
+	// Create a map of system ID to token
+	systemTokenMap := make(map[string]string)
+	for _, fingerprint := range fingerprints {
+		systemTokenMap[fingerprint.System] = fingerprint.Token
+	}
+
 	// Populate the Config struct with system data
 	for _, system := range systems {
 		userIDs := system.GetStringSlice("users")
@@ -171,11 +210,12 @@ func (h *Hub) generateConfigYAML() (string, error) {
 			}
 		}
 
-		sysConfig := SystemConfig{
+		sysConfig := systemConfig{
 			Name:  system.GetString("name"),
 			Host:  system.GetString("host"),
 			Port:  cast.ToUint16(system.Get("port")),
 			Users: userEmails,
+			Token: systemTokenMap[system.Id],
 		}
 		config.Systems = append(config.Systems, sysConfig)
 	}
@@ -187,13 +227,13 @@ func (h *Hub) generateConfigYAML() (string, error) {
 	}
 
 	// Add a header to the YAML
-	yamlData = append([]byte("# Values for port and users are optional.\n# Defaults are port 45876 and the first created user.\n\n"), yamlData...)
+	yamlData = append([]byte("# Values for port, users, and token are optional.\n# Defaults are port 45876, the first created user, and a generated UUID token.\n\n"), yamlData...)
 
 	return string(yamlData), nil
 }
 
 // New helper function to get a map of user IDs to emails
-func (h *Hub) getUserEmailMap(userIDs []string) (map[string]string, error) {
+func getUserEmailMap(h core.App, userIDs []string) (map[string]string, error) {
 	users, err := h.FindRecordsByIds("users", userIDs)
 	if err != nil {
 		return nil, err
@@ -207,13 +247,41 @@ func (h *Hub) getUserEmailMap(userIDs []string) (map[string]string, error) {
 	return userEmailMap, nil
 }
 
-// Returns the current config.yml file as a JSON object
+// Helper function to update or create fingerprint token for an existing system
-func (h *Hub) getYamlConfig(e *core.RequestEvent) error {
+func updateFingerprintToken(app core.App, systemID, token string) error {
-	info, _ := e.RequestInfo()
+	// Try to find existing fingerprint record
-	if info.Auth == nil || info.Auth.GetString("role") != "admin" {
+	fingerprint, err := app.FindFirstRecordByFilter("fingerprints", "system = {:system}", dbx.Params{"system": systemID})
-		return apis.NewForbiddenError("Forbidden", nil)
+	if err != nil {
+		// If no fingerprint record exists, create one
+		return createFingerprintRecord(app, systemID, token)
 	}
-	configContent, err := h.generateConfigYAML()
+
+	// Update existing fingerprint record with new token (keep existing fingerprint)
+	fingerprint.Set("token", token)
+	return app.Save(fingerprint)
+}
+
+// Helper function to create a new fingerprint record for a system
+func createFingerprintRecord(app core.App, systemID, token string) error {
+	fingerprintsCollection, err := app.FindCollectionByNameOrId("fingerprints")
+	if err != nil {
+		return fmt.Errorf("failed to find fingerprints collection: %v", err)
+	}
+
+	newFingerprint := core.NewRecord(fingerprintsCollection)
+	newFingerprint.Set("system", systemID)
+	newFingerprint.Set("token", token)
+	newFingerprint.Set("fingerprint", "") // Empty fingerprint, will be set on first connection
+
+	return app.Save(newFingerprint)
+}
+
+// Returns the current config.yml file as a JSON object
+func GetYamlConfig(e *core.RequestEvent) error {
+	if e.Auth.GetString("role") != "admin" {
+		return e.ForbiddenError("Requires admin role", nil)
+	}
+	configContent, err := generateYAML(e.App)
 	if err != nil {
 		return err
 	}

beszel/internal/hub/config/config_test.go

@@ -0,0 +1,245 @@
+//go:build testing
+// +build testing
+
+package config_test
+
+import (
+	"beszel/internal/hub/config"
+	"beszel/internal/tests"
+	"os"
+	"path/filepath"
+	"testing"
+
+	"github.com/pocketbase/pocketbase/core"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+	"gopkg.in/yaml.v3"
+)
+
+// Config struct for testing (copied from config package since it's not exported)
+type testConfig struct {
+	Systems []testSystemConfig `yaml:"systems"`
+}
+
+type testSystemConfig struct {
+	Name  string   `yaml:"name"`
+	Host  string   `yaml:"host"`
+	Port  uint16   `yaml:"port,omitempty"`
+	Users []string `yaml:"users"`
+	Token string   `yaml:"token,omitempty"`
+}
+
+// Helper function to create a test system for config tests
+// func createConfigTestSystem(app core.App, name, host string, port uint16, userIDs []string) (*core.Record, error) {
+// 	systemCollection, err := app.FindCollectionByNameOrId("systems")
+// 	if err != nil {
+// 		return nil, err
+// 	}
+
+// 	system := core.NewRecord(systemCollection)
+// 	system.Set("name", name)
+// 	system.Set("host", host)
+// 	system.Set("port", port)
+// 	system.Set("users", userIDs)
+// 	system.Set("status", "pending")
+
+// 	return system, app.Save(system)
+// }
+
+// Helper function to create a fingerprint record
+func createConfigTestFingerprint(app core.App, systemID, token, fingerprint string) (*core.Record, error) {
+	fingerprintCollection, err := app.FindCollectionByNameOrId("fingerprints")
+	if err != nil {
+		return nil, err
+	}
+
+	fp := core.NewRecord(fingerprintCollection)
+	fp.Set("system", systemID)
+	fp.Set("token", token)
+	fp.Set("fingerprint", fingerprint)
+
+	return fp, app.Save(fp)
+}
+
+// TestConfigSyncWithTokens tests the config.SyncSystems function with various token scenarios
+func TestConfigSyncWithTokens(t *testing.T) {
+	testHub, err := tests.NewTestHub()
+	require.NoError(t, err)
+	defer testHub.Cleanup()
+
+	// Create test user
+	user, err := tests.CreateUser(testHub.App, "admin@example.com", "testtesttest")
+	require.NoError(t, err)
+
+	testCases := []struct {
+		name        string
+		setupFunc   func() (string, *core.Record, *core.Record) // Returns: existing token, system record, fingerprint record
+		configYAML  string
+		expectToken string // Expected token after sync
+		description string
+	}{
+		{
+			name: "new system with token in config",
+			setupFunc: func() (string, *core.Record, *core.Record) {
+				return "", nil, nil // No existing system
+			},
+			configYAML: `systems:
+  - name: "new-server"
+    host: "new.example.com"
+    port: 45876
+    users:
+      - "admin@example.com"
+    token: "explicit-token-123"`,
+			expectToken: "explicit-token-123",
+			description: "New system should use token from config",
+		},
+		{
+			name: "existing system without token in config (preserve existing)",
+			setupFunc: func() (string, *core.Record, *core.Record) {
+				// Create existing system and fingerprint
+				system, err := tests.CreateRecord(testHub.App, "systems", map[string]any{
+					"name":  "preserve-server",
+					"host":  "preserve.example.com",
+					"port":  45876,
+					"users": []string{user.Id},
+				})
+				require.NoError(t, err)
+
+				fingerprint, err := createConfigTestFingerprint(testHub.App, system.Id, "preserve-token-999", "preserve-fingerprint")
+				require.NoError(t, err)
+
+				return "preserve-token-999", system, fingerprint
+			},
+			configYAML: `systems:
+  - name: "preserve-server"
+    host: "preserve.example.com"
+    port: 45876
+    users:
+      - "admin@example.com"`,
+			expectToken: "preserve-token-999",
+			description: "Existing system should preserve original token when config doesn't specify one",
+		},
+	}
+
+	for _, tc := range testCases {
+		t.Run(tc.name, func(t *testing.T) {
+			// Setup test data
+			_, existingSystem, existingFingerprint := tc.setupFunc()
+
+			// Write config file
+			configPath := filepath.Join(testHub.DataDir(), "config.yml")
+			err := os.WriteFile(configPath, []byte(tc.configYAML), 0644)
+			require.NoError(t, err)
+
+			// Create serve event and sync
+			event := &core.ServeEvent{App: testHub.App}
+			err = config.SyncSystems(event)
+			require.NoError(t, err)
+
+			// Parse the config to get the system name for verification
+			var configData testConfig
+			err = yaml.Unmarshal([]byte(tc.configYAML), &configData)
+			require.NoError(t, err)
+			require.Len(t, configData.Systems, 1)
+			systemName := configData.Systems[0].Name
+
+			// Find the system after sync
+			systems, err := testHub.FindRecordsByFilter("systems", "name = {:name}", "", -1, 0, map[string]any{"name": systemName})
+			require.NoError(t, err)
+			require.Len(t, systems, 1)
+			system := systems[0]
+
+			// Find the fingerprint record
+			fingerprints, err := testHub.FindRecordsByFilter("fingerprints", "system = {:system}", "", -1, 0, map[string]any{"system": system.Id})
+			require.NoError(t, err)
+			require.Len(t, fingerprints, 1)
+			fingerprint := fingerprints[0]
+
+			// Verify token
+			actualToken := fingerprint.GetString("token")
+			if tc.expectToken == "" {
+				// For generated tokens, just verify it's not empty and is a valid UUID format
+				assert.NotEmpty(t, actualToken, tc.description)
+				assert.Len(t, actualToken, 36, "Generated token should be UUID format") // UUID length
+			} else {
+				assert.Equal(t, tc.expectToken, actualToken, tc.description)
+			}
+
+			// For existing systems, verify fingerprint is preserved
+			if existingFingerprint != nil {
+				actualFingerprint := fingerprint.GetString("fingerprint")
+				expectedFingerprint := existingFingerprint.GetString("fingerprint")
+				assert.Equal(t, expectedFingerprint, actualFingerprint, "Fingerprint should be preserved")
+			}
+
+			// Cleanup for next test
+			if existingSystem != nil {
+				testHub.Delete(existingSystem)
+			}
+			if existingFingerprint != nil {
+				testHub.Delete(existingFingerprint)
+			}
+			// Clean up the new records
+			testHub.Delete(system)
+			testHub.Delete(fingerprint)
+		})
+	}
+}
+
+// TestConfigMigrationScenario tests the specific migration scenario mentioned in the discussion
+func TestConfigMigrationScenario(t *testing.T) {
+	testHub, err := tests.NewTestHub(t.TempDir())
+	require.NoError(t, err)
+	defer testHub.Cleanup()
+
+	// Create test user
+	user, err := tests.CreateUser(testHub.App, "admin@example.com", "testtesttest")
+	require.NoError(t, err)
+
+	// Simulate migration scenario: system exists with token from migration
+	existingSystem, err := tests.CreateRecord(testHub.App, "systems", map[string]any{
+		"name":  "migrated-server",
+		"host":  "migrated.example.com",
+		"port":  45876,
+		"users": []string{user.Id},
+	})
+	require.NoError(t, err)
+
+	migrationToken := "migration-generated-token-123"
+	existingFingerprint, err := createConfigTestFingerprint(testHub.App, existingSystem.Id, migrationToken, "existing-fingerprint-from-agent")
+	require.NoError(t, err)
+
+	// User exports config BEFORE this update (so no token field in YAML)
+	oldConfigYAML := `systems:
+  - name: "migrated-server"
+    host: "migrated.example.com"
+    port: 45876
+    users:
+      - "admin@example.com"`
+
+	// Write old config file and import
+	configPath := filepath.Join(testHub.DataDir(), "config.yml")
+	err = os.WriteFile(configPath, []byte(oldConfigYAML), 0644)
+	require.NoError(t, err)
+
+	event := &core.ServeEvent{App: testHub.App}
+	err = config.SyncSystems(event)
+	require.NoError(t, err)
+
+	// Verify the original token is preserved
+	updatedFingerprint, err := testHub.FindRecordById("fingerprints", existingFingerprint.Id)
+	require.NoError(t, err)
+
+	actualToken := updatedFingerprint.GetString("token")
+	assert.Equal(t, migrationToken, actualToken, "Migration token should be preserved when config doesn't specify a token")
+
+	// Verify fingerprint is also preserved
+	actualFingerprint := updatedFingerprint.GetString("fingerprint")
+	assert.Equal(t, "existing-fingerprint-from-agent", actualFingerprint, "Existing fingerprint should be preserved")
+
+	// Verify system still exists and is updated correctly
+	updatedSystem, err := testHub.FindRecordById("systems", existingSystem.Id)
+	require.NoError(t, err)
+	assert.Equal(t, "migrated-server", updatedSystem.GetString("name"))
+	assert.Equal(t, "migrated.example.com", updatedSystem.GetString("host"))
+}

beszel/internal/hub/expirymap/expirymap.go

@@ -0,0 +1,104 @@
+package expirymap
+
+import (
+	"reflect"
+	"time"
+
+	"github.com/pocketbase/pocketbase/tools/store"
+)
+
+type val[T any] struct {
+	value   T
+	expires time.Time
+}
+
+type ExpiryMap[T any] struct {
+	store           *store.Store[string, *val[T]]
+	cleanupInterval time.Duration
+}
+
+// New creates a new expiry map with custom cleanup interval
+func New[T any](cleanupInterval time.Duration) *ExpiryMap[T] {
+	m := &ExpiryMap[T]{
+		store:           store.New(map[string]*val[T]{}),
+		cleanupInterval: cleanupInterval,
+	}
+	m.startCleaner()
+	return m
+}
+
+// Set stores a value with the given TTL
+func (m *ExpiryMap[T]) Set(key string, value T, ttl time.Duration) {
+	m.store.Set(key, &val[T]{
+		value:   value,
+		expires: time.Now().Add(ttl),
+	})
+}
+
+// GetOk retrieves a value and checks if it exists and hasn't expired
+// Performs lazy cleanup of expired entries on access
+func (m *ExpiryMap[T]) GetOk(key string) (T, bool) {
+	value, ok := m.store.GetOk(key)
+	if !ok {
+		return *new(T), false
+	}
+
+	// Check if expired and perform lazy cleanup
+	if value.expires.Before(time.Now()) {
+		m.store.Remove(key)
+		return *new(T), false
+	}
+
+	return value.value, true
+}
+
+// GetByValue retrieves a value by value
+func (m *ExpiryMap[T]) GetByValue(val T) (key string, value T, ok bool) {
+	for key, v := range m.store.GetAll() {
+		if reflect.DeepEqual(v.value, val) {
+			// check if expired
+			if v.expires.Before(time.Now()) {
+				m.store.Remove(key)
+				break
+			}
+			return key, v.value, true
+		}
+	}
+	return "", *new(T), false
+}
+
+// Remove explicitly removes a key
+func (m *ExpiryMap[T]) Remove(key string) {
+	m.store.Remove(key)
+}
+
+// RemovebyValue removes a value by value
+func (m *ExpiryMap[T]) RemovebyValue(value T) (T, bool) {
+	for key, val := range m.store.GetAll() {
+		if reflect.DeepEqual(val.value, value) {
+			m.store.Remove(key)
+			return val.value, true
+		}
+	}
+	return *new(T), false
+}
+
+// startCleaner runs the background cleanup process
+func (m *ExpiryMap[T]) startCleaner() {
+	go func() {
+		tick := time.Tick(m.cleanupInterval)
+		for range tick {
+			m.cleanup()
+		}
+	}()
+}
+
+// cleanup removes all expired entries
+func (m *ExpiryMap[T]) cleanup() {
+	now := time.Now()
+	for key, val := range m.store.GetAll() {
+		if val.expires.Before(now) {
+			m.store.Remove(key)
+		}
+	}
+}

beszel/internal/hub/expirymap/expirymap_test.go

@@ -0,0 +1,477 @@
+//go:build testing
+// +build testing
+
+package expirymap
+
+import (
+	"testing"
+	"time"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+// Not using the following methods but are useful for testing
+
+// TESTING: Has checks if a key exists and hasn't expired
+func (m *ExpiryMap[T]) Has(key string) bool {
+	_, ok := m.GetOk(key)
+	return ok
+}
+
+// TESTING: Get retrieves a value, returns zero value if not found or expired
+func (m *ExpiryMap[T]) Get(key string) T {
+	value, _ := m.GetOk(key)
+	return value
+}
+
+// TESTING: Len returns the number of non-expired entries
+func (m *ExpiryMap[T]) Len() int {
+	count := 0
+	now := time.Now()
+	for _, val := range m.store.Values() {
+		if val.expires.After(now) {
+			count++
+		}
+	}
+	return count
+}
+
+func TestExpiryMap_BasicOperations(t *testing.T) {
+	em := New[string](time.Hour)
+
+	// Test Set and GetOk
+	em.Set("key1", "value1", time.Hour)
+	value, ok := em.GetOk("key1")
+	assert.True(t, ok)
+	assert.Equal(t, "value1", value)
+
+	// Test Get
+	value = em.Get("key1")
+	assert.Equal(t, "value1", value)
+
+	// Test Has
+	assert.True(t, em.Has("key1"))
+	assert.False(t, em.Has("nonexistent"))
+
+	// Test Remove
+	em.Remove("key1")
+	assert.False(t, em.Has("key1"))
+}
+
+func TestExpiryMap_Expiration(t *testing.T) {
+	em := New[string](time.Hour)
+
+	// Set a value with very short TTL
+	em.Set("shortlived", "value", time.Millisecond*10)
+
+	// Should exist immediately
+	assert.True(t, em.Has("shortlived"))
+
+	// Wait for expiration
+	time.Sleep(time.Millisecond * 20)
+
+	// Should be expired and automatically cleaned up on access
+	assert.False(t, em.Has("shortlived"))
+	value, ok := em.GetOk("shortlived")
+	assert.False(t, ok)
+	assert.Equal(t, "", value) // zero value for string
+}
+
+func TestExpiryMap_LazyCleanup(t *testing.T) {
+	em := New[int](time.Hour)
+
+	// Set multiple values with short TTL
+	em.Set("key1", 1, time.Millisecond*10)
+	em.Set("key2", 2, time.Millisecond*10)
+	em.Set("key3", 3, time.Hour) // This one won't expire
+
+	// Wait for expiration
+	time.Sleep(time.Millisecond * 20)
+
+	// Access expired keys should trigger lazy cleanup
+	_, ok := em.GetOk("key1")
+	assert.False(t, ok)
+
+	// Non-expired key should still exist
+	value, ok := em.GetOk("key3")
+	assert.True(t, ok)
+	assert.Equal(t, 3, value)
+}
+
+func TestExpiryMap_Len(t *testing.T) {
+	em := New[string](time.Hour)
+
+	// Initially empty
+	assert.Equal(t, 0, em.Len())
+
+	// Add some values
+	em.Set("key1", "value1", time.Hour)
+	em.Set("key2", "value2", time.Hour)
+	em.Set("key3", "value3", time.Millisecond*10) // Will expire soon
+
+	// Should count all initially
+	assert.Equal(t, 3, em.Len())
+
+	// Wait for one to expire
+	time.Sleep(time.Millisecond * 20)
+
+	// Len should reflect only non-expired entries
+	assert.Equal(t, 2, em.Len())
+}
+
+func TestExpiryMap_CustomInterval(t *testing.T) {
+	// Create with very short cleanup interval for testing
+	em := New[string](time.Millisecond * 50)
+
+	// Set a value that expires quickly
+	em.Set("test", "value", time.Millisecond*10)
+
+	// Should exist initially
+	assert.True(t, em.Has("test"))
+
+	// Wait for expiration + cleanup cycle
+	time.Sleep(time.Millisecond * 100)
+
+	// Should be cleaned up by background process
+	// Note: This test might be flaky due to timing, but demonstrates the concept
+	assert.False(t, em.Has("test"))
+}
+
+func TestExpiryMap_GenericTypes(t *testing.T) {
+	// Test with different types
+	t.Run("Int", func(t *testing.T) {
+		em := New[int](time.Hour)
+
+		em.Set("num", 42, time.Hour)
+		value, ok := em.GetOk("num")
+		assert.True(t, ok)
+		assert.Equal(t, 42, value)
+	})
+
+	t.Run("Struct", func(t *testing.T) {
+		type TestStruct struct {
+			Name string
+			Age  int
+		}
+
+		em := New[TestStruct](time.Hour)
+
+		expected := TestStruct{Name: "John", Age: 30}
+		em.Set("person", expected, time.Hour)
+
+		value, ok := em.GetOk("person")
+		assert.True(t, ok)
+		assert.Equal(t, expected, value)
+	})
+
+	t.Run("Pointer", func(t *testing.T) {
+		em := New[*string](time.Hour)
+
+		str := "hello"
+		em.Set("ptr", &str, time.Hour)
+
+		value, ok := em.GetOk("ptr")
+		assert.True(t, ok)
+		require.NotNil(t, value)
+		assert.Equal(t, "hello", *value)
+	})
+}
+
+func TestExpiryMap_ZeroValues(t *testing.T) {
+	em := New[string](time.Hour)
+
+	// Test getting non-existent key returns zero value
+	value := em.Get("nonexistent")
+	assert.Equal(t, "", value)
+
+	// Test getting expired key returns zero value
+	em.Set("expired", "value", time.Millisecond*10)
+	time.Sleep(time.Millisecond * 20)
+
+	value = em.Get("expired")
+	assert.Equal(t, "", value)
+}
+
+func TestExpiryMap_Concurrent(t *testing.T) {
+	em := New[int](time.Hour)
+
+	// Simple concurrent access test
+	done := make(chan bool, 2)
+
+	// Writer goroutine
+	go func() {
+		for i := 0; i < 100; i++ {
+			em.Set("key", i, time.Hour)
+			time.Sleep(time.Microsecond)
+		}
+		done <- true
+	}()
+
+	// Reader goroutine
+	go func() {
+		for i := 0; i < 100; i++ {
+			_ = em.Get("key")
+			time.Sleep(time.Microsecond)
+		}
+		done <- true
+	}()
+
+	// Wait for both to complete
+	<-done
+	<-done
+
+	// Should not panic and should have some value
+	assert.True(t, em.Has("key"))
+}
+
+func TestExpiryMap_GetByValue(t *testing.T) {
+	em := New[string](time.Hour)
+
+	// Test getting by value when value exists
+	em.Set("key1", "value1", time.Hour)
+	em.Set("key2", "value2", time.Hour)
+	em.Set("key3", "value1", time.Hour) // Duplicate value - should return first match
+
+	// Test successful retrieval
+	key, value, ok := em.GetByValue("value1")
+	assert.True(t, ok)
+	assert.Equal(t, "value1", value)
+	assert.Contains(t, []string{"key1", "key3"}, key) // Should be one of the keys with this value
+
+	// Test retrieval of unique value
+	key, value, ok = em.GetByValue("value2")
+	assert.True(t, ok)
+	assert.Equal(t, "value2", value)
+	assert.Equal(t, "key2", key)
+
+	// Test getting non-existent value
+	key, value, ok = em.GetByValue("nonexistent")
+	assert.False(t, ok)
+	assert.Equal(t, "", value) // zero value for string
+	assert.Equal(t, "", key)   // zero value for string
+}
+
+func TestExpiryMap_GetByValue_Expiration(t *testing.T) {
+	em := New[string](time.Hour)
+
+	// Set a value with short TTL
+	em.Set("shortkey", "shortvalue", time.Millisecond*10)
+	em.Set("longkey", "longvalue", time.Hour)
+
+	// Should find the short-lived value initially
+	key, value, ok := em.GetByValue("shortvalue")
+	assert.True(t, ok)
+	assert.Equal(t, "shortvalue", value)
+	assert.Equal(t, "shortkey", key)
+
+	// Wait for expiration
+	time.Sleep(time.Millisecond * 20)
+
+	// Should not find expired value and should trigger lazy cleanup
+	key, value, ok = em.GetByValue("shortvalue")
+	assert.False(t, ok)
+	assert.Equal(t, "", value)
+	assert.Equal(t, "", key)
+
+	// Should still find non-expired value
+	key, value, ok = em.GetByValue("longvalue")
+	assert.True(t, ok)
+	assert.Equal(t, "longvalue", value)
+	assert.Equal(t, "longkey", key)
+}
+
+func TestExpiryMap_GetByValue_GenericTypes(t *testing.T) {
+	t.Run("Int", func(t *testing.T) {
+		em := New[int](time.Hour)
+
+		em.Set("num1", 42, time.Hour)
+		em.Set("num2", 84, time.Hour)
+
+		key, value, ok := em.GetByValue(42)
+		assert.True(t, ok)
+		assert.Equal(t, 42, value)
+		assert.Equal(t, "num1", key)
+
+		key, value, ok = em.GetByValue(99)
+		assert.False(t, ok)
+		assert.Equal(t, 0, value)
+		assert.Equal(t, "", key)
+	})
+
+	t.Run("Struct", func(t *testing.T) {
+		type TestStruct struct {
+			Name string
+			Age  int
+		}
+
+		em := New[TestStruct](time.Hour)
+
+		person1 := TestStruct{Name: "John", Age: 30}
+		person2 := TestStruct{Name: "Jane", Age: 25}
+
+		em.Set("person1", person1, time.Hour)
+		em.Set("person2", person2, time.Hour)
+
+		key, value, ok := em.GetByValue(person1)
+		assert.True(t, ok)
+		assert.Equal(t, person1, value)
+		assert.Equal(t, "person1", key)
+
+		nonexistent := TestStruct{Name: "Bob", Age: 40}
+		key, value, ok = em.GetByValue(nonexistent)
+		assert.False(t, ok)
+		assert.Equal(t, TestStruct{}, value)
+		assert.Equal(t, "", key)
+	})
+}
+
+func TestExpiryMap_RemoveValue(t *testing.T) {
+	em := New[string](time.Hour)
+
+	// Test removing existing value
+	em.Set("key1", "value1", time.Hour)
+	em.Set("key2", "value2", time.Hour)
+	em.Set("key3", "value1", time.Hour) // Duplicate value
+
+	// Remove by value should remove one instance
+	removedValue, ok := em.RemovebyValue("value1")
+	assert.True(t, ok)
+	assert.Equal(t, "value1", removedValue)
+
+	// Should still have the other instance or value2
+	assert.True(t, em.Has("key2")) // value2 should still exist
+
+	// Check if one of the duplicate values was removed
+	// At least one key with "value1" should be gone
+	key1Exists := em.Has("key1")
+	key3Exists := em.Has("key3")
+	assert.False(t, key1Exists && key3Exists) // Both shouldn't exist
+	assert.True(t, key1Exists || key3Exists)  // At least one should be gone
+
+	// Test removing non-existent value
+	removedValue, ok = em.RemovebyValue("nonexistent")
+	assert.False(t, ok)
+	assert.Equal(t, "", removedValue) // zero value for string
+}
+
+func TestExpiryMap_RemoveValue_GenericTypes(t *testing.T) {
+	t.Run("Int", func(t *testing.T) {
+		em := New[int](time.Hour)
+
+		em.Set("num1", 42, time.Hour)
+		em.Set("num2", 84, time.Hour)
+
+		// Remove existing value
+		removedValue, ok := em.RemovebyValue(42)
+		assert.True(t, ok)
+		assert.Equal(t, 42, removedValue)
+		assert.False(t, em.Has("num1"))
+		assert.True(t, em.Has("num2"))
+
+		// Remove non-existent value
+		removedValue, ok = em.RemovebyValue(99)
+		assert.False(t, ok)
+		assert.Equal(t, 0, removedValue)
+	})
+
+	t.Run("Struct", func(t *testing.T) {
+		type TestStruct struct {
+			Name string
+			Age  int
+		}
+
+		em := New[TestStruct](time.Hour)
+
+		person1 := TestStruct{Name: "John", Age: 30}
+		person2 := TestStruct{Name: "Jane", Age: 25}
+
+		em.Set("person1", person1, time.Hour)
+		em.Set("person2", person2, time.Hour)
+
+		// Remove existing struct
+		removedValue, ok := em.RemovebyValue(person1)
+		assert.True(t, ok)
+		assert.Equal(t, person1, removedValue)
+		assert.False(t, em.Has("person1"))
+		assert.True(t, em.Has("person2"))
+
+		// Remove non-existent struct
+		nonexistent := TestStruct{Name: "Bob", Age: 40}
+		removedValue, ok = em.RemovebyValue(nonexistent)
+		assert.False(t, ok)
+		assert.Equal(t, TestStruct{}, removedValue)
+	})
+}
+
+func TestExpiryMap_RemoveValue_WithExpiration(t *testing.T) {
+	em := New[string](time.Hour)
+
+	// Set values with different TTLs
+	em.Set("key1", "value1", time.Millisecond*10) // Will expire
+	em.Set("key2", "value2", time.Hour)           // Won't expire
+	em.Set("key3", "value1", time.Hour)           // Won't expire, duplicate value
+
+	// Wait for first value to expire
+	time.Sleep(time.Millisecond * 20)
+
+	// Try to remove the expired value - should remove one of the "value1" entries
+	removedValue, ok := em.RemovebyValue("value1")
+	assert.True(t, ok)
+	assert.Equal(t, "value1", removedValue)
+
+	// Should still have key2 (different value)
+	assert.True(t, em.Has("key2"))
+
+	// Should have removed one of the "value1" entries (either key1 or key3)
+	// But we can't predict which one due to map iteration order
+	key1Exists := em.Has("key1")
+	key3Exists := em.Has("key3")
+
+	// Exactly one of key1 or key3 should be gone
+	assert.False(t, key1Exists && key3Exists) // Both shouldn't exist
+	assert.True(t, key1Exists || key3Exists)  // At least one should still exist
+}
+
+func TestExpiryMap_ValueOperations_Integration(t *testing.T) {
+	em := New[string](time.Hour)
+
+	// Test integration of GetByValue and RemoveValue
+	em.Set("key1", "shared", time.Hour)
+	em.Set("key2", "unique", time.Hour)
+	em.Set("key3", "shared", time.Hour)
+
+	// Find shared value
+	key, value, ok := em.GetByValue("shared")
+	assert.True(t, ok)
+	assert.Equal(t, "shared", value)
+	assert.Contains(t, []string{"key1", "key3"}, key)
+
+	// Remove shared value
+	removedValue, ok := em.RemovebyValue("shared")
+	assert.True(t, ok)
+	assert.Equal(t, "shared", removedValue)
+
+	// Should still be able to find the other shared value
+	key, value, ok = em.GetByValue("shared")
+	assert.True(t, ok)
+	assert.Equal(t, "shared", value)
+	assert.Contains(t, []string{"key1", "key3"}, key)
+
+	// Remove the other shared value
+	removedValue, ok = em.RemovebyValue("shared")
+	assert.True(t, ok)
+	assert.Equal(t, "shared", removedValue)
+
+	// Should not find shared value anymore
+	key, value, ok = em.GetByValue("shared")
+	assert.False(t, ok)
+	assert.Equal(t, "", value)
+	assert.Equal(t, "", key)
+
+	// Unique value should still exist
+	key, value, ok = em.GetByValue("unique")
+	assert.True(t, ok)
+	assert.Equal(t, "unique", value)
+	assert.Equal(t, "key2", key)
+}

beszel/internal/hub/hub.go

@@ -4,19 +4,21 @@ package hub
 import (
 	"beszel"
 	"beszel/internal/alerts"
+	"beszel/internal/hub/config"
 	"beszel/internal/hub/systems"
 	"beszel/internal/records"
 	"beszel/internal/users"
-	"beszel/site"
 	"crypto/ed25519"
 	"encoding/pem"
-	"io/fs"
+	"fmt"
 	"net/http"
-	"net/http/httputil"
 	"net/url"
 	"os"
+	"path"
 	"strings"
+	"time"
 
+	"github.com/google/uuid"
 	"github.com/pocketbase/pocketbase"
 	"github.com/pocketbase/pocketbase/apis"
 	"github.com/pocketbase/pocketbase/core"
@@ -30,6 +32,7 @@ type Hub struct {
 	rm     *records.RecordManager
 	sm     *systems.SystemManager
 	pubKey string
+	signer ssh.Signer
 	appURL string
 }
 
@@ -56,14 +59,13 @@ func GetEnv(key string) (value string, exists bool) {
 }
 
 func (h *Hub) StartHub() error {
-
 	h.App.OnServe().BindFunc(func(e *core.ServeEvent) error {
 		// initialize settings / collections
 		if err := h.initialize(e); err != nil {
 			return err
 		}
 		// sync systems with config
-		if err := syncSystemsWithConfig(e); err != nil {
+		if err := config.SyncSystems(e); err != nil {
 			return err
 		}
 		// register api routes
@@ -110,6 +112,11 @@ func (h *Hub) initialize(e *core.ServeEvent) error {
 	// set URL if BASE_URL env is set
 	if h.appURL != "" {
 		settings.Meta.AppURL = h.appURL
+	} else {
+		h.appURL = settings.Meta.AppURL
+	}
+	if err := e.App.Save(settings); err != nil {
+		return err
 	}
 	// set auth settings
 	usersCollection, err := e.App.FindCollectionByNameOrId("users")
@@ -156,57 +163,9 @@ func (h *Hub) initialize(e *core.ServeEvent) error {
 	return nil
 }
 
-// startServer starts the server for the Beszel (not PocketBase)
-func (h *Hub) startServer(se *core.ServeEvent) error {
-	// TODO: exclude dev server from production binary
-	switch h.IsDev() {
-	case true:
-		proxy := httputil.NewSingleHostReverseProxy(&url.URL{
-			Scheme: "http",
-			Host:   "localhost:5173",
-		})
-		se.Router.GET("/{path...}", func(e *core.RequestEvent) error {
-			proxy.ServeHTTP(e.Response, e.Request)
-			return nil
-		})
-	default:
-		// parse app url
-		parsedURL, err := url.Parse(h.appURL)
-		if err != nil {
-			return err
-		}
-		// fix base paths in html if using subpath
-		basePath := strings.TrimSuffix(parsedURL.Path, "/") + "/"
-		indexFile, _ := fs.ReadFile(site.DistDirFS, "index.html")
-		indexContent := strings.ReplaceAll(string(indexFile), "./", basePath)
-		indexContent = strings.Replace(indexContent, "{{V}}", beszel.Version, 1)
-		// set up static asset serving
-		staticPaths := [2]string{"/static/", "/assets/"}
-		serveStatic := apis.Static(site.DistDirFS, false)
-		// get CSP configuration
-		csp, cspExists := GetEnv("CSP")
-		// add route
-		se.Router.GET("/{path...}", func(e *core.RequestEvent) error {
-			// serve static assets if path is in staticPaths
-			for i := range staticPaths {
-				if strings.Contains(e.Request.URL.Path, staticPaths[i]) {
-					e.Response.Header().Set("Cache-Control", "public, max-age=2592000")
-					return serveStatic(e)
-				}
-			}
-			if cspExists {
-				e.Response.Header().Del("X-Frame-Options")
-				e.Response.Header().Set("Content-Security-Policy", csp)
-			}
-			return e.HTML(http.StatusOK, indexContent)
-		})
-	}
-	return nil
-}
-
 // registerCronJobs sets up scheduled tasks
 func (h *Hub) registerCronJobs(_ *core.ServeEvent) error {
-	// delete old records once every hour
+	// delete old system_stats and alerts_history records once every hour
 	h.Cron().MustAdd("delete old records", "8 * * * *", h.rm.DeleteOldRecords)
 	// create longer records every 10 minutes
 	h.Cron().MustAdd("create longer records", "*/10 * * * *", h.rm.CreateLongerRecords)
@@ -215,97 +174,128 @@ func (h *Hub) registerCronJobs(_ *core.ServeEvent) error {
 
 // custom api routes
 func (h *Hub) registerApiRoutes(se *core.ServeEvent) error {
-	// returns public key and version
+	// auth protected routes
-	se.Router.GET("/api/beszel/getkey", func(e *core.RequestEvent) error {
+	apiAuth := se.Router.Group("/api/beszel")
-		info, _ := e.RequestInfo()
+	apiAuth.Bind(apis.RequireAuth())
-		if info.Auth == nil {
+	// auth optional routes
-			return apis.NewForbiddenError("Forbidden", nil)
+	apiNoAuth := se.Router.Group("/api/beszel")
-		}
+
-		return e.JSON(http.StatusOK, map[string]string{"key": h.pubKey, "v": beszel.Version})
+	// create first user endpoint only needed if no users exist
-	})
+	if totalUsers, _ := se.App.CountRecords("users"); totalUsers == 0 {
+		apiNoAuth.POST("/create-user", h.um.CreateFirstUser)
+	}
 	// check if first time setup on login page
-	se.Router.GET("/api/beszel/first-run", func(e *core.RequestEvent) error {
+	apiNoAuth.GET("/first-run", func(e *core.RequestEvent) error {
-		total, err := h.CountRecords("users")
+		total, err := e.App.CountRecords("users")
 		return e.JSON(http.StatusOK, map[string]bool{"firstRun": err == nil && total == 0})
 	})
+	// get public key and version
+	apiAuth.GET("/getkey", func(e *core.RequestEvent) error {
+		return e.JSON(http.StatusOK, map[string]string{"key": h.pubKey, "v": beszel.Version})
+	})
 	// send test notification
-	se.Router.GET("/api/beszel/send-test-notification", h.SendTestNotification)
+	apiAuth.POST("/test-notification", h.SendTestNotification)
-	// API endpoint to get config.yml content
+	// get config.yml content
-	se.Router.GET("/api/beszel/config-yaml", h.getYamlConfig)
+	apiAuth.GET("/config-yaml", config.GetYamlConfig)
-	// create first user endpoint only needed if no users exist
+	// handle agent websocket connection
-	if totalUsers, _ := h.CountRecords("users"); totalUsers == 0 {
+	apiNoAuth.GET("/agent-connect", h.handleAgentConnect)
-		se.Router.POST("/api/beszel/create-user", h.um.CreateFirstUser)
+	// get or create universal tokens
-	}
+	apiAuth.GET("/universal-token", h.getUniversalToken)
+	// update / delete user alerts
+	apiAuth.POST("/user-alerts", alerts.UpsertUserAlerts)
+	apiAuth.DELETE("/user-alerts", alerts.DeleteUserAlerts)
+
 	return nil
 }
 
-// generates key pair if it doesn't exist and returns private key bytes
+// Handler for universal token API endpoint (create, read, delete)
-func (h *Hub) GetSSHKey() ([]byte, error) {
+func (h *Hub) getUniversalToken(e *core.RequestEvent) error {
-	dataDir := h.DataDir()
+	tokenMap := universalTokenMap.GetMap()
-	// check if the key pair already exists
+	userID := e.Auth.Id
-	existingKey, err := os.ReadFile(dataDir + "/id_ed25519")
+	query := e.Request.URL.Query()
-	if err == nil {
+	token := query.Get("token")
-		if pubKey, err := os.ReadFile(h.DataDir() + "/id_ed25519.pub"); err == nil {
+
-			h.pubKey = strings.TrimSuffix(string(pubKey), "\n")
+	if token == "" {
+		// return existing token if it exists
+		if token, _, ok := tokenMap.GetByValue(userID); ok {
+			return e.JSON(http.StatusOK, map[string]any{"token": token, "active": true})
 		}
-		// return existing private key
+		// if no token is provided, generate a new one
-		return existingKey, nil
+		token = uuid.New().String()
+	}
+	response := map[string]any{"token": token}
+
+	switch query.Get("enable") {
+	case "1":
+		tokenMap.Set(token, userID, time.Hour)
+	case "0":
+		tokenMap.RemovebyValue(userID)
+	}
+	_, response["active"] = tokenMap.GetOk(token)
+	return e.JSON(http.StatusOK, response)
+}
+
+// generates key pair if it doesn't exist and returns signer
+func (h *Hub) GetSSHKey(dataDir string) (ssh.Signer, error) {
+	if h.signer != nil {
+		return h.signer, nil
+	}
+
+	if dataDir == "" {
+		dataDir = h.DataDir()
+	}
+
+	privateKeyPath := path.Join(dataDir, "id_ed25519")
+
+	// check if the key pair already exists
+	existingKey, err := os.ReadFile(privateKeyPath)
+	if err == nil {
+		private, err := ssh.ParsePrivateKey(existingKey)
+		if err != nil {
+			return nil, fmt.Errorf("failed to parse private key: %s", err)
+		}
+		pubKeyBytes := ssh.MarshalAuthorizedKey(private.PublicKey())
+		h.pubKey = strings.TrimSuffix(string(pubKeyBytes), "\n")
+		return private, nil
+	} else if !os.IsNotExist(err) {
+		// File exists but couldn't be read for some other reason
+		return nil, fmt.Errorf("failed to read %s: %w", privateKeyPath, err)
 	}
 
 	// Generate the Ed25519 key pair
-	pubKey, privKey, err := ed25519.GenerateKey(nil)
+	_, privKey, err := ed25519.GenerateKey(nil)
 	if err != nil {
-		// h.Logger().Error("Error generating key pair:", "err", err.Error())
 		return nil, err
 	}
-
+	privKeyPem, err := ssh.MarshalPrivateKey(privKey, "")
-	// Get the private key in OpenSSH format
-	privKeyBytes, err := ssh.MarshalPrivateKey(privKey, "")
-	if err != nil {
-		// h.Logger().Error("Error marshaling private key:", "err", err.Error())
-		return nil, err
-	}
-
-	// Save the private key to a file
-	privateFile, err := os.Create(dataDir + "/id_ed25519")
-	if err != nil {
-		// h.Logger().Error("Error creating private key file:", "err", err.Error())
-		return nil, err
-	}
-	defer privateFile.Close()
-
-	if err := pem.Encode(privateFile, privKeyBytes); err != nil {
-		// h.Logger().Error("Error writing private key to file:", "err", err.Error())
-		return nil, err
-	}
-
-	// Generate the public key in OpenSSH format
-	publicKey, err := ssh.NewPublicKey(pubKey)
 	if err != nil {
 		return nil, err
 	}
 
-	pubKeyBytes := ssh.MarshalAuthorizedKey(publicKey)
+	if err := os.WriteFile(privateKeyPath, pem.EncodeToMemory(privKeyPem), 0600); err != nil {
+		return nil, fmt.Errorf("failed to write private key to %q: err: %w", privateKeyPath, err)
+	}
+
+	// These are fine to ignore the errors on, as we've literally just created a crypto.PublicKey | crypto.Signer
+	sshPrivate, _ := ssh.NewSignerFromSigner(privKey)
+	pubKeyBytes := ssh.MarshalAuthorizedKey(sshPrivate.PublicKey())
 	h.pubKey = strings.TrimSuffix(string(pubKeyBytes), "\n")
 
-	// Save the public key to a file
+	h.Logger().Info("ed25519 key pair generated successfully.")
-	publicFile, err := os.Create(dataDir + "/id_ed25519.pub")
+	h.Logger().Info("Saved to: " + privateKeyPath)
-	if err != nil {
-		return nil, err
-	}
-	defer publicFile.Close()
 
-	if _, err := publicFile.Write(pubKeyBytes); err != nil {
+	return sshPrivate, err
-		return nil, err
+}
-	}
+
-
+// MakeLink formats a link with the app URL and path segments.
-	h.Logger().Info("ed25519 SSH key pair generated successfully.")
+// Only path segments should be provided.
-	h.Logger().Info("Private key saved to: " + dataDir + "/id_ed25519")
+func (h *Hub) MakeLink(parts ...string) string {
-	h.Logger().Info("Public key saved to: " + dataDir + "/id_ed25519.pub")
+	base := strings.TrimSuffix(h.Settings().Meta.AppURL, "/")
-
+	for _, part := range parts {
-	existingKey, err = os.ReadFile(dataDir + "/id_ed25519")
+		if part == "" {
-	if err == nil {
+			continue
-		return existingKey, nil
+		}
-	}
+		base = fmt.Sprintf("%s/%s", base, url.PathEscape(part))
-	return nil, err
+	}
+	return base
 }

beszel/internal/hub/hub_test.go

@@ -0,0 +1,603 @@
+//go:build testing
+// +build testing
+
+package hub_test
+
+import (
+	beszelTests "beszel/internal/tests"
+	"testing"
+
+	"bytes"
+	"crypto/ed25519"
+	"encoding/json"
+	"encoding/pem"
+	"io"
+	"net/http"
+	"os"
+	"path/filepath"
+	"strings"
+
+	"github.com/pocketbase/pocketbase/core"
+	pbTests "github.com/pocketbase/pocketbase/tests"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+	"golang.org/x/crypto/ssh"
+)
+
+// marshal to json and return an io.Reader (for use in ApiScenario.Body)
+func jsonReader(v any) io.Reader {
+	data, err := json.Marshal(v)
+	if err != nil {
+		panic(err)
+	}
+	return bytes.NewReader(data)
+}
+
+func TestMakeLink(t *testing.T) {
+	hub, _ := beszelTests.NewTestHub(t.TempDir())
+
+	tests := []struct {
+		name     string
+		appURL   string
+		parts    []string
+		expected string
+	}{
+		{
+			name:     "no parts, no trailing slash in AppURL",
+			appURL:   "http://localhost:8090",
+			parts:    []string{},
+			expected: "http://localhost:8090",
+		},
+		{
+			name:     "no parts, with trailing slash in AppURL",
+			appURL:   "http://localhost:8090/",
+			parts:    []string{},
+			expected: "http://localhost:8090", // TrimSuffix should handle the trailing slash
+		},
+		{
+			name:     "one part",
+			appURL:   "http://example.com",
+			parts:    []string{"one"},
+			expected: "http://example.com/one",
+		},
+		{
+			name:     "multiple parts",
+			appURL:   "http://example.com",
+			parts:    []string{"alpha", "beta", "gamma"},
+			expected: "http://example.com/alpha/beta/gamma",
+		},
+		{
+			name:     "parts with spaces needing escaping",
+			appURL:   "http://example.com",
+			parts:    []string{"path with spaces", "another part"},
+			expected: "http://example.com/path%20with%20spaces/another%20part",
+		},
+		{
+			name:     "parts with slashes needing escaping",
+			appURL:   "http://example.com",
+			parts:    []string{"a/b", "c"},
+			expected: "http://example.com/a%2Fb/c", // url.PathEscape escapes '/'
+		},
+		{
+			name:     "AppURL with subpath, no trailing slash",
+			appURL:   "http://localhost/sub",
+			parts:    []string{"resource"},
+			expected: "http://localhost/sub/resource",
+		},
+		{
+			name:     "AppURL with subpath, with trailing slash",
+			appURL:   "http://localhost/sub/",
+			parts:    []string{"item"},
+			expected: "http://localhost/sub/item",
+		},
+		{
+			name:     "empty parts in the middle",
+			appURL:   "http://localhost",
+			parts:    []string{"first", "", "third"},
+			expected: "http://localhost/first/third",
+		},
+		{
+			name:     "leading and trailing empty parts",
+			appURL:   "http://localhost",
+			parts:    []string{"", "path", ""},
+			expected: "http://localhost/path",
+		},
+		{
+			name:     "parts with various special characters",
+			appURL:   "https://test.dev/",
+			parts:    []string{"p@th?", "key=value&"},
+			expected: "https://test.dev/p@th%3F/key=value&",
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			// Store original app URL and restore it after the test
+			originalAppURL := hub.Settings().Meta.AppURL
+			hub.Settings().Meta.AppURL = tt.appURL
+			defer func() { hub.Settings().Meta.AppURL = originalAppURL }()
+
+			got := hub.MakeLink(tt.parts...)
+			assert.Equal(t, tt.expected, got, "MakeLink generated URL does not match expected")
+		})
+	}
+}
+
+func TestGetSSHKey(t *testing.T) {
+	hub, _ := beszelTests.NewTestHub(t.TempDir())
+
+	// Test Case 1: Key generation (no existing key)
+	t.Run("KeyGeneration", func(t *testing.T) {
+		tempDir := t.TempDir()
+
+		// Ensure pubKey is initially empty or different to ensure GetSSHKey sets it
+		hub.SetPubkey("")
+
+		signer, err := hub.GetSSHKey(tempDir)
+		assert.NoError(t, err, "GetSSHKey should not error when generating a new key")
+		assert.NotNil(t, signer, "GetSSHKey should return a non-nil signer")
+
+		// Check if private key file was created
+		privateKeyPath := filepath.Join(tempDir, "id_ed25519")
+		info, err := os.Stat(privateKeyPath)
+		assert.NoError(t, err, "Private key file should be created")
+		assert.False(t, info.IsDir(), "Private key path should be a file, not a directory")
+
+		// Check if h.pubKey was set
+		assert.NotEmpty(t, hub.GetPubkey(), "h.pubKey should be set after key generation")
+		assert.True(t, strings.HasPrefix(hub.GetPubkey(), "ssh-ed25519 "), "h.pubKey should start with 'ssh-ed25519 '")
+
+		// Verify the generated private key is parsable
+		keyData, err := os.ReadFile(privateKeyPath)
+		require.NoError(t, err)
+		_, err = ssh.ParsePrivateKey(keyData)
+		assert.NoError(t, err, "Generated private key should be parsable by ssh.ParsePrivateKey")
+	})
+
+	// Test Case 2: Existing key
+	t.Run("ExistingKey", func(t *testing.T) {
+		tempDir := t.TempDir()
+
+		// Manually create a valid key pair for the test
+		rawPubKey, rawPrivKey, err := ed25519.GenerateKey(nil)
+		require.NoError(t, err, "Failed to generate raw ed25519 key pair for pre-existing key test")
+
+		// Marshal the private key into OpenSSH PEM format
+		pemBlock, err := ssh.MarshalPrivateKey(rawPrivKey, "")
+		require.NoError(t, err, "Failed to marshal private key to PEM block for pre-existing key test")
+
+		privateKeyBytes := pem.EncodeToMemory(pemBlock)
+		require.NotNil(t, privateKeyBytes, "PEM encoded private key bytes should not be nil")
+
+		privateKeyPath := filepath.Join(tempDir, "id_ed25519")
+		err = os.WriteFile(privateKeyPath, privateKeyBytes, 0600)
+		require.NoError(t, err, "Failed to write pre-existing private key")
+
+		// Determine the expected public key string
+		sshPubKey, err := ssh.NewPublicKey(rawPubKey)
+		require.NoError(t, err)
+		expectedPubKeyStr := strings.TrimSpace(string(ssh.MarshalAuthorizedKey(sshPubKey)))
+
+		// Reset h.pubKey to ensure it's set by GetSSHKey from the file
+		hub.SetPubkey("")
+
+		signer, err := hub.GetSSHKey(tempDir)
+		assert.NoError(t, err, "GetSSHKey should not error when reading an existing key")
+		assert.NotNil(t, signer, "GetSSHKey should return a non-nil signer for an existing key")
+
+		// Check if h.pubKey was set correctly to the public key from the file
+		assert.Equal(t, expectedPubKeyStr, hub.GetPubkey(), "h.pubKey should match the existing public key")
+
+		// Verify the signer's public key matches the original public key
+		signerPubKey := signer.PublicKey()
+		marshaledSignerPubKey := strings.TrimSpace(string(ssh.MarshalAuthorizedKey(signerPubKey)))
+		assert.Equal(t, expectedPubKeyStr, marshaledSignerPubKey, "Signer's public key should match the existing public key")
+	})
+
+	// Test Case 3: Error cases
+	t.Run("ErrorCases", func(t *testing.T) {
+		tests := []struct {
+			name       string
+			setupFunc  func(dir string) error
+			errorCheck func(t *testing.T, err error)
+		}{
+			{
+				name: "CorruptedKey",
+				setupFunc: func(dir string) error {
+					return os.WriteFile(filepath.Join(dir, "id_ed25519"), []byte("this is not a valid SSH key"), 0600)
+				},
+				errorCheck: func(t *testing.T, err error) {
+					assert.Error(t, err)
+					assert.Contains(t, err.Error(), "ssh: no key found")
+				},
+			},
+			{
+				name: "PermissionDenied",
+				setupFunc: func(dir string) error {
+					// Create the key file
+					keyPath := filepath.Join(dir, "id_ed25519")
+					if err := os.WriteFile(keyPath, []byte("dummy content"), 0600); err != nil {
+						return err
+					}
+					// Make it read-only (can't be opened for writing in case a new key needs to be written)
+					return os.Chmod(keyPath, 0400)
+				},
+				errorCheck: func(t *testing.T, err error) {
+					// On read-only key, the parser will attempt to parse it and fail with "ssh: no key found"
+					assert.Error(t, err)
+				},
+			},
+			{
+				name: "EmptyFile",
+				setupFunc: func(dir string) error {
+					// Create an empty file
+					return os.WriteFile(filepath.Join(dir, "id_ed25519"), []byte{}, 0600)
+				},
+				errorCheck: func(t *testing.T, err error) {
+					assert.Error(t, err)
+					// The error from attempting to parse an empty file
+					assert.Contains(t, err.Error(), "ssh: no key found")
+				},
+			},
+		}
+
+		for _, tc := range tests {
+			t.Run(tc.name, func(t *testing.T) {
+				tempDir := t.TempDir()
+
+				// Setup the test case
+				err := tc.setupFunc(tempDir)
+				require.NoError(t, err, "Setup failed")
+
+				// Reset h.pubKey before each test case
+				hub.SetPubkey("")
+
+				// Attempt to get SSH key
+				_, err = hub.GetSSHKey(tempDir)
+
+				// Verify the error
+				tc.errorCheck(t, err)
+
+				// Check that pubKey was not set in error cases
+				assert.Empty(t, hub.GetPubkey(), "h.pubKey should not be set if there was an error")
+			})
+		}
+	})
+}
+
+func TestApiRoutesAuthentication(t *testing.T) {
+	hub, _ := beszelTests.NewTestHub(t.TempDir())
+	defer hub.Cleanup()
+
+	hub.StartHub()
+
+	// Create test user and get auth token
+	user, err := beszelTests.CreateUser(hub, "testuser@example.com", "password123")
+	require.NoError(t, err, "Failed to create test user")
+
+	adminUser, err := beszelTests.CreateRecord(hub, "users", map[string]any{
+		"email":    "admin@example.com",
+		"password": "password123",
+		"role":     "admin",
+	})
+	require.NoError(t, err, "Failed to create admin user")
+	adminUserToken, err := adminUser.NewAuthToken()
+
+	// superUser, err := beszelTests.CreateRecord(hub, core.CollectionNameSuperusers, map[string]any{
+	// 	"email":    "superuser@example.com",
+	// 	"password": "password123",
+	// })
+	// require.NoError(t, err, "Failed to create superuser")
+
+	userToken, err := user.NewAuthToken()
+	require.NoError(t, err, "Failed to create auth token")
+
+	// Create test system for user-alerts endpoints
+	system, err := beszelTests.CreateRecord(hub, "systems", map[string]any{
+		"name":  "test-system",
+		"users": []string{user.Id},
+		"host":  "127.0.0.1",
+	})
+	require.NoError(t, err, "Failed to create test system")
+
+	testAppFactory := func(t testing.TB) *pbTests.TestApp {
+		return hub.TestApp
+	}
+
+	scenarios := []beszelTests.ApiScenario{
+		// Auth Protected Routes - Should require authentication
+		{
+			Name:            "POST /test-notification - no auth should fail",
+			Method:          http.MethodPost,
+			URL:             "/api/beszel/test-notification",
+			ExpectedStatus:  401,
+			ExpectedContent: []string{"requires valid"},
+			TestAppFactory:  testAppFactory,
+			Body: jsonReader(map[string]any{
+				"url": "generic://127.0.0.1",
+			}),
+		},
+		{
+			Name:           "POST /test-notification - with auth should succeed",
+			Method:         http.MethodPost,
+			URL:            "/api/beszel/test-notification",
+			TestAppFactory: testAppFactory,
+			Headers: map[string]string{
+				"Authorization": userToken,
+			},
+			Body: jsonReader(map[string]any{
+				"url": "generic://127.0.0.1",
+			}),
+			ExpectedStatus:  200,
+			ExpectedContent: []string{"sending message"},
+		},
+		{
+			Name:            "GET /config-yaml - no auth should fail",
+			Method:          http.MethodGet,
+			URL:             "/api/beszel/config-yaml",
+			ExpectedStatus:  401,
+			ExpectedContent: []string{"requires valid"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "GET /config-yaml - with user auth should fail",
+			Method: http.MethodGet,
+			URL:    "/api/beszel/config-yaml",
+			Headers: map[string]string{
+				"Authorization": userToken,
+			},
+			ExpectedStatus:  403,
+			ExpectedContent: []string{"Requires admin"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "GET /config-yaml - with admin auth should succeed",
+			Method: http.MethodGet,
+			URL:    "/api/beszel/config-yaml",
+			Headers: map[string]string{
+				"Authorization": adminUserToken,
+			},
+			ExpectedStatus:  200,
+			ExpectedContent: []string{"test-system"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:            "GET /universal-token - no auth should fail",
+			Method:          http.MethodGet,
+			URL:             "/api/beszel/universal-token",
+			ExpectedStatus:  401,
+			ExpectedContent: []string{"requires valid"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "GET /universal-token - with auth should succeed",
+			Method: http.MethodGet,
+			URL:    "/api/beszel/universal-token",
+			Headers: map[string]string{
+				"Authorization": userToken,
+			},
+			ExpectedStatus:  200,
+			ExpectedContent: []string{"active", "token"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:            "POST /user-alerts - no auth should fail",
+			Method:          http.MethodPost,
+			URL:             "/api/beszel/user-alerts",
+			ExpectedStatus:  401,
+			ExpectedContent: []string{"requires valid"},
+			TestAppFactory:  testAppFactory,
+			Body: jsonReader(map[string]any{
+				"name":    "CPU",
+				"value":   80,
+				"min":     10,
+				"systems": []string{system.Id},
+			}),
+		},
+		{
+			Name:   "POST /user-alerts - with auth should succeed",
+			Method: http.MethodPost,
+			URL:    "/api/beszel/user-alerts",
+			Headers: map[string]string{
+				"Authorization": userToken,
+			},
+			ExpectedStatus:  200,
+			ExpectedContent: []string{"\"success\":true"},
+			TestAppFactory:  testAppFactory,
+			Body: jsonReader(map[string]any{
+				"name":    "CPU",
+				"value":   80,
+				"min":     10,
+				"systems": []string{system.Id},
+			}),
+		},
+		{
+			Name:            "DELETE /user-alerts - no auth should fail",
+			Method:          http.MethodDelete,
+			URL:             "/api/beszel/user-alerts",
+			ExpectedStatus:  401,
+			ExpectedContent: []string{"requires valid"},
+			TestAppFactory:  testAppFactory,
+			Body: jsonReader(map[string]any{
+				"name":    "CPU",
+				"systems": []string{system.Id},
+			}),
+		},
+		{
+			Name:   "DELETE /user-alerts - with auth should succeed",
+			Method: http.MethodDelete,
+			URL:    "/api/beszel/user-alerts",
+			Headers: map[string]string{
+				"Authorization": userToken,
+			},
+			ExpectedStatus:  200,
+			ExpectedContent: []string{"\"success\":true"},
+			TestAppFactory:  testAppFactory,
+			Body: jsonReader(map[string]any{
+				"name":    "CPU",
+				"systems": []string{system.Id},
+			}),
+			BeforeTestFunc: func(t testing.TB, app *pbTests.TestApp, e *core.ServeEvent) {
+				// Create an alert to delete
+				beszelTests.CreateRecord(app, "alerts", map[string]any{
+					"name":   "CPU",
+					"system": system.Id,
+					"user":   user.Id,
+					"value":  80,
+					"min":    10,
+				})
+			},
+		},
+
+		// Auth Optional Routes - Should work without authentication
+		{
+			Name:            "GET /getkey - no auth should fail",
+			Method:          http.MethodGet,
+			URL:             "/api/beszel/getkey",
+			ExpectedStatus:  401,
+			ExpectedContent: []string{"requires valid"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "GET /getkey - with auth should also succeed",
+			Method: http.MethodGet,
+			URL:    "/api/beszel/getkey",
+			Headers: map[string]string{
+				"Authorization": userToken,
+			},
+			ExpectedStatus:  200,
+			ExpectedContent: []string{"\"key\":", "\"v\":"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:            "GET /first-run - no auth should succeed",
+			Method:          http.MethodGet,
+			URL:             "/api/beszel/first-run",
+			ExpectedStatus:  200,
+			ExpectedContent: []string{"\"firstRun\":false"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "GET /first-run - with auth should also succeed",
+			Method: http.MethodGet,
+			URL:    "/api/beszel/first-run",
+			Headers: map[string]string{
+				"Authorization": userToken,
+			},
+			ExpectedStatus:  200,
+			ExpectedContent: []string{"\"firstRun\":false"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:            "GET /agent-connect - no auth should succeed (websocket upgrade fails but route is accessible)",
+			Method:          http.MethodGet,
+			URL:             "/api/beszel/agent-connect",
+			ExpectedStatus:  400,
+			ExpectedContent: []string{},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "POST /test-notification - invalid auth token should fail",
+			Method: http.MethodPost,
+			URL:    "/api/beszel/test-notification",
+			Body: jsonReader(map[string]any{
+				"url": "generic://127.0.0.1",
+			}),
+			Headers: map[string]string{
+				"Authorization": "invalid-token",
+			},
+			ExpectedStatus:  401,
+			ExpectedContent: []string{"requires valid"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "POST /user-alerts - invalid auth token should fail",
+			Method: http.MethodPost,
+			URL:    "/api/beszel/user-alerts",
+			Headers: map[string]string{
+				"Authorization": "invalid-token",
+			},
+			ExpectedStatus:  401,
+			ExpectedContent: []string{"requires valid"},
+			TestAppFactory:  testAppFactory,
+			Body: jsonReader(map[string]any{
+				"name":    "CPU",
+				"value":   80,
+				"min":     10,
+				"systems": []string{system.Id},
+			}),
+		},
+	}
+
+	for _, scenario := range scenarios {
+		scenario.Test(t)
+	}
+}
+
+func TestCreateUserEndpointAvailability(t *testing.T) {
+	t.Run("CreateUserEndpoint available when no users exist", func(t *testing.T) {
+		hub, _ := beszelTests.NewTestHub(t.TempDir())
+		defer hub.Cleanup()
+
+		// Ensure no users exist
+		userCount, err := hub.CountRecords("users")
+		require.NoError(t, err)
+		require.Zero(t, userCount, "Should start with no users")
+
+		hub.StartHub()
+
+		testAppFactory := func(t testing.TB) *pbTests.TestApp {
+			return hub.TestApp
+		}
+
+		scenario := beszelTests.ApiScenario{
+			Name:   "POST /create-user - should be available when no users exist",
+			Method: http.MethodPost,
+			URL:    "/api/beszel/create-user",
+			Body: jsonReader(map[string]any{
+				"email":    "firstuser@example.com",
+				"password": "password123",
+			}),
+			ExpectedStatus:  200,
+			ExpectedContent: []string{"User created"},
+			TestAppFactory:  testAppFactory,
+		}
+
+		scenario.Test(t)
+
+		// Verify user was created
+		userCount, err = hub.CountRecords("users")
+		require.NoError(t, err)
+		require.EqualValues(t, 1, userCount, "Should have created one user")
+	})
+
+	t.Run("CreateUserEndpoint not available when users exist", func(t *testing.T) {
+		hub, _ := beszelTests.NewTestHub(t.TempDir())
+		defer hub.Cleanup()
+
+		// Create a user first
+		_, err := beszelTests.CreateUser(hub, "existing@example.com", "password")
+		require.NoError(t, err)
+
+		hub.StartHub()
+
+		testAppFactory := func(t testing.TB) *pbTests.TestApp {
+			return hub.TestApp
+		}
+
+		scenario := beszelTests.ApiScenario{
+			Name:   "POST /create-user - should not be available when users exist",
+			Method: http.MethodPost,
+			URL:    "/api/beszel/create-user",
+			Body: jsonReader(map[string]any{
+				"email":    "another@example.com",
+				"password": "password123",
+			}),
+			ExpectedStatus:  404,
+			ExpectedContent: []string{"wasn't found"},
+			TestAppFactory:  testAppFactory,
+		}
+
+		scenario.Test(t)
+	})
+}

beszel/internal/hub/hub_test_helpers.go

@@ -0,0 +1,21 @@
+//go:build testing
+// +build testing
+
+package hub
+
+import "beszel/internal/hub/systems"
+
+// TESTING ONLY: GetSystemManager returns the system manager
+func (h *Hub) GetSystemManager() *systems.SystemManager {
+	return h.sm
+}
+
+// TESTING ONLY: GetPubkey returns the public key
+func (h *Hub) GetPubkey() string {
+	return h.pubKey
+}
+
+// TESTING ONLY: SetPubkey sets the public key
+func (h *Hub) SetPubkey(pubkey string) {
+	h.pubKey = pubkey
+}

beszel/internal/hub/server_development.go

@@ -0,0 +1,79 @@
+//go:build development
+
+package hub
+
+import (
+	"beszel"
+	"fmt"
+	"io"
+	"log/slog"
+	"net/http"
+	"net/http/httputil"
+	"net/url"
+	"strings"
+
+	"github.com/pocketbase/pocketbase/core"
+)
+
+// Wraps http.RoundTripper to modify dev proxy HTML responses
+type responseModifier struct {
+	transport http.RoundTripper
+	hub       *Hub
+}
+
+func (rm *responseModifier) RoundTrip(req *http.Request) (*http.Response, error) {
+	resp, err := rm.transport.RoundTrip(req)
+	if err != nil {
+		return resp, err
+	}
+	// Only modify HTML responses
+	contentType := resp.Header.Get("Content-Type")
+	if !strings.Contains(contentType, "text/html") {
+		return resp, nil
+	}
+	body, err := io.ReadAll(resp.Body)
+	if err != nil {
+		return resp, err
+	}
+	resp.Body.Close()
+	// Create a new response with the modified body
+	modifiedBody := rm.modifyHTML(string(body))
+	resp.Body = io.NopCloser(strings.NewReader(modifiedBody))
+	resp.ContentLength = int64(len(modifiedBody))
+	resp.Header.Set("Content-Length", fmt.Sprintf("%d", len(modifiedBody)))
+
+	return resp, nil
+}
+
+func (rm *responseModifier) modifyHTML(html string) string {
+	parsedURL, err := url.Parse(rm.hub.appURL)
+	if err != nil {
+		return html
+	}
+	// fix base paths in html if using subpath
+	basePath := strings.TrimSuffix(parsedURL.Path, "/") + "/"
+	html = strings.ReplaceAll(html, "./", basePath)
+	html = strings.Replace(html, "{{V}}", beszel.Version, 1)
+	html = strings.Replace(html, "{{HUB_URL}}", rm.hub.appURL, 1)
+	return html
+}
+
+// startServer sets up the development server for Beszel
+func (h *Hub) startServer(se *core.ServeEvent) error {
+	slog.Info("starting server", "appURL", h.appURL)
+	proxy := httputil.NewSingleHostReverseProxy(&url.URL{
+		Scheme: "http",
+		Host:   "localhost:5173",
+	})
+
+	proxy.Transport = &responseModifier{
+		transport: http.DefaultTransport,
+		hub:       h,
+	}
+
+	se.Router.GET("/{path...}", func(e *core.RequestEvent) error {
+		proxy.ServeHTTP(e.Response, e.Request)
+		return nil
+	})
+	return nil
+}

beszel/internal/hub/server_production.go

@@ -0,0 +1,51 @@
+//go:build !development
+
+package hub
+
+import (
+	"beszel"
+	"beszel/site"
+	"io/fs"
+	"net/http"
+	"net/url"
+	"strings"
+
+	"github.com/pocketbase/pocketbase/apis"
+	"github.com/pocketbase/pocketbase/core"
+)
+
+// startServer sets up the production server for Beszel
+func (h *Hub) startServer(se *core.ServeEvent) error {
+	// parse app url
+	parsedURL, err := url.Parse(h.appURL)
+	if err != nil {
+		return err
+	}
+	// fix base paths in html if using subpath
+	basePath := strings.TrimSuffix(parsedURL.Path, "/") + "/"
+	indexFile, _ := fs.ReadFile(site.DistDirFS, "index.html")
+	html := strings.ReplaceAll(string(indexFile), "./", basePath)
+	html = strings.Replace(html, "{{V}}", beszel.Version, 1)
+	html = strings.Replace(html, "{{HUB_URL}}", h.appURL, 1)
+	// set up static asset serving
+	staticPaths := [2]string{"/static/", "/assets/"}
+	serveStatic := apis.Static(site.DistDirFS, false)
+	// get CSP configuration
+	csp, cspExists := GetEnv("CSP")
+	// add route
+	se.Router.GET("/{path...}", func(e *core.RequestEvent) error {
+		// serve static assets if path is in staticPaths
+		for i := range staticPaths {
+			if strings.Contains(e.Request.URL.Path, staticPaths[i]) {
+				e.Response.Header().Set("Cache-Control", "public, max-age=2592000")
+				return serveStatic(e)
+			}
+		}
+		if cspExists {
+			e.Response.Header().Del("X-Frame-Options")
+			e.Response.Header().Set("Content-Security-Policy", csp)
+		}
+		return e.HTML(http.StatusOK, html)
+	})
+	return nil
+}

beszel/internal/hub/systems/system.go

@@ -0,0 +1,387 @@
+package systems
+
+import (
+	"beszel"
+	"beszel/internal/entities/system"
+	"beszel/internal/hub/ws"
+	"context"
+	"encoding/json"
+	"errors"
+	"fmt"
+	"math/rand"
+	"net"
+	"strings"
+	"time"
+
+	"github.com/blang/semver"
+	"github.com/fxamacker/cbor/v2"
+	"github.com/pocketbase/pocketbase/core"
+	"golang.org/x/crypto/ssh"
+)
+
+type System struct {
+	Id           string               `db:"id"`
+	Host         string               `db:"host"`
+	Port         string               `db:"port"`
+	Status       string               `db:"status"`
+	manager      *SystemManager       // Manager that this system belongs to
+	client       *ssh.Client          // SSH client for fetching data
+	data         *system.CombinedData // system data from agent
+	ctx          context.Context      // Context for stopping the updater
+	cancel       context.CancelFunc   // Stops and removes system from updater
+	WsConn       *ws.WsConn           // Handler for agent WebSocket connection
+	agentVersion semver.Version       // Agent version
+	updateTicker *time.Ticker         // Ticker for updating the system
+}
+
+func (sm *SystemManager) NewSystem(systemId string) *System {
+	system := &System{
+		Id:   systemId,
+		data: &system.CombinedData{},
+	}
+	system.ctx, system.cancel = system.getContext()
+	return system
+}
+
+// StartUpdater starts the system updater.
+// It first fetches the data from the agent then updates the records.
+// If the data is not found or the system is down, it sets the system down.
+func (sys *System) StartUpdater() {
+	// Channel that can be used to set the system down. Currently only used to
+	// allow a short delay for reconnection after websocket connection is closed.
+	var downChan chan struct{}
+
+	// Add random jitter to first WebSocket connection to prevent
+	// clustering if all agents are started at the same time.
+	// SSH connections during hub startup are already staggered.
+	var jitter <-chan time.Time
+	if sys.WsConn != nil {
+		jitter = getJitter()
+		// use the websocket connection's down channel to set the system down
+		downChan = sys.WsConn.DownChan
+	} else {
+		// if the system does not have a websocket connection, wait before updating
+		// to allow the agent to connect via websocket (makes sure fingerprint is set).
+		time.Sleep(11 * time.Second)
+	}
+
+	// update immediately if system is not paused (only for ws connections)
+	// we'll wait a minute before connecting via SSH to prioritize ws connections
+	if sys.Status != paused && sys.ctx.Err() == nil {
+		if err := sys.update(); err != nil {
+			_ = sys.setDown(err)
+		}
+	}
+
+	sys.updateTicker = time.NewTicker(time.Duration(interval) * time.Millisecond)
+	// Go 1.23+ will automatically stop the ticker when the system is garbage collected, however we seem to need this or testing/synctest will block even if calling runtime.GC()
+	defer sys.updateTicker.Stop()
+
+	for {
+		select {
+		case <-sys.ctx.Done():
+			return
+		case <-sys.updateTicker.C:
+			if err := sys.update(); err != nil {
+				_ = sys.setDown(err)
+			}
+		case <-downChan:
+			sys.WsConn = nil
+			downChan = nil
+			_ = sys.setDown(nil)
+		case <-jitter:
+			sys.updateTicker.Reset(time.Duration(interval) * time.Millisecond)
+			if err := sys.update(); err != nil {
+				_ = sys.setDown(err)
+			}
+		}
+	}
+}
+
+// update updates the system data and records.
+func (sys *System) update() error {
+	if sys.Status == paused {
+		sys.handlePaused()
+		return nil
+	}
+	data, err := sys.fetchDataFromAgent()
+	if err == nil {
+		_, err = sys.createRecords(data)
+	}
+	return err
+}
+
+func (sys *System) handlePaused() {
+	if sys.WsConn == nil {
+		// if the system is paused and there's no websocket connection, remove the system
+		_ = sys.manager.RemoveSystem(sys.Id)
+	} else {
+		// Send a ping to the agent to keep the connection alive if the system is paused
+		if err := sys.WsConn.Ping(); err != nil {
+			sys.manager.hub.Logger().Warn("Failed to ping agent", "system", sys.Id, "err", err)
+			_ = sys.manager.RemoveSystem(sys.Id)
+		}
+	}
+}
+
+// createRecords updates the system record and adds system_stats and container_stats records
+func (sys *System) createRecords(data *system.CombinedData) (*core.Record, error) {
+	systemRecord, err := sys.getRecord()
+	if err != nil {
+		return nil, err
+	}
+	hub := sys.manager.hub
+	// add system_stats and container_stats records
+	systemStatsCollection, err := hub.FindCachedCollectionByNameOrId("system_stats")
+	if err != nil {
+		return nil, err
+	}
+
+	systemStatsRecord := core.NewRecord(systemStatsCollection)
+	systemStatsRecord.Set("system", systemRecord.Id)
+	systemStatsRecord.Set("stats", data.Stats)
+	systemStatsRecord.Set("type", "1m")
+	if err := hub.SaveNoValidate(systemStatsRecord); err != nil {
+		return nil, err
+	}
+	// add new container_stats record
+	if len(data.Containers) > 0 {
+		containerStatsCollection, err := hub.FindCachedCollectionByNameOrId("container_stats")
+		if err != nil {
+			return nil, err
+		}
+		containerStatsRecord := core.NewRecord(containerStatsCollection)
+		containerStatsRecord.Set("system", systemRecord.Id)
+		containerStatsRecord.Set("stats", data.Containers)
+		containerStatsRecord.Set("type", "1m")
+		if err := hub.SaveNoValidate(containerStatsRecord); err != nil {
+			return nil, err
+		}
+	}
+	// update system record (do this last because it triggers alerts and we need above records to be inserted first)
+	systemRecord.Set("status", up)
+
+	systemRecord.Set("info", data.Info)
+	if err := hub.SaveNoValidate(systemRecord); err != nil {
+		return nil, err
+	}
+	return systemRecord, nil
+}
+
+// getRecord retrieves the system record from the database.
+// If the record is not found, it removes the system from the manager.
+func (sys *System) getRecord() (*core.Record, error) {
+	record, err := sys.manager.hub.FindRecordById("systems", sys.Id)
+	if err != nil || record == nil {
+		_ = sys.manager.RemoveSystem(sys.Id)
+		return nil, err
+	}
+	return record, nil
+}
+
+// setDown marks a system as down in the database.
+// It takes the original error that caused the system to go down and returns any error
+// encountered during the process of updating the system status.
+func (sys *System) setDown(originalError error) error {
+	if sys.Status == down || sys.Status == paused {
+		return nil
+	}
+	record, err := sys.getRecord()
+	if err != nil {
+		return err
+	}
+	if originalError != nil {
+		sys.manager.hub.Logger().Error("System down", "system", record.GetString("name"), "err", originalError)
+	}
+	record.Set("status", down)
+	return sys.manager.hub.SaveNoValidate(record)
+}
+
+func (sys *System) getContext() (context.Context, context.CancelFunc) {
+	if sys.ctx == nil {
+		sys.ctx, sys.cancel = context.WithCancel(context.Background())
+	}
+	return sys.ctx, sys.cancel
+}
+
+// fetchDataFromAgent attempts to fetch data from the agent,
+// prioritizing WebSocket if available.
+func (sys *System) fetchDataFromAgent() (*system.CombinedData, error) {
+	if sys.data == nil {
+		sys.data = &system.CombinedData{}
+	}
+
+	if sys.WsConn != nil && sys.WsConn.IsConnected() {
+		wsData, err := sys.fetchDataViaWebSocket()
+		if err == nil {
+			return wsData, nil
+		}
+		// close the WebSocket connection if error and try SSH
+		sys.closeWebSocketConnection()
+	}
+
+	sshData, err := sys.fetchDataViaSSH()
+	if err != nil {
+		return nil, err
+	}
+	return sshData, nil
+}
+
+func (sys *System) fetchDataViaWebSocket() (*system.CombinedData, error) {
+	if sys.WsConn == nil || !sys.WsConn.IsConnected() {
+		return nil, errors.New("no websocket connection")
+	}
+	err := sys.WsConn.RequestSystemData(sys.data)
+	if err != nil {
+		return nil, err
+	}
+	return sys.data, nil
+}
+
+// fetchDataViaSSH handles fetching data using SSH.
+// This function encapsulates the original SSH logic.
+// It updates sys.data directly upon successful fetch.
+func (sys *System) fetchDataViaSSH() (*system.CombinedData, error) {
+	maxRetries := 1
+	for attempt := 0; attempt <= maxRetries; attempt++ {
+		if sys.client == nil || sys.Status == down {
+			if err := sys.createSSHClient(); err != nil {
+				return nil, err
+			}
+		}
+
+		session, err := sys.createSessionWithTimeout(4 * time.Second)
+		if err != nil {
+			if attempt >= maxRetries {
+				return nil, err
+			}
+			sys.manager.hub.Logger().Warn("Session closed. Retrying...", "host", sys.Host, "port", sys.Port, "err", err)
+			sys.closeSSHConnection()
+			// Reset format detection on connection failure - agent might have been upgraded
+			continue
+		}
+		defer session.Close()
+
+		stdout, err := session.StdoutPipe()
+		if err != nil {
+			return nil, err
+		}
+		if err := session.Shell(); err != nil {
+			return nil, err
+		}
+
+		*sys.data = system.CombinedData{}
+
+		if sys.agentVersion.GTE(beszel.MinVersionCbor) {
+			err = cbor.NewDecoder(stdout).Decode(sys.data)
+		} else {
+			err = json.NewDecoder(stdout).Decode(sys.data)
+		}
+
+		if err != nil {
+			sys.closeSSHConnection()
+			if attempt < maxRetries {
+				continue
+			}
+			return nil, err
+		}
+
+		// wait for the session to complete
+		if err := session.Wait(); err != nil {
+			return nil, err
+		}
+
+		return sys.data, nil
+	}
+
+	// this should never be reached due to the return in the loop
+	return nil, fmt.Errorf("failed to fetch data")
+}
+
+// createSSHClient creates a new SSH client for the system
+func (s *System) createSSHClient() error {
+	if s.manager.sshConfig == nil {
+		if err := s.manager.createSSHClientConfig(); err != nil {
+			return err
+		}
+	}
+	network := "tcp"
+	host := s.Host
+	if strings.HasPrefix(host, "/") {
+		network = "unix"
+	} else {
+		host = net.JoinHostPort(host, s.Port)
+	}
+	var err error
+	s.client, err = ssh.Dial(network, host, s.manager.sshConfig)
+	if err != nil {
+		return err
+	}
+	s.agentVersion, _ = extractAgentVersion(string(s.client.Conn.ServerVersion()))
+	return nil
+}
+
+// createSessionWithTimeout creates a new SSH session with a timeout to avoid hanging
+// in case of network issues
+func (sys *System) createSessionWithTimeout(timeout time.Duration) (*ssh.Session, error) {
+	if sys.client == nil {
+		return nil, fmt.Errorf("client not initialized")
+	}
+
+	ctx, cancel := context.WithTimeout(sys.ctx, timeout)
+	defer cancel()
+
+	sessionChan := make(chan *ssh.Session, 1)
+	errChan := make(chan error, 1)
+
+	go func() {
+		if session, err := sys.client.NewSession(); err != nil {
+			errChan <- err
+		} else {
+			sessionChan <- session
+		}
+	}()
+
+	select {
+	case session := <-sessionChan:
+		return session, nil
+	case err := <-errChan:
+		return nil, err
+	case <-ctx.Done():
+		return nil, fmt.Errorf("timeout")
+	}
+}
+
+// closeSSHConnection closes the SSH connection but keeps the system in the manager
+func (sys *System) closeSSHConnection() {
+	if sys.client != nil {
+		sys.client.Close()
+		sys.client = nil
+	}
+}
+
+// closeWebSocketConnection closes the WebSocket connection but keeps the system in the manager
+// to allow updating via SSH. It will be removed if the WS connection is re-established.
+// The system will be set as down a few seconds later if the connection is not re-established.
+func (sys *System) closeWebSocketConnection() {
+	if sys.WsConn != nil {
+		sys.WsConn.Close(nil)
+	}
+}
+
+// extractAgentVersion extracts the beszel version from SSH server version string
+func extractAgentVersion(versionString string) (semver.Version, error) {
+	_, after, _ := strings.Cut(versionString, "_")
+	return semver.Parse(after)
+}
+
+// getJitter returns a channel that will be triggered after a random delay
+// between 40% and 90% of the interval.
+// This is used to stagger the initial WebSocket connections to prevent clustering.
+func getJitter() <-chan time.Time {
+	minPercent := 40
+	maxPercent := 90
+	jitterRange := maxPercent - minPercent
+	msDelay := (interval * minPercent / 100) + rand.Intn(interval*jitterRange/100)
+	return time.After(time.Duration(msDelay) * time.Millisecond)
+}

beszel/internal/hub/systems/system_manager.go

@@ -0,0 +1,346 @@
+package systems
+
+import (
+	"beszel"
+	"beszel/internal/common"
+	"beszel/internal/entities/system"
+	"beszel/internal/hub/ws"
+	"errors"
+	"fmt"
+	"time"
+
+	"github.com/blang/semver"
+	"github.com/pocketbase/pocketbase/core"
+	"github.com/pocketbase/pocketbase/tools/store"
+	"golang.org/x/crypto/ssh"
+)
+
+// System status constants
+const (
+	up      string = "up"      // System is online and responding
+	down    string = "down"    // System is offline or not responding
+	paused  string = "paused"  // System monitoring is paused
+	pending string = "pending" // System is waiting on initial connection result
+
+	// interval is the default update interval in milliseconds (60 seconds)
+	interval int = 60_000
+	// interval int = 10_000 // Debug interval for faster updates
+
+	// sessionTimeout is the maximum time to wait for SSH connections
+	sessionTimeout = 4 * time.Second
+)
+
+var (
+	// errSystemExists is returned when attempting to add a system that already exists
+	errSystemExists = errors.New("system exists")
+)
+
+// SystemManager manages a collection of monitored systems and their connections.
+// It handles system lifecycle, status updates, and maintains both SSH and WebSocket connections.
+type SystemManager struct {
+	hub       hubLike                       // Hub interface for database and alert operations
+	systems   *store.Store[string, *System] // Thread-safe store of active systems
+	sshConfig *ssh.ClientConfig             // SSH client configuration for system connections
+}
+
+// hubLike defines the interface requirements for the hub dependency.
+// It extends core.App with system-specific functionality.
+type hubLike interface {
+	core.App
+	GetSSHKey(dataDir string) (ssh.Signer, error)
+	HandleSystemAlerts(systemRecord *core.Record, data *system.CombinedData) error
+	HandleStatusAlerts(status string, systemRecord *core.Record) error
+}
+
+// NewSystemManager creates a new SystemManager instance with the provided hub.
+// The hub must implement the hubLike interface to provide database and alert functionality.
+func NewSystemManager(hub hubLike) *SystemManager {
+	return &SystemManager{
+		systems: store.New(map[string]*System{}),
+		hub:     hub,
+	}
+}
+
+// Initialize sets up the system manager by binding event hooks and starting existing systems.
+// It configures SSH client settings and begins monitoring all non-paused systems from the database.
+// Systems are started with staggered delays to prevent overwhelming the hub during startup.
+func (sm *SystemManager) Initialize() error {
+	sm.bindEventHooks()
+
+	// Initialize SSH client configuration
+	err := sm.createSSHClientConfig()
+	if err != nil {
+		return err
+	}
+
+	// Load existing systems from database (excluding paused ones)
+	var systems []*System
+	err = sm.hub.DB().NewQuery("SELECT id, host, port, status FROM systems WHERE status != 'paused'").All(&systems)
+	if err != nil || len(systems) == 0 {
+		return err
+	}
+
+	// Start systems in background with staggered timing
+	go func() {
+		// Calculate staggered delay between system starts (max 2 seconds per system)
+		delta := interval / max(1, len(systems))
+		delta = min(delta, 2_000)
+		sleepTime := time.Duration(delta) * time.Millisecond
+
+		for _, system := range systems {
+			time.Sleep(sleepTime)
+			_ = sm.AddSystem(system)
+		}
+	}()
+	return nil
+}
+
+// bindEventHooks registers event handlers for system and fingerprint record changes.
+// These hooks ensure the system manager stays synchronized with database changes.
+func (sm *SystemManager) bindEventHooks() {
+	sm.hub.OnRecordCreate("systems").BindFunc(sm.onRecordCreate)
+	sm.hub.OnRecordAfterCreateSuccess("systems").BindFunc(sm.onRecordAfterCreateSuccess)
+	sm.hub.OnRecordUpdate("systems").BindFunc(sm.onRecordUpdate)
+	sm.hub.OnRecordAfterUpdateSuccess("systems").BindFunc(sm.onRecordAfterUpdateSuccess)
+	sm.hub.OnRecordAfterDeleteSuccess("systems").BindFunc(sm.onRecordAfterDeleteSuccess)
+	sm.hub.OnRecordAfterUpdateSuccess("fingerprints").BindFunc(sm.onTokenRotated)
+}
+
+// onTokenRotated handles fingerprint token rotation events.
+// When a system's authentication token is rotated, any existing WebSocket connection
+// must be closed to force re-authentication with the new token.
+func (sm *SystemManager) onTokenRotated(e *core.RecordEvent) error {
+	systemID := e.Record.GetString("system")
+	system, ok := sm.systems.GetOk(systemID)
+	if !ok {
+		return e.Next()
+	}
+	// No need to close connection if not connected via websocket
+	if system.WsConn == nil {
+		return e.Next()
+	}
+	system.setDown(nil)
+	sm.RemoveSystem(systemID)
+	return e.Next()
+}
+
+// onRecordCreate is called before a new system record is committed to the database.
+// It initializes the record with default values: empty info and pending status.
+func (sm *SystemManager) onRecordCreate(e *core.RecordEvent) error {
+	e.Record.Set("info", system.Info{})
+	e.Record.Set("status", pending)
+	return e.Next()
+}
+
+// onRecordAfterCreateSuccess is called after a new system record is successfully created.
+// It adds the new system to the manager to begin monitoring.
+func (sm *SystemManager) onRecordAfterCreateSuccess(e *core.RecordEvent) error {
+	if err := sm.AddRecord(e.Record, nil); err != nil {
+		e.App.Logger().Error("Error adding record", "err", err)
+	}
+	return e.Next()
+}
+
+// onRecordUpdate is called before a system record is updated in the database.
+// It clears system info when the status is changed to paused.
+func (sm *SystemManager) onRecordUpdate(e *core.RecordEvent) error {
+	if e.Record.GetString("status") == paused {
+		e.Record.Set("info", system.Info{})
+	}
+	return e.Next()
+}
+
+// onRecordAfterUpdateSuccess handles system record updates after they're committed to the database.
+// It manages system lifecycle based on status changes and triggers appropriate alerts.
+// Status transitions are handled as follows:
+// - paused: Closes SSH connection and deactivates alerts
+// - pending: Starts monitoring (reuses WebSocket if available)
+// - up: Triggers system alerts
+// - down: Triggers status change alerts
+func (sm *SystemManager) onRecordAfterUpdateSuccess(e *core.RecordEvent) error {
+	newStatus := e.Record.GetString("status")
+	prevStatus := pending
+	system, ok := sm.systems.GetOk(e.Record.Id)
+	if ok {
+		prevStatus = system.Status
+		system.Status = newStatus
+	}
+
+	switch newStatus {
+	case paused:
+		if ok {
+			// Pause monitoring but keep system in manager for potential resume
+			system.closeSSHConnection()
+		}
+		_ = deactivateAlerts(e.App, e.Record.Id)
+		return e.Next()
+	case pending:
+		// Resume monitoring, preferring existing WebSocket connection
+		if ok && system.WsConn != nil {
+			go system.update()
+			return e.Next()
+		}
+		// Start new monitoring session
+		if err := sm.AddRecord(e.Record, nil); err != nil {
+			e.App.Logger().Error("Error adding record", "err", err)
+		}
+		_ = deactivateAlerts(e.App, e.Record.Id)
+		return e.Next()
+	}
+
+	// Handle systems not in manager
+	if !ok {
+		return sm.AddRecord(e.Record, nil)
+	}
+
+	// Trigger system alerts when system comes online
+	if newStatus == up {
+		if err := sm.hub.HandleSystemAlerts(e.Record, system.data); err != nil {
+			e.App.Logger().Error("Error handling system alerts", "err", err)
+		}
+	}
+
+	// Trigger status change alerts for up/down transitions
+	if (newStatus == down && prevStatus == up) || (newStatus == up && prevStatus == down) {
+		if err := sm.hub.HandleStatusAlerts(newStatus, e.Record); err != nil {
+			e.App.Logger().Error("Error handling status alerts", "err", err)
+		}
+	}
+	return e.Next()
+}
+
+// onRecordAfterDeleteSuccess is called after a system record is successfully deleted.
+// It removes the system from the manager and cleans up all associated resources.
+func (sm *SystemManager) onRecordAfterDeleteSuccess(e *core.RecordEvent) error {
+	sm.RemoveSystem(e.Record.Id)
+	return e.Next()
+}
+
+// AddSystem adds a system to the manager and starts monitoring it.
+// It validates required fields, initializes the system context, and starts the update goroutine.
+// Returns error if a system with the same ID already exists.
+func (sm *SystemManager) AddSystem(sys *System) error {
+	if sm.systems.Has(sys.Id) {
+		return errSystemExists
+	}
+	if sys.Id == "" || sys.Host == "" {
+		return errors.New("system missing required fields")
+	}
+
+	// Initialize system for monitoring
+	sys.manager = sm
+	sys.ctx, sys.cancel = sys.getContext()
+	sys.data = &system.CombinedData{}
+	sm.systems.Set(sys.Id, sys)
+
+	// Start monitoring in background
+	go sys.StartUpdater()
+	return nil
+}
+
+// RemoveSystem removes a system from the manager and cleans up all associated resources.
+// It cancels the system's context, closes all connections, and removes it from the store.
+// Returns an error if the system is not found.
+func (sm *SystemManager) RemoveSystem(systemID string) error {
+	system, ok := sm.systems.GetOk(systemID)
+	if !ok {
+		return errors.New("system not found")
+	}
+
+	// Stop the update goroutine
+	if system.cancel != nil {
+		system.cancel()
+	}
+
+	// Clean up all connections
+	system.closeSSHConnection()
+	system.closeWebSocketConnection()
+	sm.systems.Remove(systemID)
+	return nil
+}
+
+// AddRecord creates a System instance from a database record and adds it to the manager.
+// If a system with the same ID already exists, it's removed first to ensure clean state.
+// If no system instance is provided, a new one is created.
+// This method is typically called when systems are created or their status changes to pending.
+func (sm *SystemManager) AddRecord(record *core.Record, system *System) (err error) {
+	// Remove existing system to ensure clean state
+	if sm.systems.Has(record.Id) {
+		_ = sm.RemoveSystem(record.Id)
+	}
+
+	// Create new system if none provided
+	if system == nil {
+		system = sm.NewSystem(record.Id)
+	}
+
+	// Populate system from record
+	system.Status = record.GetString("status")
+	system.Host = record.GetString("host")
+	system.Port = record.GetString("port")
+
+	return sm.AddSystem(system)
+}
+
+// AddWebSocketSystem creates and adds a system with an established WebSocket connection.
+// This method is called when an agent connects via WebSocket with valid authentication.
+// The system is immediately added to monitoring with the provided connection and version info.
+func (sm *SystemManager) AddWebSocketSystem(systemId string, agentVersion semver.Version, wsConn *ws.WsConn) error {
+	systemRecord, err := sm.hub.FindRecordById("systems", systemId)
+	if err != nil {
+		return err
+	}
+
+	system := sm.NewSystem(systemId)
+	system.WsConn = wsConn
+	system.agentVersion = agentVersion
+
+	if err := sm.AddRecord(systemRecord, system); err != nil {
+		return err
+	}
+	return nil
+}
+
+// createSSHClientConfig initializes the SSH client configuration for connecting to an agent's server
+func (sm *SystemManager) createSSHClientConfig() error {
+	privateKey, err := sm.hub.GetSSHKey("")
+	if err != nil {
+		return err
+	}
+
+	sm.sshConfig = &ssh.ClientConfig{
+		User: "u",
+		Auth: []ssh.AuthMethod{
+			ssh.PublicKeys(privateKey),
+		},
+		Config: ssh.Config{
+			Ciphers:      common.DefaultCiphers,
+			KeyExchanges: common.DefaultKeyExchanges,
+			MACs:         common.DefaultMACs,
+		},
+		HostKeyCallback: ssh.InsecureIgnoreHostKey(),
+		ClientVersion:   fmt.Sprintf("SSH-2.0-%s_%s", beszel.AppName, beszel.Version),
+		Timeout:         sessionTimeout,
+	}
+	return nil
+}
+
+// deactivateAlerts finds all triggered alerts for a system and sets them to inactive.
+// This is called when a system is paused or goes offline to prevent continued alerts.
+func deactivateAlerts(app core.App, systemID string) error {
+	// Note: Direct SQL updates don't trigger SSE, so we use the PocketBase API
+	// _, err := app.DB().NewQuery(fmt.Sprintf("UPDATE alerts SET triggered = false WHERE system = '%s'", systemID)).Execute()
+
+	alerts, err := app.FindRecordsByFilter("alerts", fmt.Sprintf("system = '%s' && triggered = 1", systemID), "", -1, 0)
+	if err != nil {
+		return err
+	}
+
+	for _, alert := range alerts {
+		alert.Set("triggered", false)
+		if err := app.SaveNoValidate(alert); err != nil {
+			return err
+		}
+	}
+	return nil
+}

beszel/internal/hub/systems/systems.go

@@ -1,435 +0,0 @@
-package systems
-
-import (
-	"beszel/internal/entities/system"
-	"context"
-	"fmt"
-	"net"
-	"strings"
-	"time"
-
-	"github.com/goccy/go-json"
-	"github.com/pocketbase/pocketbase/core"
-	"github.com/pocketbase/pocketbase/tools/store"
-	"golang.org/x/crypto/ssh"
-)
-
-const (
-	up      string = "up"
-	down    string = "down"
-	paused  string = "paused"
-	pending string = "pending"
-
-	interval int = 60_000
-
-	sessionTimeout = 4 * time.Second
-)
-
-type SystemManager struct {
-	hub       hubLike
-	systems   *store.Store[string, *System]
-	sshConfig *ssh.ClientConfig
-}
-
-type System struct {
-	Id      string `db:"id"`
-	Host    string `db:"host"`
-	Port    string `db:"port"`
-	Status  string `db:"status"`
-	manager *SystemManager
-	client  *ssh.Client
-	data    *system.CombinedData
-	ctx     context.Context
-	cancel  context.CancelFunc
-}
-
-type hubLike interface {
-	core.App
-	GetSSHKey() ([]byte, error)
-	HandleSystemAlerts(systemRecord *core.Record, data *system.CombinedData) error
-	HandleStatusAlerts(status string, systemRecord *core.Record) error
-}
-
-func NewSystemManager(hub hubLike) *SystemManager {
-	return &SystemManager{
-		systems: store.New(map[string]*System{}),
-		hub:     hub,
-	}
-}
-
-// Initialize initializes the system manager.
-// It binds the event hooks and starts updating existing systems.
-func (sm *SystemManager) Initialize() error {
-	sm.bindEventHooks()
-	// ssh setup
-	key, err := sm.hub.GetSSHKey()
-	if err != nil {
-		return err
-	}
-	if err := sm.createSSHClientConfig(key); err != nil {
-		return err
-	}
-	// start updating existing systems
-	var systems []*System
-	err = sm.hub.DB().NewQuery("SELECT id, host, port, status FROM systems WHERE status != 'paused'").All(&systems)
-	if err != nil || len(systems) == 0 {
-		return err
-	}
-	go func() {
-		// time between initial system updates
-		delta := interval / max(1, len(systems))
-		delta = min(delta, 2_000)
-		sleepTime := time.Duration(delta) * time.Millisecond
-		for _, system := range systems {
-			time.Sleep(sleepTime)
-			_ = sm.AddSystem(system)
-		}
-	}()
-	return nil
-}
-
-func (sm *SystemManager) bindEventHooks() {
-	sm.hub.OnRecordCreate("systems").BindFunc(sm.onRecordCreate)
-	sm.hub.OnRecordAfterCreateSuccess("systems").BindFunc(sm.onRecordAfterCreateSuccess)
-	sm.hub.OnRecordUpdate("systems").BindFunc(sm.onRecordUpdate)
-	sm.hub.OnRecordAfterUpdateSuccess("systems").BindFunc(sm.onRecordAfterUpdateSuccess)
-	sm.hub.OnRecordAfterDeleteSuccess("systems").BindFunc(sm.onRecordAfterDeleteSuccess)
-}
-
-// Runs before the record is committed to the database
-func (sm *SystemManager) onRecordCreate(e *core.RecordEvent) error {
-	e.Record.Set("info", system.Info{})
-	e.Record.Set("status", pending)
-	return e.Next()
-}
-
-// Runs after the record is committed to the database
-func (sm *SystemManager) onRecordAfterCreateSuccess(e *core.RecordEvent) error {
-	if err := sm.AddRecord(e.Record); err != nil {
-		e.App.Logger().Error("Error adding record", "err", err)
-	}
-	return e.Next()
-}
-
-// Runs before the record is updated
-func (sm *SystemManager) onRecordUpdate(e *core.RecordEvent) error {
-	if e.Record.GetString("status") == paused {
-		e.Record.Set("info", system.Info{})
-	}
-	return e.Next()
-}
-
-// Runs after the record is updated
-func (sm *SystemManager) onRecordAfterUpdateSuccess(e *core.RecordEvent) error {
-	newStatus := e.Record.GetString("status")
-	switch newStatus {
-	case paused:
-		sm.RemoveSystem(e.Record.Id)
-		return e.Next()
-	case pending:
-		if err := sm.AddRecord(e.Record); err != nil {
-			e.App.Logger().Error("Error adding record", "err", err)
-		}
-		return e.Next()
-	}
-	system, ok := sm.systems.GetOk(e.Record.Id)
-	if !ok {
-		return sm.AddRecord(e.Record)
-	}
-	prevStatus := system.Status
-	system.Status = newStatus
-	// system alerts if system is up
-	if system.Status == up {
-		if err := sm.hub.HandleSystemAlerts(e.Record, system.data); err != nil {
-			e.App.Logger().Error("Error handling system alerts", "err", err)
-		}
-	}
-	if (system.Status == down && prevStatus == up) || (system.Status == up && prevStatus == down) {
-		if err := sm.hub.HandleStatusAlerts(system.Status, e.Record); err != nil {
-			e.App.Logger().Error("Error handling status alerts", "err", err)
-		}
-	}
-	return e.Next()
-}
-
-// Runs after the record is deleted
-func (sm *SystemManager) onRecordAfterDeleteSuccess(e *core.RecordEvent) error {
-	sm.RemoveSystem(e.Record.Id)
-	return e.Next()
-}
-
-// AddSystem adds a system to the manager
-func (sm *SystemManager) AddSystem(sys *System) error {
-	if sm.systems.Has(sys.Id) {
-		return fmt.Errorf("system exists")
-	}
-	if sys.Id == "" || sys.Host == "" {
-		return fmt.Errorf("system is missing required fields")
-	}
-	sys.manager = sm
-	sys.ctx, sys.cancel = context.WithCancel(context.Background())
-	sys.data = &system.CombinedData{}
-	sm.systems.Set(sys.Id, sys)
-	go sys.StartUpdater()
-	return nil
-}
-
-// RemoveSystem removes a system from the manager
-func (sm *SystemManager) RemoveSystem(systemID string) error {
-	system, ok := sm.systems.GetOk(systemID)
-	if !ok {
-		return fmt.Errorf("system not found")
-	}
-	// cancel the context to signal stop
-	if system.cancel != nil {
-		system.cancel()
-	}
-	system.resetSSHClient()
-	sm.systems.Remove(systemID)
-	return nil
-}
-
-// AddRecord adds a record to the system manager.
-// It first removes any existing system with the same ID, then creates a new System
-// instance from the record data and adds it to the manager.
-// This function is typically called when a new system is created or when an existing
-// system's status changes to pending.
-func (sm *SystemManager) AddRecord(record *core.Record) (err error) {
-	_ = sm.RemoveSystem(record.Id)
-	system := &System{
-		Id:     record.Id,
-		Status: record.GetString("status"),
-		Host:   record.GetString("host"),
-		Port:   record.GetString("port"),
-	}
-	return sm.AddSystem(system)
-}
-
-// StartUpdater starts the system updater.
-// It first fetches the data from the agent then updates the records.
-// If the data is not found or the system is down, it sets the system down.
-func (sys *System) StartUpdater() {
-	if sys.data == nil {
-		sys.data = &system.CombinedData{}
-	}
-	if err := sys.update(); err != nil {
-		_ = sys.setDown(err)
-	}
-
-	c := time.Tick(time.Duration(interval) * time.Millisecond)
-
-	for {
-		select {
-		case <-sys.ctx.Done():
-			return
-		case <-c:
-			err := sys.update()
-			if err != nil {
-				_ = sys.setDown(err)
-			}
-		}
-	}
-}
-
-// update updates the system data and records.
-// It first fetches the data from the agent then updates the records.
-func (sys *System) update() error {
-	_, err := sys.fetchDataFromAgent()
-	if err == nil {
-		_, err = sys.createRecords()
-	}
-	return err
-}
-
-// createRecords updates the system record and adds system_stats and container_stats records
-func (sys *System) createRecords() (*core.Record, error) {
-	systemRecord, err := sys.getRecord()
-	if err != nil {
-		return nil, err
-	}
-	hub := sys.manager.hub
-	// add system_stats and container_stats records
-	systemStats, err := hub.FindCachedCollectionByNameOrId("system_stats")
-	if err != nil {
-		return nil, err
-	}
-	systemStatsRecord := core.NewRecord(systemStats)
-	systemStatsRecord.Set("system", systemRecord.Id)
-	systemStatsRecord.Set("stats", sys.data.Stats)
-	systemStatsRecord.Set("type", "1m")
-	if err := hub.SaveNoValidate(systemStatsRecord); err != nil {
-		return nil, err
-	}
-	// add new container_stats record
-	if len(sys.data.Containers) > 0 {
-		containerStats, err := hub.FindCachedCollectionByNameOrId("container_stats")
-		if err != nil {
-			return nil, err
-		}
-		containerStatsRecord := core.NewRecord(containerStats)
-		containerStatsRecord.Set("system", systemRecord.Id)
-		containerStatsRecord.Set("stats", sys.data.Containers)
-		containerStatsRecord.Set("type", "1m")
-		if err := hub.SaveNoValidate(containerStatsRecord); err != nil {
-			return nil, err
-		}
-	}
-	// update system record (do this last because it triggers alerts and we need above records to be inserted first)
-	systemRecord.Set("status", up)
-	systemRecord.Set("info", sys.data.Info)
-	if err := hub.SaveNoValidate(systemRecord); err != nil {
-		return nil, err
-	}
-	return systemRecord, nil
-}
-
-// getRecord retrieves the system record from the database.
-// If the record is not found or the system is paused, it removes the system from the manager.
-func (sys *System) getRecord() (*core.Record, error) {
-	record, err := sys.manager.hub.FindRecordById("systems", sys.Id)
-	if err != nil || record == nil {
-		_ = sys.manager.RemoveSystem(sys.Id)
-		return nil, err
-	}
-	return record, nil
-}
-
-// setDown marks a system as down in the database.
-// It takes the original error that caused the system to go down and returns any error
-// encountered during the process of updating the system status.
-func (sys *System) setDown(OriginalError error) error {
-	if sys.Status == down {
-		return nil
-	}
-	record, err := sys.getRecord()
-	if err != nil {
-		return err
-	}
-	sys.manager.hub.Logger().Error("System down", "system", record.GetString("name"), "err", OriginalError)
-	record.Set("status", down)
-	err = sys.manager.hub.SaveNoValidate(record)
-	if err != nil {
-		return err
-	}
-	return nil
-}
-
-// fetchDataFromAgent fetches the data from the agent.
-// It first creates a new SSH client if it doesn't exist or the system is down.
-// Then it creates a new SSH session and fetches the data from the agent.
-// If the data is not found or the system is down, it sets the system down.
-func (sys *System) fetchDataFromAgent() (*system.CombinedData, error) {
-	maxRetries := 1
-	for attempt := 0; attempt <= maxRetries; attempt++ {
-		if sys.client == nil || sys.Status == down {
-			if err := sys.createSSHClient(); err != nil {
-				return nil, err
-			}
-		}
-
-		session, err := sys.createSessionWithTimeout(4 * time.Second)
-		if err != nil {
-			if attempt >= maxRetries {
-				return nil, err
-			}
-			sys.manager.hub.Logger().Warn("Session closed. Retrying...", "host", sys.Host, "port", sys.Port, "err", err)
-			sys.resetSSHClient()
-			continue
-		}
-		defer session.Close()
-
-		stdout, err := session.StdoutPipe()
-		if err != nil {
-			return nil, err
-		}
-		if err := session.Shell(); err != nil {
-			return nil, err
-		}
-
-		// this is initialized in startUpdater, should never be nil
-		*sys.data = system.CombinedData{}
-		if err := json.NewDecoder(stdout).Decode(sys.data); err != nil {
-			return nil, err
-		}
-		// wait for the session to complete
-		if err := session.Wait(); err != nil {
-			return nil, err
-		}
-		return sys.data, nil
-	}
-
-	// this should never be reached due to the return in the loop
-	return nil, fmt.Errorf("failed to fetch data")
-}
-
-func (sm *SystemManager) createSSHClientConfig(key []byte) error {
-	signer, err := ssh.ParsePrivateKey(key)
-	if err != nil {
-		return err
-	}
-	sm.sshConfig = &ssh.ClientConfig{
-		User: "u",
-		Auth: []ssh.AuthMethod{
-			ssh.PublicKeys(signer),
-		},
-		HostKeyCallback: ssh.InsecureIgnoreHostKey(),
-		Timeout:         sessionTimeout,
-	}
-	return nil
-}
-
-// createSSHClient creates a new SSH client for the system
-func (s *System) createSSHClient() error {
-	network := "tcp"
-	host := s.Host
-	if strings.HasPrefix(host, "/") {
-		network = "unix"
-	} else {
-		host = net.JoinHostPort(host, s.Port)
-	}
-	var err error
-	s.client, err = ssh.Dial(network, host, s.manager.sshConfig)
-	if err != nil {
-		return err
-	}
-	return nil
-}
-
-// createSessionWithTimeout creates a new SSH session with a timeout to avoid hanging
-// in case of network issues
-func (sys *System) createSessionWithTimeout(timeout time.Duration) (*ssh.Session, error) {
-	if sys.client == nil {
-		return nil, fmt.Errorf("client not initialized")
-	}
-
-	ctx, cancel := context.WithTimeout(sys.ctx, timeout)
-	defer cancel()
-
-	sessionChan := make(chan *ssh.Session, 1)
-	errChan := make(chan error, 1)
-
-	go func() {
-		if session, err := sys.client.NewSession(); err != nil {
-			errChan <- err
-		} else {
-			sessionChan <- session
-		}
-	}()
-
-	select {
-	case session := <-sessionChan:
-		return session, nil
-	case err := <-errChan:
-		return nil, err
-	case <-ctx.Done():
-		return nil, fmt.Errorf("timeout")
-	}
-}
-
-// resetSSHClient closes the SSH connection and resets the client to nil
-func (sys *System) resetSSHClient() {
-	if sys.client != nil {
-		sys.client.Close()
-	}
-	sys.client = nil
-}

beszel/internal/hub/systems/systems_test.go

@@ -11,70 +11,135 @@ import (
 	"fmt"
 	"sync"
 	"testing"
+	"testing/synctest"
 	"time"
 
-	"github.com/pocketbase/dbx"
-	"github.com/pocketbase/pocketbase/core"
 	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"
 )
 
-// createTestSystem creates a test system record with a unique host name
+func TestSystemManagerNew(t *testing.T) {
-// and returns the created record and any error
+	hub, err := tests.NewTestHub(t.TempDir())
-func createTestSystem(t *testing.T, hub *tests.TestHub, options map[string]any) (*core.Record, error) {
-	collection, err := hub.FindCachedCollectionByNameOrId("systems")
-	if err != nil {
-		return nil, err
-	}
-
-	// get user record
-	var firstUser *core.Record
-	users, err := hub.FindAllRecords("users", dbx.NewExp("id != ''"))
-	if err != nil {
-		t.Fatal(err)
-	}
-	if len(users) > 0 {
-		firstUser = users[0]
-	}
-	// Generate a unique host name to ensure we're adding a new system
-	uniqueHost := fmt.Sprintf("test-host-%d.example.com", time.Now().UnixNano())
-
-	// Create the record
-	record := core.NewRecord(collection)
-	record.Set("name", uniqueHost)
-	record.Set("host", uniqueHost)
-	record.Set("port", "45876")
-	record.Set("status", "pending")
-	record.Set("users", []string{firstUser.Id})
-
-	// Apply any custom options
-	for key, value := range options {
-		record.Set(key, value)
-	}
-
-	// Save the record to the database
-	err = hub.Save(record)
-	if err != nil {
-		return nil, err
-	}
-
-	return record, nil
-}
-
-func TestSystemManagerIntegration(t *testing.T) {
-	// Create a test hub
-	hub, err := tests.NewTestHub()
 	if err != nil {
 		t.Fatal(err)
 	}
 	defer hub.Cleanup()
+	sm := hub.GetSystemManager()
 
-	// Create independent system manager
+	user, err := tests.CreateUser(hub, "test@test.com", "testtesttest")
-	sm := systems.NewSystemManager(hub)
+	require.NoError(t, err)
+
+	synctest.Test(t, func(t *testing.T) {
+		sm.Initialize()
+
+		record, err := tests.CreateRecord(hub, "systems", map[string]any{
+			"name":  "it-was-coney-island",
+			"host":  "the-playground-of-the-world",
+			"port":  "33914",
+			"users": []string{user.Id},
+		})
+		require.NoError(t, err)
+
+		assert.Equal(t, "pending", record.GetString("status"), "System status should be 'pending'")
+		assert.Equal(t, "pending", sm.GetSystemStatusFromStore(record.Id), "System status should be 'pending'")
+
+		// Verify the system host and port
+		host, port := sm.GetSystemHostPort(record.Id)
+		assert.Equal(t, record.GetString("host"), host, "System host should match")
+		assert.Equal(t, record.GetString("port"), port, "System port should match")
+
+		time.Sleep(13 * time.Second)
+		synctest.Wait()
+
+		assert.Equal(t, "pending", record.Fresh().GetString("status"), "System status should be 'pending'")
+		// Verify the system was added by checking if it exists
+		assert.True(t, sm.HasSystem(record.Id), "System should exist in the store")
+
+		time.Sleep(10 * time.Second)
+		synctest.Wait()
+
+		// system should be set to down after 15 seconds (no websocket connection)
+		assert.Equal(t, "down", sm.GetSystemStatusFromStore(record.Id), "System status should be 'down'")
+		// make sure the system is down in the db
+		record, err = hub.FindRecordById("systems", record.Id)
+		require.NoError(t, err)
+		assert.Equal(t, "down", record.GetString("status"), "System status should be 'down'")
+
+		assert.Equal(t, 1, sm.GetSystemCount(), "System count should be 1")
+
+		err = sm.RemoveSystem(record.Id)
+		assert.NoError(t, err)
+
+		assert.Equal(t, 0, sm.GetSystemCount(), "System count should be 0")
+		assert.False(t, sm.HasSystem(record.Id), "System should not exist in the store after removal")
+
+		// let's also make sure a system is removed from the store when the record is deleted
+		record, err = tests.CreateRecord(hub, "systems", map[string]any{
+			"name":  "there-was-no-place-like-it",
+			"host":  "in-the-whole-world",
+			"port":  "33914",
+			"users": []string{user.Id},
+		})
+		require.NoError(t, err)
+
+		assert.True(t, sm.HasSystem(record.Id), "System should exist in the store after creation")
+
+		time.Sleep(8 * time.Second)
+		synctest.Wait()
+		assert.Equal(t, "pending", sm.GetSystemStatusFromStore(record.Id), "System status should be 'pending'")
+
+		sm.SetSystemStatusInDB(record.Id, "up")
+		time.Sleep(time.Second)
+		synctest.Wait()
+		assert.Equal(t, "up", sm.GetSystemStatusFromStore(record.Id), "System status should be 'up'")
+
+		// make sure the system switches to down after 11 seconds
+		sm.RemoveSystem(record.Id)
+		sm.AddRecord(record, nil)
+		assert.Equal(t, "pending", sm.GetSystemStatusFromStore(record.Id), "System status should be 'pending'")
+		time.Sleep(12 * time.Second)
+		synctest.Wait()
+		assert.Equal(t, "down", sm.GetSystemStatusFromStore(record.Id), "System status should be 'down'")
+
+		// sm.SetSystemStatusInDB(record.Id, "paused")
+		// time.Sleep(time.Second)
+		// synctest.Wait()
+		// assert.Equal(t, "paused", sm.GetSystemStatusFromStore(record.Id), "System status should be 'paused'")
+
+		// delete the record
+		err = hub.Delete(record)
+		require.NoError(t, err)
+		assert.False(t, sm.HasSystem(record.Id), "System should not exist in the store after deletion")
+	})
+
+	testOld(t, hub)
+
+	synctest.Test(t, func(t *testing.T) {
+		time.Sleep(time.Second)
+		synctest.Wait()
+
+		for _, systemId := range sm.GetAllSystemIDs() {
+			err = sm.RemoveSystem(systemId)
+			require.NoError(t, err)
+			assert.False(t, sm.HasSystem(systemId), "System should not exist in the store after deletion")
+		}
+
+		assert.Equal(t, 0, sm.GetSystemCount(), "System count should be 0")
+
+		// TODO: test with websocket client
+	})
+}
+
+func testOld(t *testing.T, hub *tests.TestHub) {
+	user, err := tests.CreateUser(hub, "test@testy.com", "testtesttest")
+	require.NoError(t, err)
+
+	sm := hub.GetSystemManager()
 	assert.NotNil(t, sm)
 
-	// Test initialization
+	// error expected when creating a user with a duplicate email
-	sm.Initialize()
+	_, err = tests.CreateUser(hub, "test@test.com", "testtesttest")
+	require.Error(t, err)
 
 	// Test collection existence. todo: move to hub package tests
 	t.Run("CollectionExistence", func(t *testing.T) {
@@ -92,81 +157,17 @@ func TestSystemManagerIntegration(t *testing.T) {
 		assert.NotNil(t, containerStats)
 	})
 
-	// Test adding a system record
-	t.Run("AddRecord", func(t *testing.T) {
-		var wg sync.WaitGroup
-		wg.Add(2)
-
-		// Get the count before adding the system
-		countBefore := sm.GetSystemCount()
-
-		// record should be pending on create
-		hub.OnRecordCreate("systems").BindFunc(func(e *core.RecordEvent) error {
-			record := e.Record
-			if record.GetString("name") == "welcometoarcoampm" {
-				assert.Equal(t, "pending", e.Record.GetString("status"), "System status should be 'pending'")
-				wg.Done()
-			}
-			return e.Next()
-		})
-
-		// record should be down on update
-		hub.OnRecordAfterUpdateSuccess("systems").BindFunc(func(e *core.RecordEvent) error {
-			record := e.Record
-			if record.GetString("name") == "welcometoarcoampm" {
-				assert.Equal(t, "down", e.Record.GetString("status"), "System status should be 'pending'")
-				wg.Done()
-			}
-			return e.Next()
-		})
-		// Create a test system with the first user assigned
-		record, err := createTestSystem(t, hub, map[string]any{
-			"name": "welcometoarcoampm",
-			"host": "localhost",
-			"port": "33914",
-		})
-		require.NoError(t, err)
-
-		wg.Wait()
-
-		// system should be down if grabbed from the store
-		assert.Equal(t, "down", sm.GetSystemStatusFromStore(record.Id), "System status should be 'down'")
-
-		// Check that the system count increased
-		countAfter := sm.GetSystemCount()
-		assert.Equal(t, countBefore+1, countAfter, "System count should increase after adding a system via event hook")
-
-		// Verify the system was added by checking if it exists
-		assert.True(t, sm.HasSystem(record.Id), "System should exist in the store")
-
-		// Verify the system host and port
-		host, port := sm.GetSystemHostPort(record.Id)
-		assert.Equal(t, record.Get("host"), host, "System host should match")
-		assert.Equal(t, record.Get("port"), port, "System port should match")
-
-		// Verify the system is in the list of all system IDs
-		ids := sm.GetAllSystemIDs()
-		assert.Contains(t, ids, record.Id, "System ID should be in the list of all system IDs")
-
-		// Verify the system was added by checking if removing it works
-		err = sm.RemoveSystem(record.Id)
-		assert.NoError(t, err, "System should exist and be removable")
-
-		// Verify the system no longer exists
-		assert.False(t, sm.HasSystem(record.Id), "System should not exist in the store after removal")
-
-		// Verify the system is not in the list of all system IDs
-		newIds := sm.GetAllSystemIDs()
-		assert.NotContains(t, newIds, record.Id, "System ID should not be in the list of all system IDs after removal")
-
-	})
-
 	t.Run("RemoveSystem", func(t *testing.T) {
 		// Get the count before adding the system
 		countBefore := sm.GetSystemCount()
 
 		// Create a test system record
-		record, err := createTestSystem(t, hub, map[string]any{})
+		record, err := tests.CreateRecord(hub, "systems", map[string]any{
+			"name":  "i-even-got-lost-at-coney-island",
+			"host":  "but-they-found-me",
+			"port":  "33914",
+			"users": []string{user.Id},
+		})
 		require.NoError(t, err)
 
 		// Verify the system count increased
@@ -202,11 +203,16 @@ func TestSystemManagerIntegration(t *testing.T) {
 
 	t.Run("NewRecordPending", func(t *testing.T) {
 		// Create a test system
-		record, err := createTestSystem(t, hub, map[string]any{})
+		record, err := tests.CreateRecord(hub, "systems", map[string]any{
+			"name":  "and-you-know",
+			"host":  "i-feel-very-bad",
+			"port":  "33914",
+			"users": []string{user.Id},
+		})
 		require.NoError(t, err)
 
 		// Add the record to the system manager
-		err = sm.AddRecord(record)
+		err = sm.AddRecord(record, nil)
 		require.NoError(t, err)
 
 		// Test filtering records by status - should be "pending" now
@@ -218,11 +224,16 @@ func TestSystemManagerIntegration(t *testing.T) {
 
 	t.Run("SystemStatusUpdate", func(t *testing.T) {
 		// Create a test system record
-		record, err := createTestSystem(t, hub, map[string]any{})
+		record, err := tests.CreateRecord(hub, "systems", map[string]any{
+			"name":  "we-used-to-sleep-on-the-beach",
+			"host":  "sleep-overnight-here",
+			"port":  "33914",
+			"users": []string{user.Id},
+		})
 		require.NoError(t, err)
 
 		// Add the record to the system manager
-		err = sm.AddRecord(record)
+		err = sm.AddRecord(record, nil)
 		require.NoError(t, err)
 
 		// Test status changes
@@ -244,7 +255,12 @@ func TestSystemManagerIntegration(t *testing.T) {
 
 	t.Run("HandleSystemData", func(t *testing.T) {
 		// Create a test system record
-		record, err := createTestSystem(t, hub, map[string]any{})
+		record, err := tests.CreateRecord(hub, "systems", map[string]any{
+			"name":  "things-changed-you-know",
+			"host":  "they-dont-sleep-anymore-on-the-beach",
+			"port":  "33914",
+			"users": []string{user.Id},
+		})
 		require.NoError(t, err)
 
 		// Create test system data
@@ -295,54 +311,14 @@ func TestSystemManagerIntegration(t *testing.T) {
 		assert.Error(t, err)
 	})
 
-	t.Run("DeleteRecord", func(t *testing.T) {
-		var wg sync.WaitGroup
-		wg.Add(2)
-
-		runs := 0
-
-		hub.OnRecordUpdate("systems").BindFunc(func(e *core.RecordEvent) error {
-			runs++
-			record := e.Record
-			if record.GetString("name") == "deadflagblues" {
-				if runs == 1 {
-					assert.Equal(t, "up", e.Record.GetString("status"), "System status should be 'up'")
-					wg.Done()
-				} else if runs == 2 {
-					assert.Equal(t, "paused", e.Record.GetString("status"), "System status should be 'paused'")
-					wg.Done()
-				}
-			}
-			return e.Next()
-		})
-
-		// Create a test system record
-		record, err := createTestSystem(t, hub, map[string]any{
-			"name": "deadflagblues",
-		})
-		require.NoError(t, err)
-
-		// Verify the system exists
-		assert.True(t, sm.HasSystem(record.Id), "System should exist in the store")
-
-		// set the status manually to up
-		sm.SetSystemStatusInDB(record.Id, "up")
-
-		// verify the status is up
-		assert.Equal(t, "up", sm.GetSystemStatusFromStore(record.Id), "System status should be 'up'")
-
-		// Set the status to "paused" which should cause it to be deleted from the store
-		sm.SetSystemStatusInDB(record.Id, "paused")
-
-		wg.Wait()
-
-		// Verify the system no longer exists
-		assert.False(t, sm.HasSystem(record.Id), "System should not exist in the store after deletion")
-	})
-
 	t.Run("ConcurrentOperations", func(t *testing.T) {
 		// Create a test system
-		record, err := createTestSystem(t, hub, map[string]any{})
+		record, err := tests.CreateRecord(hub, "systems", map[string]any{
+			"name":  "jfkjahkfajs",
+			"host":  "localhost",
+			"port":  "33914",
+			"users": []string{user.Id},
+		})
 		require.NoError(t, err)
 
 		// Run concurrent operations
@@ -377,7 +353,12 @@ func TestSystemManagerIntegration(t *testing.T) {
 
 	t.Run("ContextCancellation", func(t *testing.T) {
 		// Create a test system record
-		record, err := createTestSystem(t, hub, map[string]any{})
+		record, err := tests.CreateRecord(hub, "systems", map[string]any{
+			"name":  "lkhsdfsjf",
+			"host":  "localhost",
+			"port":  "33914",
+			"users": []string{user.Id},
+		})
 		require.NoError(t, err)
 
 		// Verify the system exists in the store
@@ -420,7 +401,7 @@ func TestSystemManagerIntegration(t *testing.T) {
 		assert.Error(t, err, "RemoveSystem should fail for non-existent system")
 
 		// Add the system back
-		err = sm.AddRecord(record)
+		err = sm.AddRecord(record, nil)
 		require.NoError(t, err, "AddRecord should succeed")
 
 		// Verify the system is back in the store

beszel/internal/hub/systems/systems_test_helpers.go

@@ -9,17 +9,17 @@ import (
 	"fmt"
 )
 
-// GetSystemCount returns the number of systems in the store
+// TESTING ONLY: GetSystemCount returns the number of systems in the store
 func (sm *SystemManager) GetSystemCount() int {
 	return sm.systems.Length()
 }
 
-// HasSystem checks if a system with the given ID exists in the store
+// TESTING ONLY: HasSystem checks if a system with the given ID exists in the store
 func (sm *SystemManager) HasSystem(systemID string) bool {
 	return sm.systems.Has(systemID)
 }
 
-// GetSystemStatusFromStore returns the status of a system with the given ID
+// TESTING ONLY: GetSystemStatusFromStore returns the status of a system with the given ID
 // Returns an empty string if the system doesn't exist
 func (sm *SystemManager) GetSystemStatusFromStore(systemID string) string {
 	sys, ok := sm.systems.GetOk(systemID)
@@ -29,7 +29,7 @@ func (sm *SystemManager) GetSystemStatusFromStore(systemID string) string {
 	return sys.Status
 }
 
-// GetSystemContextFromStore returns the context and cancel function for a system
+// TESTING ONLY: GetSystemContextFromStore returns the context and cancel function for a system
 func (sm *SystemManager) GetSystemContextFromStore(systemID string) (context.Context, context.CancelFunc, error) {
 	sys, ok := sm.systems.GetOk(systemID)
 	if !ok {
@@ -38,7 +38,7 @@ func (sm *SystemManager) GetSystemContextFromStore(systemID string) (context.Con
 	return sys.ctx, sys.cancel, nil
 }
 
-// GetSystemFromStore returns a store from the system
+// TESTING ONLY: GetSystemFromStore returns a store from the system
 func (sm *SystemManager) GetSystemFromStore(systemID string) (*System, error) {
 	sys, ok := sm.systems.GetOk(systemID)
 	if !ok {
@@ -47,7 +47,7 @@ func (sm *SystemManager) GetSystemFromStore(systemID string) (*System, error) {
 	return sys, nil
 }
 
-// GetAllSystemIDs returns a slice of all system IDs in the store
+// TESTING ONLY: GetAllSystemIDs returns a slice of all system IDs in the store
 func (sm *SystemManager) GetAllSystemIDs() []string {
 	data := sm.systems.GetAll()
 	ids := make([]string, 0, len(data))
@@ -57,7 +57,7 @@ func (sm *SystemManager) GetAllSystemIDs() []string {
 	return ids
 }
 
-// GetSystemData returns the combined data for a system with the given ID
+// TESTING ONLY: GetSystemData returns the combined data for a system with the given ID
 // Returns nil if the system doesn't exist
 // This method is intended for testing
 func (sm *SystemManager) GetSystemData(systemID string) *entities.CombinedData {
@@ -68,7 +68,7 @@ func (sm *SystemManager) GetSystemData(systemID string) *entities.CombinedData {
 	return sys.data
 }
 
-// GetSystemHostPort returns the host and port for a system with the given ID
+// TESTING ONLY: GetSystemHostPort returns the host and port for a system with the given ID
 // Returns empty strings if the system doesn't exist
 func (sm *SystemManager) GetSystemHostPort(systemID string) (string, string) {
 	sys, ok := sm.systems.GetOk(systemID)
@@ -78,22 +78,7 @@ func (sm *SystemManager) GetSystemHostPort(systemID string) (string, string) {
 	return sys.Host, sys.Port
 }
 
-// DisableAutoUpdater disables the automatic updater for a system
+// TESTING ONLY: SetSystemStatusInDB sets the status of a system directly and updates the database record
-// This is intended for testing
-// Returns false if the system doesn't exist
-// func (sm *SystemManager) DisableAutoUpdater(systemID string) bool {
-// 	sys, ok := sm.systems.GetOk(systemID)
-// 	if !ok {
-// 		return false
-// 	}
-// 	if sys.cancel != nil {
-// 		sys.cancel()
-// 		sys.cancel = nil
-// 	}
-// 	return true
-// }
-
-// SetSystemStatusInDB sets the status of a system directly and updates the database record
 // This is intended for testing
 // Returns false if the system doesn't exist
 func (sm *SystemManager) SetSystemStatusInDB(systemID string, status string) bool {

beszel/internal/hub/update.go

@@ -1,57 +1,85 @@
 package hub
 
 import (
-	"beszel"
+	"beszel/internal/ghupdate"
 	"fmt"
+	"log"
 	"os"
-	"strings"
+	"os/exec"
 
-	"github.com/blang/semver"
-	"github.com/rhysd/go-github-selfupdate/selfupdate"
 	"github.com/spf13/cobra"
 )
 
 // Update updates beszel to the latest version
-func Update(_ *cobra.Command, _ []string) {
+func Update(cmd *cobra.Command, _ []string) {
-	var latest *selfupdate.Release
+	dataDir := os.TempDir()
-	var found bool
+
-	var err error
+	// set dataDir to ./beszel_data if it exists
-	currentVersion := semver.MustParse(beszel.Version)
+	if _, err := os.Stat("./beszel_data"); err == nil {
-	fmt.Println("beszel", currentVersion)
+		dataDir = "./beszel_data"
-	fmt.Println("Checking for updates...")
+	}
-	updater, _ := selfupdate.NewUpdater(selfupdate.Config{
+
-		Filters: []string{"beszel_"},
+	// Check if china-mirrors flag is set
+	useMirror, _ := cmd.Flags().GetBool("china-mirrors")
+
+	updated, err := ghupdate.Update(ghupdate.Config{
+		ArchiveExecutable: "beszel",
+		DataDir:           dataDir,
+		UseMirror:         useMirror,
 	})
-	latest, found, err = updater.DetectLatest("henrygd/beszel")
-
 	if err != nil {
-		fmt.Println("Error checking for updates:", err)
+		log.Fatal(err)
-		os.Exit(1)
 	}
-
+	if !updated {
-	if !found {
-		fmt.Println("No updates found")
-		os.Exit(0)
-	}
-
-	fmt.Println("Latest version:", latest.Version)
-
-	if latest.Version.LTE(currentVersion) {
-		fmt.Println("You are up to date")
 		return
 	}
 
-	var binaryPath string
+	// make sure the file is executable
-	fmt.Printf("Updating from %s to %s...\n", currentVersion, latest.Version)
+	exePath, err := os.Executable()
-	binaryPath, err = os.Executable()
+	if err == nil {
-	if err != nil {
+		if err := os.Chmod(exePath, 0755); err != nil {
-		fmt.Println("Error getting binary path:", err)
+			fmt.Printf("Warning: failed to set executable permissions: %v\n", err)
-		os.Exit(1)
+		}
 	}
-	err = selfupdate.UpdateTo(latest.AssetURL, binaryPath)
+
-	if err != nil {
+	// Try to restart the service if it's running
-		fmt.Println("Please try rerunning with sudo. Error:", err)
+	restartService()
-		os.Exit(1)
+}
-	}
+
-	fmt.Printf("Successfully updated to %s\n\n%s\n", latest.Version, strings.TrimSpace(latest.ReleaseNotes))
+// restartService attempts to restart the beszel service
+func restartService() {
+	// Check if we're running as a service by looking for systemd
+	if _, err := exec.LookPath("systemctl"); err == nil {
+		// Check if beszel service exists and is active
+		cmd := exec.Command("systemctl", "is-active", "beszel.service")
+		if err := cmd.Run(); err == nil {
+			ghupdate.ColorPrint(ghupdate.ColorYellow, "Restarting beszel service...")
+			restartCmd := exec.Command("systemctl", "restart", "beszel.service")
+			if err := restartCmd.Run(); err != nil {
+				ghupdate.ColorPrintf(ghupdate.ColorYellow, "Warning: Failed to restart service: %v\n", err)
+				ghupdate.ColorPrint(ghupdate.ColorYellow, "Please restart the service manually: sudo systemctl restart beszel")
+			} else {
+				ghupdate.ColorPrint(ghupdate.ColorGreen, "Service restarted successfully")
+			}
+			return
+		}
+	}
+
+	// Check for OpenRC (Alpine Linux)
+	if _, err := exec.LookPath("rc-service"); err == nil {
+		cmd := exec.Command("rc-service", "beszel", "status")
+		if err := cmd.Run(); err == nil {
+			ghupdate.ColorPrint(ghupdate.ColorYellow, "Restarting beszel service...")
+			restartCmd := exec.Command("rc-service", "beszel", "restart")
+			if err := restartCmd.Run(); err != nil {
+				ghupdate.ColorPrintf(ghupdate.ColorYellow, "Warning: Failed to restart service: %v\n", err)
+				ghupdate.ColorPrint(ghupdate.ColorYellow, "Please restart the service manually: sudo rc-service beszel restart")
+			} else {
+				ghupdate.ColorPrint(ghupdate.ColorGreen, "Service restarted successfully")
+			}
+			return
+		}
+	}
+
+	ghupdate.ColorPrint(ghupdate.ColorYellow, "Service restart not attempted. If running as a service, restart manually.")
 }

beszel/internal/hub/ws/ws.go

@@ -0,0 +1,180 @@
+package ws
+
+import (
+	"beszel/internal/common"
+	"beszel/internal/entities/system"
+	"errors"
+	"time"
+	"weak"
+
+	"github.com/fxamacker/cbor/v2"
+	"github.com/lxzan/gws"
+	"golang.org/x/crypto/ssh"
+)
+
+const (
+	deadline = 70 * time.Second
+)
+
+// Handler implements the WebSocket event handler for agent connections.
+type Handler struct {
+	gws.BuiltinEventHandler
+}
+
+// WsConn represents a WebSocket connection to an agent.
+type WsConn struct {
+	conn         *gws.Conn
+	responseChan chan *gws.Message
+	DownChan     chan struct{}
+}
+
+// FingerprintRecord is fingerprints collection record data in the hub
+type FingerprintRecord struct {
+	Id          string `db:"id"`
+	SystemId    string `db:"system"`
+	Fingerprint string `db:"fingerprint"`
+	Token       string `db:"token"`
+}
+
+var upgrader *gws.Upgrader
+
+// GetUpgrader returns a singleton WebSocket upgrader instance.
+func GetUpgrader() *gws.Upgrader {
+	if upgrader != nil {
+		return upgrader
+	}
+	handler := &Handler{}
+	upgrader = gws.NewUpgrader(handler, &gws.ServerOption{})
+	return upgrader
+}
+
+// NewWsConnection creates a new WebSocket connection wrapper.
+func NewWsConnection(conn *gws.Conn) *WsConn {
+	return &WsConn{
+		conn:         conn,
+		responseChan: make(chan *gws.Message, 1),
+		DownChan:     make(chan struct{}, 1),
+	}
+}
+
+// OnOpen sets a deadline for the WebSocket connection.
+func (h *Handler) OnOpen(conn *gws.Conn) {
+	conn.SetDeadline(time.Now().Add(deadline))
+}
+
+// OnMessage routes incoming WebSocket messages to the response channel.
+func (h *Handler) OnMessage(conn *gws.Conn, message *gws.Message) {
+	conn.SetDeadline(time.Now().Add(deadline))
+	if message.Opcode != gws.OpcodeBinary || message.Data.Len() == 0 {
+		return
+	}
+	wsConn, ok := conn.Session().Load("wsConn")
+	if !ok {
+		_ = conn.WriteClose(1000, nil)
+		return
+	}
+	select {
+	case wsConn.(*WsConn).responseChan <- message:
+	default:
+		// close if the connection is not expecting a response
+		wsConn.(*WsConn).Close(nil)
+	}
+}
+
+// OnClose handles WebSocket connection closures and triggers system down status after delay.
+func (h *Handler) OnClose(conn *gws.Conn, err error) {
+	wsConn, ok := conn.Session().Load("wsConn")
+	if !ok {
+		return
+	}
+	wsConn.(*WsConn).conn = nil
+	// wait 5 seconds to allow reconnection before setting system down
+	// use a weak pointer to avoid keeping references if the system is removed
+	go func(downChan weak.Pointer[chan struct{}]) {
+		time.Sleep(5 * time.Second)
+		downChanValue := downChan.Value()
+		if downChanValue != nil {
+			*downChanValue <- struct{}{}
+		}
+	}(weak.Make(&wsConn.(*WsConn).DownChan))
+}
+
+// Close terminates the WebSocket connection gracefully.
+func (ws *WsConn) Close(msg []byte) {
+	if ws.IsConnected() {
+		ws.conn.WriteClose(1000, msg)
+	}
+}
+
+// Ping sends a ping frame to keep the connection alive.
+func (ws *WsConn) Ping() error {
+	ws.conn.SetDeadline(time.Now().Add(deadline))
+	return ws.conn.WritePing(nil)
+}
+
+// sendMessage encodes data to CBOR and sends it as a binary message to the agent.
+func (ws *WsConn) sendMessage(data common.HubRequest[any]) error {
+	if ws.conn == nil {
+		return gws.ErrConnClosed
+	}
+	bytes, err := cbor.Marshal(data)
+	if err != nil {
+		return err
+	}
+	return ws.conn.WriteMessage(gws.OpcodeBinary, bytes)
+}
+
+// RequestSystemData requests system metrics from the agent and unmarshals the response.
+func (ws *WsConn) RequestSystemData(data *system.CombinedData) error {
+	var message *gws.Message
+
+	ws.sendMessage(common.HubRequest[any]{
+		Action: common.GetData,
+	})
+	select {
+	case <-time.After(10 * time.Second):
+		ws.Close(nil)
+		return gws.ErrConnClosed
+	case message = <-ws.responseChan:
+	}
+	defer message.Close()
+	return cbor.Unmarshal(message.Data.Bytes(), data)
+}
+
+// GetFingerprint authenticates with the agent using SSH signature and returns the agent's fingerprint.
+func (ws *WsConn) GetFingerprint(token string, signer ssh.Signer, needSysInfo bool) (common.FingerprintResponse, error) {
+	var clientFingerprint common.FingerprintResponse
+	challenge := []byte(token)
+
+	signature, err := signer.Sign(nil, challenge)
+	if err != nil {
+		return clientFingerprint, err
+	}
+
+	err = ws.sendMessage(common.HubRequest[any]{
+		Action: common.CheckFingerprint,
+		Data: common.FingerprintRequest{
+			Signature:   signature.Blob,
+			NeedSysInfo: needSysInfo,
+		},
+	})
+	if err != nil {
+		return clientFingerprint, err
+	}
+
+	var message *gws.Message
+	select {
+	case message = <-ws.responseChan:
+	case <-time.After(10 * time.Second):
+		return clientFingerprint, errors.New("request expired")
+	}
+	defer message.Close()
+
+	err = cbor.Unmarshal(message.Data.Bytes(), &clientFingerprint)
+	return clientFingerprint, err
+}
+
+// IsConnected returns true if the WebSocket connection is active.
+func (ws *WsConn) IsConnected() bool {
+	return ws.conn != nil
+}

beszel/internal/hub/ws/ws_test.go

@@ -0,0 +1,221 @@
+//go:build testing
+// +build testing
+
+package ws
+
+import (
+	"beszel/internal/common"
+	"crypto/ed25519"
+	"testing"
+	"time"
+
+	"github.com/fxamacker/cbor/v2"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+	"golang.org/x/crypto/ssh"
+)
+
+// TestGetUpgrader tests the singleton upgrader
+func TestGetUpgrader(t *testing.T) {
+	// Reset the global upgrader to test singleton behavior
+	upgrader = nil
+
+	// First call should create the upgrader
+	upgrader1 := GetUpgrader()
+	assert.NotNil(t, upgrader1, "Upgrader should not be nil")
+
+	// Second call should return the same instance
+	upgrader2 := GetUpgrader()
+	assert.Same(t, upgrader1, upgrader2, "Should return the same upgrader instance")
+
+	// Verify it's properly configured
+	assert.NotNil(t, upgrader1, "Upgrader should be configured")
+}
+
+// TestNewWsConnection tests WebSocket connection creation
+func TestNewWsConnection(t *testing.T) {
+	// We can't easily mock gws.Conn, so we'll pass nil and test the structure
+	wsConn := NewWsConnection(nil)
+
+	assert.NotNil(t, wsConn, "WebSocket connection should not be nil")
+	assert.Nil(t, wsConn.conn, "Connection should be nil as passed")
+	assert.NotNil(t, wsConn.responseChan, "Response channel should be initialized")
+	assert.NotNil(t, wsConn.DownChan, "Down channel should be initialized")
+	assert.Equal(t, 1, cap(wsConn.responseChan), "Response channel should have capacity of 1")
+	assert.Equal(t, 1, cap(wsConn.DownChan), "Down channel should have capacity of 1")
+}
+
+// TestWsConn_IsConnected tests the connection status check
+func TestWsConn_IsConnected(t *testing.T) {
+	// Test with nil connection
+	wsConn := NewWsConnection(nil)
+	assert.False(t, wsConn.IsConnected(), "Should not be connected when conn is nil")
+}
+
+// TestWsConn_Close tests the connection closing with nil connection
+func TestWsConn_Close(t *testing.T) {
+	wsConn := NewWsConnection(nil)
+
+	// Should handle nil connection gracefully
+	assert.NotPanics(t, func() {
+		wsConn.Close([]byte("test message"))
+	}, "Should not panic when closing nil connection")
+}
+
+// TestWsConn_SendMessage_CBOR tests CBOR encoding in sendMessage
+func TestWsConn_SendMessage_CBOR(t *testing.T) {
+	wsConn := NewWsConnection(nil)
+
+	testData := common.HubRequest[any]{
+		Action: common.GetData,
+		Data:   "test data",
+	}
+
+	// This will fail because conn is nil, but we can test the CBOR encoding logic
+	// by checking that the function properly encodes to CBOR before failing
+	err := wsConn.sendMessage(testData)
+	assert.Error(t, err, "Should error with nil connection")
+
+	// Test CBOR encoding separately
+	bytes, err := cbor.Marshal(testData)
+	assert.NoError(t, err, "Should encode to CBOR successfully")
+
+	// Verify we can decode it back
+	var decodedData common.HubRequest[any]
+	err = cbor.Unmarshal(bytes, &decodedData)
+	assert.NoError(t, err, "Should decode from CBOR successfully")
+	assert.Equal(t, testData.Action, decodedData.Action, "Action should match")
+}
+
+// TestWsConn_GetFingerprint_SignatureGeneration tests signature creation logic
+func TestWsConn_GetFingerprint_SignatureGeneration(t *testing.T) {
+	// Generate test key pair
+	_, privKey, err := ed25519.GenerateKey(nil)
+	require.NoError(t, err)
+
+	signer, err := ssh.NewSignerFromKey(privKey)
+	require.NoError(t, err)
+
+	token := "test-token"
+
+	// This will timeout since conn is nil, but we can verify the signature logic
+	// We can't test the full flow, but we can test that the signature is created properly
+	challenge := []byte(token)
+	signature, err := signer.Sign(nil, challenge)
+	assert.NoError(t, err, "Should create signature successfully")
+	assert.NotEmpty(t, signature.Blob, "Signature blob should not be empty")
+	assert.Equal(t, signer.PublicKey().Type(), signature.Format, "Signature format should match key type")
+
+	// Test the fingerprint request structure
+	fpRequest := common.FingerprintRequest{
+		Signature:   signature.Blob,
+		NeedSysInfo: true,
+	}
+
+	// Test CBOR encoding of fingerprint request
+	fpData, err := cbor.Marshal(fpRequest)
+	assert.NoError(t, err, "Should encode fingerprint request to CBOR")
+
+	var decodedFpRequest common.FingerprintRequest
+	err = cbor.Unmarshal(fpData, &decodedFpRequest)
+	assert.NoError(t, err, "Should decode fingerprint request from CBOR")
+	assert.Equal(t, fpRequest.Signature, decodedFpRequest.Signature, "Signature should match")
+	assert.Equal(t, fpRequest.NeedSysInfo, decodedFpRequest.NeedSysInfo, "NeedSysInfo should match")
+
+	// Test the full hub request structure
+	hubRequest := common.HubRequest[any]{
+		Action: common.CheckFingerprint,
+		Data:   fpRequest,
+	}
+
+	hubData, err := cbor.Marshal(hubRequest)
+	assert.NoError(t, err, "Should encode hub request to CBOR")
+
+	var decodedHubRequest common.HubRequest[cbor.RawMessage]
+	err = cbor.Unmarshal(hubData, &decodedHubRequest)
+	assert.NoError(t, err, "Should decode hub request from CBOR")
+	assert.Equal(t, common.CheckFingerprint, decodedHubRequest.Action, "Action should be CheckFingerprint")
+}
+
+// TestWsConn_RequestSystemData_RequestFormat tests system data request format
+func TestWsConn_RequestSystemData_RequestFormat(t *testing.T) {
+	// Test the request format that would be sent
+	request := common.HubRequest[any]{
+		Action: common.GetData,
+	}
+
+	// Test CBOR encoding
+	data, err := cbor.Marshal(request)
+	assert.NoError(t, err, "Should encode request to CBOR")
+
+	// Test decoding
+	var decodedRequest common.HubRequest[any]
+	err = cbor.Unmarshal(data, &decodedRequest)
+	assert.NoError(t, err, "Should decode request from CBOR")
+	assert.Equal(t, common.GetData, decodedRequest.Action, "Should have GetData action")
+}
+
+// TestFingerprintRecord tests the FingerprintRecord struct
+func TestFingerprintRecord(t *testing.T) {
+	record := FingerprintRecord{
+		Id:          "test-id",
+		SystemId:    "system-123",
+		Fingerprint: "test-fingerprint",
+		Token:       "test-token",
+	}
+
+	assert.Equal(t, "test-id", record.Id)
+	assert.Equal(t, "system-123", record.SystemId)
+	assert.Equal(t, "test-fingerprint", record.Fingerprint)
+	assert.Equal(t, "test-token", record.Token)
+}
+
+// TestDeadlineConstant tests that the deadline constant is reasonable
+func TestDeadlineConstant(t *testing.T) {
+	assert.Equal(t, 70*time.Second, deadline, "Deadline should be 70 seconds")
+}
+
+// TestCommonActions tests that the common actions are properly defined
+func TestCommonActions(t *testing.T) {
+	// Test that the actions we use exist and have expected values
+	assert.Equal(t, common.WebSocketAction(0), common.GetData, "GetData should be action 0")
+	assert.Equal(t, common.WebSocketAction(1), common.CheckFingerprint, "CheckFingerprint should be action 1")
+}
+
+// TestHandler tests that we can create a Handler
+func TestHandler(t *testing.T) {
+	handler := &Handler{}
+	assert.NotNil(t, handler, "Handler should be created successfully")
+
+	// The Handler embeds gws.BuiltinEventHandler, so it should have the embedded type
+	assert.NotNil(t, handler.BuiltinEventHandler, "Should have embedded BuiltinEventHandler")
+}
+
+// TestWsConnChannelBehavior tests channel behavior without WebSocket connections
+func TestWsConnChannelBehavior(t *testing.T) {
+	wsConn := NewWsConnection(nil)
+
+	// Test that channels are properly initialized and can be used
+	select {
+	case wsConn.DownChan <- struct{}{}:
+		// Should be able to write to channel
+	default:
+		t.Error("Should be able to write to DownChan")
+	}
+
+	// Test reading from DownChan
+	select {
+	case <-wsConn.DownChan:
+		// Should be able to read from channel
+	case <-time.After(10 * time.Millisecond):
+		t.Error("Should be able to read from DownChan")
+	}
+
+	// Response channel should be empty initially
+	select {
+	case <-wsConn.responseChan:
+		t.Error("Response channel should be empty initially")
+	default:
+		// Expected - channel should be empty
+	}
+}

beszel/internal/records/records.go

@@ -4,13 +4,13 @@ package records
 import (
 	"beszel/internal/entities/container"
 	"beszel/internal/entities/system"
+	"encoding/json"
 	"fmt"
 	"log"
 	"math"
 	"strings"
 	"time"
 
-	"github.com/goccy/go-json"
 	"github.com/pocketbase/dbx"
 	"github.com/pocketbase/pocketbase/core"
 )
@@ -26,14 +26,26 @@ type LongerRecordData struct {
 	minShorterRecords  int
 }
 
-type RecordStats []struct {
+type RecordIds []struct {
-	Stats []byte `db:"stats"`
+	Id string `db:"id"`
 }
 
 func NewRecordManager(app core.App) *RecordManager {
 	return &RecordManager{app}
 }
 
+type StatsRecord struct {
+	Stats []byte `db:"stats"`
+}
+
+// global variables for reusing allocations
+var statsRecord StatsRecord
+var containerStats []container.Stats
+var sumStats system.Stats
+var tempStats system.Stats
+var queryParams = make(dbx.Params, 1)
+var containerSums = make(map[string]*container.Stats)
+
 // Create longer records by averaging shorter records
 func (rm *RecordManager) CreateLongerRecords() {
 	// start := time.Now()
@@ -76,11 +88,10 @@ func (rm *RecordManager) CreateLongerRecords() {
 		if err != nil {
 			return err
 		}
-		var systems []struct {
+		var systems RecordIds
-			Id string `db:"id"`
+		db := txApp.DB()
-		}
 
-		txApp.DB().NewQuery("SELECT id FROM systems WHERE status='up'").All(&systems)
+		db.NewQuery("SELECT id FROM systems WHERE status='up'").All(&systems)
 
 		// loop through all active systems, time periods, and collections
 		for _, system := range systems {
@@ -96,22 +107,23 @@ func (rm *RecordManager) CreateLongerRecords() {
 				for _, collection := range collections {
 					// check creation time of last longer record if not 10m, since 10m is created every run
 					if recordData.longerType != "10m" {
-						lastLongerRecord, err := txApp.FindFirstRecordByFilter(
+						count, err := txApp.CountRecords(
 							collection.Id,
-							"system = {:system} && type = {:type} && created > {:created}",
+							dbx.NewExp(
-							dbx.Params{"type": recordData.longerType, "system": system.Id, "created": longerRecordPeriod},
+								"system = {:system} AND type = {:type} AND created > {:created}",
+								dbx.Params{"type": recordData.longerType, "system": system.Id, "created": longerRecordPeriod},
+							),
 						)
 						// continue if longer record exists
-						if err == nil || lastLongerRecord != nil {
+						if err != nil || count > 0 {
-							// log.Println("longer record found. continuing")
 							continue
 						}
 					}
 					// get shorter records from the past x minutes
-					var stats RecordStats
+					var recordIds RecordIds
 
 					err := txApp.DB().
-						Select("stats").
+						Select("id").
 						From(collection.Name).
 						AndWhere(dbx.NewExp(
 							"system={:system} AND type={:type} AND created > {:created}",
@@ -121,10 +133,10 @@ func (rm *RecordManager) CreateLongerRecords() {
 								"created": shorterRecordPeriod,
 							},
 						)).
-						All(&stats)
+						All(&recordIds)
 
 					// continue if not enough shorter records
-					if err != nil || len(stats) < recordData.minShorterRecords {
+					if err != nil || len(recordIds) < recordData.minShorterRecords {
 						continue
 					}
 					// average the shorter records and create longer record
@@ -133,9 +145,10 @@ func (rm *RecordManager) CreateLongerRecords() {
 					longerRecord.Set("type", recordData.longerType)
 					switch collection.Name {
 					case "system_stats":
-						longerRecord.Set("stats", rm.AverageSystemStats(stats))
+						longerRecord.Set("stats", rm.AverageSystemStats(db, recordIds))
 					case "container_stats":
-						longerRecord.Set("stats", rm.AverageContainerStats(stats))
+
+						longerRecord.Set("stats", rm.AverageContainerStats(db, recordIds))
 					}
 					if err := txApp.SaveNoValidate(longerRecord); err != nil {
 						log.Println("failed to save longer record", "err", err)
@@ -147,24 +160,36 @@ func (rm *RecordManager) CreateLongerRecords() {
 		return nil
 	})
 
+	statsRecord.Stats = statsRecord.Stats[:0]
+
 	// log.Println("finished creating longer records", "time (ms)", time.Since(start).Milliseconds())
 }
 
 // Calculate the average stats of a list of system_stats records without reflect
-func (rm *RecordManager) AverageSystemStats(records RecordStats) *system.Stats {
+func (rm *RecordManager) AverageSystemStats(db dbx.Builder, records RecordIds) *system.Stats {
-	sum := &system.Stats{}
+	// Clear/reset global structs for reuse
+	sumStats = system.Stats{}
+	tempStats = system.Stats{}
+	sum := &sumStats
+	stats := &tempStats
+	// necessary because uint8 is not big enough for the sum
+	batterySum := 0
+
 	count := float64(len(records))
 	tempCount := float64(0)
 
-	// Temporary struct for unmarshaling
-	stats := &system.Stats{}
-
 	// Accumulate totals
-	for i := range records {
+	for _, record := range records {
-		*stats = system.Stats{} // Reset tempStats for unmarshaling
+		id := record.Id
-		if err := json.Unmarshal(records[i].Stats, stats); err != nil {
+		// clear global statsRecord for reuse
+		statsRecord.Stats = statsRecord.Stats[:0]
+
+		queryParams["id"] = id
+		db.NewQuery("SELECT stats FROM system_stats WHERE id = {:id}").Bind(queryParams).One(&statsRecord)
+		if err := json.Unmarshal(statsRecord.Stats, stats); err != nil {
 			continue
 		}
+
 		sum.Cpu += stats.Cpu
 		sum.Mem += stats.Mem
 		sum.MemUsed += stats.MemUsed
@@ -180,12 +205,22 @@ func (rm *RecordManager) AverageSystemStats(records RecordStats) *system.Stats {
 		sum.DiskWritePs += stats.DiskWritePs
 		sum.NetworkSent += stats.NetworkSent
 		sum.NetworkRecv += stats.NetworkRecv
+		sum.LoadAvg[0] += stats.LoadAvg[0]
+		sum.LoadAvg[1] += stats.LoadAvg[1]
+		sum.LoadAvg[2] += stats.LoadAvg[2]
+		sum.Bandwidth[0] += stats.Bandwidth[0]
+		sum.Bandwidth[1] += stats.Bandwidth[1]
+		batterySum += int(stats.Battery[0])
+		sum.Battery[1] = stats.Battery[1]
 		// Set peak values
 		sum.MaxCpu = max(sum.MaxCpu, stats.MaxCpu, stats.Cpu)
+		sum.MaxMem = max(sum.MaxMem, stats.MaxMem, stats.MemUsed)
 		sum.MaxNetworkSent = max(sum.MaxNetworkSent, stats.MaxNetworkSent, stats.NetworkSent)
 		sum.MaxNetworkRecv = max(sum.MaxNetworkRecv, stats.MaxNetworkRecv, stats.NetworkRecv)
 		sum.MaxDiskReadPs = max(sum.MaxDiskReadPs, stats.MaxDiskReadPs, stats.DiskReadPs)
 		sum.MaxDiskWritePs = max(sum.MaxDiskWritePs, stats.MaxDiskWritePs, stats.DiskWritePs)
+		sum.MaxBandwidth[0] = max(sum.MaxBandwidth[0], stats.MaxBandwidth[0], stats.Bandwidth[0])
+		sum.MaxBandwidth[1] = max(sum.MaxBandwidth[1], stats.MaxBandwidth[1], stats.Bandwidth[1])
 
 		// Accumulate temperatures
 		if stats.Temperatures != nil {
@@ -255,7 +290,12 @@ func (rm *RecordManager) AverageSystemStats(records RecordStats) *system.Stats {
 		sum.DiskWritePs = twoDecimals(sum.DiskWritePs / count)
 		sum.NetworkSent = twoDecimals(sum.NetworkSent / count)
 		sum.NetworkRecv = twoDecimals(sum.NetworkRecv / count)
-
+		sum.LoadAvg[0] = twoDecimals(sum.LoadAvg[0] / count)
+		sum.LoadAvg[1] = twoDecimals(sum.LoadAvg[1] / count)
+		sum.LoadAvg[2] = twoDecimals(sum.LoadAvg[2] / count)
+		sum.Bandwidth[0] = sum.Bandwidth[0] / uint64(count)
+		sum.Bandwidth[1] = sum.Bandwidth[1] / uint64(count)
+		sum.Battery[0] = uint8(batterySum / int(count))
 		// Average temperatures
 		if sum.Temperatures != nil && tempCount > 0 {
 			for key := range sum.Temperatures {
@@ -293,14 +333,24 @@ func (rm *RecordManager) AverageSystemStats(records RecordStats) *system.Stats {
 }
 
 // Calculate the average stats of a list of container_stats records
-func (rm *RecordManager) AverageContainerStats(records RecordStats) []container.Stats {
+func (rm *RecordManager) AverageContainerStats(db dbx.Builder, records RecordIds) []container.Stats {
-	sums := make(map[string]*container.Stats)
+	// Clear global map for reuse
+	for k := range containerSums {
+		delete(containerSums, k)
+	}
+	sums := containerSums
 	count := float64(len(records))
-	containerStats := make([]container.Stats, 0, 50)
+
 	for i := range records {
-		// reset slice
+		id := records[i].Id
+		// clear global statsRecord and containerStats for reuse
+		statsRecord.Stats = statsRecord.Stats[:0]
 		containerStats = containerStats[:0]
-		if err := json.Unmarshal(records[i].Stats, &containerStats); err != nil {
+
+		queryParams["id"] = id
+		db.NewQuery("SELECT stats FROM container_stats WHERE id = {:id}").Bind(queryParams).One(&statsRecord)
+
+		if err := json.Unmarshal(statsRecord.Stats, &containerStats); err != nil {
 			return []container.Stats{}
 		}
 		for i := range containerStats {
@@ -328,12 +378,46 @@ func (rm *RecordManager) AverageContainerStats(records RecordStats) []container.
 	return result
 }
 
-// Deletes records older than what is displayed in the UI
+// Delete old records
 func (rm *RecordManager) DeleteOldRecords() {
-	// Define the collections to process
+	rm.app.RunInTransaction(func(txApp core.App) error {
-	collections := []string{"system_stats", "container_stats"}
+		err := deleteOldSystemStats(txApp)
+		if err != nil {
+			return err
+		}
+		err = deleteOldAlertsHistory(txApp, 200, 250)
+		if err != nil {
+			return err
+		}
+		return nil
+	})
+}
 
-	// Define record types and their retention periods
+// Delete old alerts history records
+func deleteOldAlertsHistory(app core.App, countToKeep, countBeforeDeletion int) error {
+	db := app.DB()
+	var users []struct {
+		Id string `db:"user"`
+	}
+	err := db.NewQuery("SELECT user, COUNT(*) as count FROM alerts_history GROUP BY user HAVING count > {:countBeforeDeletion}").Bind(dbx.Params{"countBeforeDeletion": countBeforeDeletion}).All(&users)
+	if err != nil {
+		return err
+	}
+	for _, user := range users {
+		_, err = db.NewQuery("DELETE FROM alerts_history WHERE user = {:user} AND id NOT IN (SELECT id FROM alerts_history WHERE user = {:user} ORDER BY created DESC LIMIT {:countToKeep})").Bind(dbx.Params{"user": user.Id, "countToKeep": countToKeep}).Execute()
+		if err != nil {
+			return err
+		}
+	}
+	return nil
+}
+
+// Deletes system_stats records older than what is displayed in the UI
+func deleteOldSystemStats(app core.App) error {
+	// Collections to process
+	collections := [2]string{"system_stats", "container_stats"}
+
+	// Record types and their retention periods
 	type RecordDeletionData struct {
 		recordType string
 		retention  time.Duration
@@ -346,31 +430,28 @@ func (rm *RecordManager) DeleteOldRecords() {
 		{recordType: "480m", retention: 30 * 24 * time.Hour}, // 30 days
 	}
 
-	// Process each collection
+	now := time.Now().UTC()
+
 	for _, collection := range collections {
-		// Build the WHERE clause dynamically
+		// Build the WHERE clause
 		var conditionParts []string
 		var params dbx.Params = make(map[string]any)
-
+		for i := range recordData {
-		for i, rd := range recordData {
+			rd := recordData[i]
 			// Create parameterized condition for this record type
 			dateParam := fmt.Sprintf("date%d", i)
 			conditionParts = append(conditionParts, fmt.Sprintf("(type = '%s' AND created < {:%s})", rd.recordType, dateParam))
-			params[dateParam] = time.Now().UTC().Add(-rd.retention)
+			params[dateParam] = now.Add(-rd.retention)
 		}
-
 		// Combine conditions with OR
 		conditionStr := strings.Join(conditionParts, " OR ")
-
+		// Construct and execute the full raw query
-		// Construct the full raw query
 		rawQuery := fmt.Sprintf("DELETE FROM %s WHERE %s", collection, conditionStr)
-
+		if _, err := app.DB().NewQuery(rawQuery).Bind(params).Execute(); err != nil {
-		// Execute the query with parameters
+			return fmt.Errorf("failed to delete from %s: %v", collection, err)
-		if _, err := rm.app.DB().NewQuery(rawQuery).Bind(params).Execute(); err != nil {
-			// return fmt.Errorf("failed to delete from %s: %v", collection, err)
-			rm.app.Logger().Error("failed to delete", "collection", collection, "error", err)
 		}
 	}
+	return nil
 }
 
 /* Round float to two decimals */

beszel/internal/records/records_test.go

@@ -0,0 +1,381 @@
+//go:build testing
+// +build testing
+
+package records_test
+
+import (
+	"beszel/internal/records"
+	"beszel/internal/tests"
+	"fmt"
+	"testing"
+	"time"
+
+	"github.com/pocketbase/dbx"
+	"github.com/pocketbase/pocketbase/core"
+	"github.com/pocketbase/pocketbase/tools/types"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+// TestDeleteOldRecords tests the main DeleteOldRecords function
+func TestDeleteOldRecords(t *testing.T) {
+	hub, err := tests.NewTestHub(t.TempDir())
+	require.NoError(t, err)
+	defer hub.Cleanup()
+
+	rm := records.NewRecordManager(hub)
+
+	// Create test user for alerts history
+	user, err := tests.CreateUser(hub, "test@example.com", "testtesttest")
+	require.NoError(t, err)
+
+	// Create test system
+	system, err := tests.CreateRecord(hub, "systems", map[string]any{
+		"name":   "test-system",
+		"host":   "localhost",
+		"port":   "45876",
+		"status": "up",
+		"users":  []string{user.Id},
+	})
+	require.NoError(t, err)
+
+	now := time.Now()
+
+	// Create old system_stats records that should be deleted
+	var record *core.Record
+	record, err = tests.CreateRecord(hub, "system_stats", map[string]any{
+		"system": system.Id,
+		"type":   "1m",
+		"stats":  `{"cpu": 50.0, "mem": 1024}`,
+	})
+	require.NoError(t, err)
+	// created is autodate field, so we need to set it manually
+	record.SetRaw("created", now.UTC().Add(-2*time.Hour).Format(types.DefaultDateLayout))
+	err = hub.SaveNoValidate(record)
+	require.NoError(t, err)
+	require.NotNil(t, record)
+	require.InDelta(t, record.GetDateTime("created").Time().UTC().Unix(), now.UTC().Add(-2*time.Hour).Unix(), 1)
+	require.Equal(t, record.Get("system"), system.Id)
+	require.Equal(t, record.Get("type"), "1m")
+
+	// Create recent system_stats record that should be kept
+	_, err = tests.CreateRecord(hub, "system_stats", map[string]any{
+		"system":  system.Id,
+		"type":    "1m",
+		"stats":   `{"cpu": 30.0, "mem": 512}`,
+		"created": now.Add(-30 * time.Minute), // 30 minutes old, should be kept
+	})
+	require.NoError(t, err)
+
+	// Create many alerts history records to trigger deletion
+	for i := range 260 { // More than countBeforeDeletion (250)
+		_, err = tests.CreateRecord(hub, "alerts_history", map[string]any{
+			"user":    user.Id,
+			"name":    "CPU",
+			"value":   i + 1,
+			"system":  system.Id,
+			"created": now.Add(-time.Duration(i) * time.Minute),
+		})
+		require.NoError(t, err)
+	}
+
+	// Count records before deletion
+	systemStatsCountBefore, err := hub.CountRecords("system_stats")
+	require.NoError(t, err)
+	alertsCountBefore, err := hub.CountRecords("alerts_history")
+	require.NoError(t, err)
+
+	// Run deletion
+	rm.DeleteOldRecords()
+
+	// Count records after deletion
+	systemStatsCountAfter, err := hub.CountRecords("system_stats")
+	require.NoError(t, err)
+	alertsCountAfter, err := hub.CountRecords("alerts_history")
+	require.NoError(t, err)
+
+	// Verify old system stats were deleted
+	assert.Less(t, systemStatsCountAfter, systemStatsCountBefore, "Old system stats should be deleted")
+
+	// Verify alerts history was trimmed
+	assert.Less(t, alertsCountAfter, alertsCountBefore, "Excessive alerts history should be deleted")
+	assert.Equal(t, alertsCountAfter, int64(200), "Alerts count should be equal to countToKeep (200)")
+}
+
+// TestDeleteOldSystemStats tests the deleteOldSystemStats function
+func TestDeleteOldSystemStats(t *testing.T) {
+	hub, err := tests.NewTestHub(t.TempDir())
+	require.NoError(t, err)
+	defer hub.Cleanup()
+
+	// Create test system
+	user, err := tests.CreateUser(hub, "test@example.com", "testtesttest")
+	require.NoError(t, err)
+
+	system, err := tests.CreateRecord(hub, "systems", map[string]any{
+		"name":   "test-system",
+		"host":   "localhost",
+		"port":   "45876",
+		"status": "up",
+		"users":  []string{user.Id},
+	})
+	require.NoError(t, err)
+
+	now := time.Now().UTC()
+
+	// Test data for different record types and their retention periods
+	testCases := []struct {
+		recordType   string
+		retention    time.Duration
+		shouldBeKept bool
+		ageFromNow   time.Duration
+		description  string
+	}{
+		{"1m", time.Hour, true, 30 * time.Minute, "1m record within 1 hour should be kept"},
+		{"1m", time.Hour, false, 2 * time.Hour, "1m record older than 1 hour should be deleted"},
+		{"10m", 12 * time.Hour, true, 6 * time.Hour, "10m record within 12 hours should be kept"},
+		{"10m", 12 * time.Hour, false, 24 * time.Hour, "10m record older than 12 hours should be deleted"},
+		{"20m", 24 * time.Hour, true, 12 * time.Hour, "20m record within 24 hours should be kept"},
+		{"20m", 24 * time.Hour, false, 48 * time.Hour, "20m record older than 24 hours should be deleted"},
+		{"120m", 7 * 24 * time.Hour, true, 3 * 24 * time.Hour, "120m record within 7 days should be kept"},
+		{"120m", 7 * 24 * time.Hour, false, 10 * 24 * time.Hour, "120m record older than 7 days should be deleted"},
+		{"480m", 30 * 24 * time.Hour, true, 15 * 24 * time.Hour, "480m record within 30 days should be kept"},
+		{"480m", 30 * 24 * time.Hour, false, 45 * 24 * time.Hour, "480m record older than 30 days should be deleted"},
+	}
+
+	// Create test records for both system_stats and container_stats
+	collections := []string{"system_stats", "container_stats"}
+	recordIds := make(map[string][]string)
+
+	for _, collection := range collections {
+		recordIds[collection] = make([]string, 0)
+
+		for i, tc := range testCases {
+			recordTime := now.Add(-tc.ageFromNow)
+
+			var stats string
+			if collection == "system_stats" {
+				stats = fmt.Sprintf(`{"cpu": %d.0, "mem": %d}`, i*10, i*100)
+			} else {
+				stats = fmt.Sprintf(`[{"name": "container%d", "cpu": %d.0, "mem": %d}]`, i, i*5, i*50)
+			}
+
+			record, err := tests.CreateRecord(hub, collection, map[string]any{
+				"system": system.Id,
+				"type":   tc.recordType,
+				"stats":  stats,
+			})
+			require.NoError(t, err)
+			record.SetRaw("created", recordTime.Format(types.DefaultDateLayout))
+			err = hub.SaveNoValidate(record)
+			require.NoError(t, err)
+			recordIds[collection] = append(recordIds[collection], record.Id)
+		}
+	}
+
+	// Run deletion
+	err = records.TestDeleteOldSystemStats(hub)
+	require.NoError(t, err)
+
+	// Verify results
+	for _, collection := range collections {
+		for i, tc := range testCases {
+			recordId := recordIds[collection][i]
+
+			// Try to find the record
+			_, err := hub.FindRecordById(collection, recordId)
+
+			if tc.shouldBeKept {
+				assert.NoError(t, err, "Record should exist: %s", tc.description)
+			} else {
+				assert.Error(t, err, "Record should be deleted: %s", tc.description)
+			}
+		}
+	}
+}
+
+// TestDeleteOldAlertsHistory tests the deleteOldAlertsHistory function
+func TestDeleteOldAlertsHistory(t *testing.T) {
+	hub, err := tests.NewTestHub(t.TempDir())
+	require.NoError(t, err)
+	defer hub.Cleanup()
+
+	// Create test users
+	user1, err := tests.CreateUser(hub, "user1@example.com", "testtesttest")
+	require.NoError(t, err)
+
+	user2, err := tests.CreateUser(hub, "user2@example.com", "testtesttest")
+	require.NoError(t, err)
+
+	system, err := tests.CreateRecord(hub, "systems", map[string]any{
+		"name":   "test-system",
+		"host":   "localhost",
+		"port":   "45876",
+		"status": "up",
+		"users":  []string{user1.Id, user2.Id},
+	})
+	require.NoError(t, err)
+	now := time.Now().UTC()
+
+	testCases := []struct {
+		name                  string
+		user                  *core.Record
+		alertCount            int
+		countToKeep           int
+		countBeforeDeletion   int
+		expectedAfterDeletion int
+		description           string
+	}{
+		{
+			name:                  "User with few alerts (below threshold)",
+			user:                  user1,
+			alertCount:            100,
+			countToKeep:           50,
+			countBeforeDeletion:   150,
+			expectedAfterDeletion: 100, // No deletion because below threshold
+			description:           "User with alerts below countBeforeDeletion should not have any deleted",
+		},
+		{
+			name:                  "User with many alerts (above threshold)",
+			user:                  user2,
+			alertCount:            300,
+			countToKeep:           100,
+			countBeforeDeletion:   200,
+			expectedAfterDeletion: 100, // Should be trimmed to countToKeep
+			description:           "User with alerts above countBeforeDeletion should be trimmed to countToKeep",
+		},
+	}
+
+	for _, tc := range testCases {
+		t.Run(tc.name, func(t *testing.T) {
+			// Create alerts for this user
+			for i := 0; i < tc.alertCount; i++ {
+				_, err := tests.CreateRecord(hub, "alerts_history", map[string]any{
+					"user":    tc.user.Id,
+					"name":    "CPU",
+					"value":   i + 1,
+					"system":  system.Id,
+					"created": now.Add(-time.Duration(i) * time.Minute),
+				})
+				require.NoError(t, err)
+			}
+
+			// Count before deletion
+			countBefore, err := hub.CountRecords("alerts_history",
+				dbx.NewExp("user = {:user}", dbx.Params{"user": tc.user.Id}))
+			require.NoError(t, err)
+			assert.Equal(t, int64(tc.alertCount), countBefore, "Initial count should match")
+
+			// Run deletion
+			err = records.TestDeleteOldAlertsHistory(hub, tc.countToKeep, tc.countBeforeDeletion)
+			require.NoError(t, err)
+
+			// Count after deletion
+			countAfter, err := hub.CountRecords("alerts_history",
+				dbx.NewExp("user = {:user}", dbx.Params{"user": tc.user.Id}))
+			require.NoError(t, err)
+
+			assert.Equal(t, int64(tc.expectedAfterDeletion), countAfter, tc.description)
+
+			// If deletion occurred, verify the most recent records were kept
+			if tc.expectedAfterDeletion < tc.alertCount {
+				records, err := hub.FindRecordsByFilter("alerts_history",
+					"user = {:user}",
+					"-created", // Order by created DESC
+					tc.countToKeep,
+					0,
+					map[string]any{"user": tc.user.Id})
+				require.NoError(t, err)
+				assert.Len(t, records, tc.expectedAfterDeletion, "Should have exactly countToKeep records")
+
+				// Verify records are in descending order by created time
+				for i := 1; i < len(records); i++ {
+					prev := records[i-1].GetDateTime("created").Time()
+					curr := records[i].GetDateTime("created").Time()
+					assert.True(t, prev.After(curr) || prev.Equal(curr),
+						"Records should be ordered by created time (newest first)")
+				}
+			}
+		})
+	}
+}
+
+// TestDeleteOldAlertsHistoryEdgeCases tests edge cases for alerts history deletion
+func TestDeleteOldAlertsHistoryEdgeCases(t *testing.T) {
+	hub, err := tests.NewTestHub(t.TempDir())
+	require.NoError(t, err)
+	defer hub.Cleanup()
+
+	t.Run("No users with excessive alerts", func(t *testing.T) {
+		// Create user with few alerts
+		user, err := tests.CreateUser(hub, "few@example.com", "testtesttest")
+		require.NoError(t, err)
+
+		system, err := tests.CreateRecord(hub, "systems", map[string]any{
+			"name":   "test-system",
+			"host":   "localhost",
+			"port":   "45876",
+			"status": "up",
+			"users":  []string{user.Id},
+		})
+
+		// Create only 5 alerts (well below threshold)
+		for i := range 5 {
+			_, err := tests.CreateRecord(hub, "alerts_history", map[string]any{
+				"user":   user.Id,
+				"name":   "CPU",
+				"value":  i + 1,
+				"system": system.Id,
+			})
+			require.NoError(t, err)
+		}
+
+		// Should not error and should not delete anything
+		err = records.TestDeleteOldAlertsHistory(hub, 10, 20)
+		require.NoError(t, err)
+
+		count, err := hub.CountRecords("alerts_history")
+		require.NoError(t, err)
+		assert.Equal(t, int64(5), count, "All alerts should remain")
+	})
+
+	t.Run("Empty alerts_history table", func(t *testing.T) {
+		// Clear any existing alerts
+		_, err := hub.DB().NewQuery("DELETE FROM alerts_history").Execute()
+		require.NoError(t, err)
+
+		// Should not error with empty table
+		err = records.TestDeleteOldAlertsHistory(hub, 10, 20)
+		require.NoError(t, err)
+	})
+}
+
+// TestRecordManagerCreation tests RecordManager creation
+func TestRecordManagerCreation(t *testing.T) {
+	hub, err := tests.NewTestHub(t.TempDir())
+	require.NoError(t, err)
+	defer hub.Cleanup()
+
+	rm := records.NewRecordManager(hub)
+	assert.NotNil(t, rm, "RecordManager should not be nil")
+}
+
+// TestTwoDecimals tests the twoDecimals helper function
+func TestTwoDecimals(t *testing.T) {
+	testCases := []struct {
+		input    float64
+		expected float64
+	}{
+		{1.234567, 1.23},
+		{1.235, 1.24}, // Should round up
+		{1.0, 1.0},
+		{0.0, 0.0},
+		{-1.234567, -1.23},
+		{-1.235, -1.23}, // Negative rounding
+	}
+
+	for _, tc := range testCases {
+		result := records.TestTwoDecimals(tc.input)
+		assert.InDelta(t, tc.expected, result, 0.02, "twoDecimals(%f) should equal %f", tc.input, tc.expected)
+	}
+}

beszel/internal/records/records_test_helpers.go

@@ -0,0 +1,23 @@
+//go:build testing
+// +build testing
+
+package records
+
+import (
+	"github.com/pocketbase/pocketbase/core"
+)
+
+// TestDeleteOldSystemStats exposes deleteOldSystemStats for testing
+func TestDeleteOldSystemStats(app core.App) error {
+	return deleteOldSystemStats(app)
+}
+
+// TestDeleteOldAlertsHistory exposes deleteOldAlertsHistory for testing
+func TestDeleteOldAlertsHistory(app core.App, countToKeep, countBeforeDeletion int) error {
+	return deleteOldAlertsHistory(app, countToKeep, countBeforeDeletion)
+}
+
+// TestTwoDecimals exposes twoDecimals for testing
+func TestTwoDecimals(value float64) float64 {
+	return twoDecimals(value)
+}

beszel/internal/tests/api.go

@@ -0,0 +1,309 @@
+package tests
+
+import (
+	"bytes"
+	"context"
+	"encoding/json"
+	"fmt"
+	"io"
+	"maps"
+	"net/http"
+	"net/http/httptest"
+	"strings"
+	"testing"
+	"time"
+
+	"github.com/pocketbase/pocketbase/apis"
+	"github.com/pocketbase/pocketbase/core"
+	pbtests "github.com/pocketbase/pocketbase/tests"
+	"github.com/pocketbase/pocketbase/tools/hook"
+)
+
+// NOTE: This is a copy of https://github.com/pocketbase/pocketbase/blob/master/tests/api.go
+// with the following changes:
+// - Removed automatic cleanup of the test app in ApiScenario.Test (Aug 17 2025)
+
+// ApiScenario defines a single api request test case/scenario.
+type ApiScenario struct {
+	// Name is the test name.
+	Name string
+
+	// Method is the HTTP method of the test request to use.
+	Method string
+
+	// URL is the url/path of the endpoint you want to test.
+	URL string
+
+	// Body specifies the body to send with the request.
+	//
+	// For example:
+	//
+	//	strings.NewReader(`{"title":"abc"}`)
+	Body io.Reader
+
+	// Headers specifies the headers to send with the request (e.g. "Authorization": "abc")
+	Headers map[string]string
+
+	// Delay adds a delay before checking the expectations usually
+	// to ensure that all fired non-awaited go routines have finished
+	Delay time.Duration
+
+	// Timeout specifies how long to wait before cancelling the request context.
+	//
+	// A zero or negative value means that there will be no timeout.
+	Timeout time.Duration
+
+	// expectations
+	// ---------------------------------------------------------------
+
+	// ExpectedStatus specifies the expected response HTTP status code.
+	ExpectedStatus int
+
+	// List of keywords that MUST exist in the response body.
+	//
+	// Either ExpectedContent or NotExpectedContent must be set if the response body is non-empty.
+	// Leave both fields empty if you want to ensure that the response didn't have any body (e.g. 204).
+	ExpectedContent []string
+
+	// List of keywords that MUST NOT exist in the response body.
+	//
+	// Either ExpectedContent or NotExpectedContent must be set if the response body is non-empty.
+	// Leave both fields empty if you want to ensure that the response didn't have any body (e.g. 204).
+	NotExpectedContent []string
+
+	// List of hook events to check whether they were fired or not.
+	//
+	// You can use the wildcard "*" event key if you want to ensure
+	// that no other hook events except those listed have been fired.
+	//
+	// For example:
+	//
+	//	map[string]int{ "*": 0 } // no hook events were fired
+	//	map[string]int{ "*": 0, "EventA": 2 } // no hook events, except EventA were fired
+	//	map[string]int{ "EventA": 2, "EventB": 0 } // ensures that EventA was fired exactly 2 times and EventB exactly 0 times.
+	ExpectedEvents map[string]int
+
+	// test hooks
+	// ---------------------------------------------------------------
+
+	TestAppFactory func(t testing.TB) *pbtests.TestApp
+	BeforeTestFunc func(t testing.TB, app *pbtests.TestApp, e *core.ServeEvent)
+	AfterTestFunc  func(t testing.TB, app *pbtests.TestApp, res *http.Response)
+}
+
+// Test executes the test scenario.
+//
+// Example:
+//
+//	func TestListExample(t *testing.T) {
+//	    scenario := tests.ApiScenario{
+//	        Name:           "list example collection",
+//	        Method:         http.MethodGet,
+//	        URL:            "/api/collections/example/records",
+//	        ExpectedStatus: 200,
+//	        ExpectedContent: []string{
+//	            `"totalItems":3`,
+//	            `"id":"0yxhwia2amd8gec"`,
+//	            `"id":"achvryl401bhse3"`,
+//	            `"id":"llvuca81nly1qls"`,
+//	        },
+//	        ExpectedEvents: map[string]int{
+//	            "OnRecordsListRequest": 1,
+//	            "OnRecordEnrich":       3,
+//	        },
+//	    }
+//
+//	    scenario.Test(t)
+//	}
+func (scenario *ApiScenario) Test(t *testing.T) {
+	t.Run(scenario.normalizedName(), func(t *testing.T) {
+		scenario.test(t)
+	})
+}
+
+// Benchmark benchmarks the test scenario.
+//
+// Example:
+//
+//	func BenchmarkListExample(b *testing.B) {
+//	    scenario := tests.ApiScenario{
+//	        Name:           "list example collection",
+//	        Method:         http.MethodGet,
+//	        URL:            "/api/collections/example/records",
+//	        ExpectedStatus: 200,
+//	        ExpectedContent: []string{
+//	            `"totalItems":3`,
+//	            `"id":"0yxhwia2amd8gec"`,
+//	            `"id":"achvryl401bhse3"`,
+//	            `"id":"llvuca81nly1qls"`,
+//	        },
+//	        ExpectedEvents: map[string]int{
+//	            "OnRecordsListRequest": 1,
+//	            "OnRecordEnrich":       3,
+//	        },
+//	    }
+//
+//	    scenario.Benchmark(b)
+//	}
+func (scenario *ApiScenario) Benchmark(b *testing.B) {
+	b.Run(scenario.normalizedName(), func(b *testing.B) {
+		for i := 0; i < b.N; i++ {
+			scenario.test(b)
+		}
+	})
+}
+
+func (scenario *ApiScenario) normalizedName() string {
+	var name = scenario.Name
+
+	if name == "" {
+		name = fmt.Sprintf("%s:%s", scenario.Method, scenario.URL)
+	}
+
+	return name
+}
+
+func (scenario *ApiScenario) test(t testing.TB) {
+	var testApp *pbtests.TestApp
+	if scenario.TestAppFactory != nil {
+		testApp = scenario.TestAppFactory(t)
+		if testApp == nil {
+			t.Fatal("TestAppFactory must return a non-nill app instance")
+		}
+	} else {
+		var testAppErr error
+		testApp, testAppErr = pbtests.NewTestApp()
+		if testAppErr != nil {
+			t.Fatalf("Failed to initialize the test app instance: %v", testAppErr)
+		}
+	}
+	// defer testApp.Cleanup()
+
+	baseRouter, err := apis.NewRouter(testApp)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	// manually trigger the serve event to ensure that custom app routes and middlewares are registered
+	serveEvent := new(core.ServeEvent)
+	serveEvent.App = testApp
+	serveEvent.Router = baseRouter
+
+	serveErr := testApp.OnServe().Trigger(serveEvent, func(e *core.ServeEvent) error {
+		if scenario.BeforeTestFunc != nil {
+			scenario.BeforeTestFunc(t, testApp, e)
+		}
+
+		// reset the event counters in case a hook was triggered from a before func (eg. db save)
+		testApp.ResetEventCalls()
+
+		// add middleware to timeout long-running requests (eg. keep-alive routes)
+		e.Router.Bind(&hook.Handler[*core.RequestEvent]{
+			Func: func(re *core.RequestEvent) error {
+				slowTimer := time.AfterFunc(3*time.Second, func() {
+					t.Logf("[WARN] Long running test %q", scenario.Name)
+				})
+				defer slowTimer.Stop()
+
+				if scenario.Timeout > 0 {
+					ctx, cancelFunc := context.WithTimeout(re.Request.Context(), scenario.Timeout)
+					defer cancelFunc()
+					re.Request = re.Request.Clone(ctx)
+				}
+
+				return re.Next()
+			},
+			Priority: -9999,
+		})
+
+		recorder := httptest.NewRecorder()
+
+		req := httptest.NewRequest(scenario.Method, scenario.URL, scenario.Body)
+
+		// set default header
+		req.Header.Set("content-type", "application/json")
+
+		// set scenario headers
+		for k, v := range scenario.Headers {
+			req.Header.Set(k, v)
+		}
+
+		// execute request
+		mux, err := e.Router.BuildMux()
+		if err != nil {
+			t.Fatalf("Failed to build router mux: %v", err)
+		}
+		mux.ServeHTTP(recorder, req)
+
+		res := recorder.Result()
+
+		if res.StatusCode != scenario.ExpectedStatus {
+			t.Errorf("Expected status code %d, got %d", scenario.ExpectedStatus, res.StatusCode)
+		}
+
+		if scenario.Delay > 0 {
+			time.Sleep(scenario.Delay)
+		}
+
+		if len(scenario.ExpectedContent) == 0 && len(scenario.NotExpectedContent) == 0 {
+			if len(recorder.Body.Bytes()) != 0 {
+				t.Errorf("Expected empty body, got \n%v", recorder.Body.String())
+			}
+		} else {
+			// normalize json response format
+			buffer := new(bytes.Buffer)
+			err := json.Compact(buffer, recorder.Body.Bytes())
+			var normalizedBody string
+			if err != nil {
+				// not a json...
+				normalizedBody = recorder.Body.String()
+			} else {
+				normalizedBody = buffer.String()
+			}
+
+			for _, item := range scenario.ExpectedContent {
+				if !strings.Contains(normalizedBody, item) {
+					t.Errorf("Cannot find %v in response body \n%v", item, normalizedBody)
+					break
+				}
+			}
+
+			for _, item := range scenario.NotExpectedContent {
+				if strings.Contains(normalizedBody, item) {
+					t.Errorf("Didn't expect %v in response body \n%v", item, normalizedBody)
+					break
+				}
+			}
+		}
+
+		remainingEvents := maps.Clone(testApp.EventCalls)
+
+		var noOtherEventsShouldRemain bool
+		for event, expectedNum := range scenario.ExpectedEvents {
+			if event == "*" && expectedNum <= 0 {
+				noOtherEventsShouldRemain = true
+				continue
+			}
+
+			actualNum := remainingEvents[event]
+			if actualNum != expectedNum {
+				t.Errorf("Expected event %s to be called %d, got %d", event, expectedNum, actualNum)
+			}
+
+			delete(remainingEvents, event)
+		}
+
+		if noOtherEventsShouldRemain && len(remainingEvents) > 0 {
+			t.Errorf("Missing expected remaining events:\n%#v\nAll triggered app events are:\n%#v", remainingEvents, testApp.EventCalls)
+		}
+
+		if scenario.AfterTestFunc != nil {
+			scenario.AfterTestFunc(t, testApp, res)
+		}
+
+		return nil
+	})
+	if serveErr != nil {
+		t.Fatalf("Failed to trigger app serve hook: %v", serveErr)
+	}
+}

beszel/internal/tests/hub.go

@@ -1,11 +1,17 @@
+//go:build testing
+// +build testing
+
 // Package tests provides helpers for testing the application.
 package tests
 
 import (
 	"beszel/internal/hub"
+	"fmt"
+	"testing"
 
 	"github.com/pocketbase/pocketbase/core"
 	"github.com/pocketbase/pocketbase/tests"
+	"github.com/stretchr/testify/assert"
 
 	_ "github.com/pocketbase/pocketbase/migrations"
 )
@@ -56,3 +62,37 @@ func NewTestHubWithConfig(config core.BaseAppConfig) (*TestHub, error) {
 
 	return t, nil
 }
+
+// Helper function to create a test user for config tests
+func CreateUser(app core.App, email string, password string) (*core.Record, error) {
+	userCollection, err := app.FindCachedCollectionByNameOrId("users")
+	if err != nil {
+		return nil, err
+	}
+
+	user := core.NewRecord(userCollection)
+	user.Set("email", email)
+	user.Set("password", password)
+
+	return user, app.Save(user)
+}
+
+// Helper function to create a test record
+func CreateRecord(app core.App, collectionName string, fields map[string]any) (*core.Record, error) {
+	collection, err := app.FindCachedCollectionByNameOrId(collectionName)
+	if err != nil {
+		return nil, err
+	}
+
+	record := core.NewRecord(collection)
+	record.Load(fields)
+
+	return record, app.Save(record)
+}
+
+func ClearCollection(t testing.TB, app core.App, collectionName string) error {
+	_, err := app.DB().NewQuery(fmt.Sprintf("DELETE from %s", collectionName)).Execute()
+	recordCount, err := app.CountRecords(collectionName)
+	assert.EqualValues(t, recordCount, 0, "should have 0 records after clearing")
+	return err
+}

beszel/internal/users/users.go

@@ -6,6 +6,7 @@ import (
 	"log"
 	"net/http"
 
+	"github.com/pocketbase/dbx"
 	"github.com/pocketbase/pocketbase/core"
 )
 
@@ -13,13 +14,6 @@ type UserManager struct {
 	app core.App
 }
 
-type UserSettings struct {
-	ChartTime            string   `json:"chartTime"`
-	NotificationEmails   []string `json:"emails"`
-	NotificationWebhooks []string `json:"webhooks"`
-	// Language             string   `json:"lang"`
-}
-
 func NewUserManager(app core.App) *UserManager {
 	return &UserManager{
 		app: app,
@@ -37,30 +31,26 @@ func (um *UserManager) InitializeUserRole(e *core.RecordEvent) error {
 // Initialize user settings with defaults if not set
 func (um *UserManager) InitializeUserSettings(e *core.RecordEvent) error {
 	record := e.Record
-	// intialize settings with defaults
+	// intialize settings with defaults (zero values can be ignored)
-	settings := UserSettings{
+	settings := struct {
-		// Language:             "en",
+		ChartTime string   `json:"chartTime"`
-		ChartTime:            "1h",
+		Emails    []string `json:"emails"`
-		NotificationEmails:   []string{},
+	}{
-		NotificationWebhooks: []string{},
+		ChartTime: "1h",
 	}
 	record.UnmarshalJSONField("settings", &settings)
-	if len(settings.NotificationEmails) == 0 {
+	// get user email from auth record
-		// get user email from auth record
+	var user struct {
-		if errs := um.app.ExpandRecord(record, []string{"user"}, nil); len(errs) == 0 {
+		Email string `db:"email"`
-			// app.Logger().Error("failed to expand user relation", "errs", errs)
-			if user := record.ExpandedOne("user"); user != nil {
-				settings.NotificationEmails = []string{user.GetString("email")}
-			} else {
-				log.Println("Failed to get user email from auth record")
-			}
-		} else {
-			log.Println("failed to expand user relation", "errs", errs)
-		}
 	}
-	// if len(settings.NotificationWebhooks) == 0 {
+	err := e.App.DB().NewQuery("SELECT email FROM users WHERE id = {:id}").Bind(dbx.Params{
-	// 	settings.NotificationWebhooks = []string{""}
+		"id": record.GetString("user"),
-	// }
+	}).One(&user)
+	if err != nil {
+		log.Println("failed to get user email", "err", err)
+		return err
+	}
+	settings.Emails = []string{user.Email}
 	record.Set("settings", settings)
 	return e.Next()
 }

beszel/migrations/0_collections_snapshot_0_12_0_7.go

@@ -1,23 +1,14 @@
 package migrations
 
 import (
+	"github.com/google/uuid"
 	"github.com/pocketbase/pocketbase/core"
 	m "github.com/pocketbase/pocketbase/migrations"
 )
 
 func init() {
 	m.Register(func(app core.App) error {
-		// delete duplicate alerts
+		// update collections
-		app.DB().NewQuery(`
-		DELETE FROM alerts
-		WHERE rowid NOT IN (
-				SELECT MAX(rowid)
-				FROM alerts
-				GROUP BY user, system, name
-		);
-	`).Execute()
-
-		// import collections
 		jsonData := `[
 	{
 		"id": "elngm8x1l60zi2v",
@@ -84,7 +75,10 @@ func init() {
 					"Memory",
 					"Disk",
 					"Temperature",
-					"Bandwidth"
+					"Bandwidth",
+					"LoadAvg1",
+					"LoadAvg5",
+					"LoadAvg15"
 				]
 			},
 			{
@@ -146,6 +140,124 @@ func init() {
 		],
 		"system": false
 	},
+	{
+		"id": "pbc_1697146157",
+		"listRule": "@request.auth.id != \"\" && user.id = @request.auth.id",
+		"viewRule": "@request.auth.id != \"\" && user.id = @request.auth.id",
+		"createRule": null,
+		"updateRule": null,
+		"deleteRule": "@request.auth.id != \"\" && user.id = @request.auth.id",
+		"name": "alerts_history",
+		"type": "base",
+		"fields": [
+			{
+					"autogeneratePattern": "[a-z0-9]{15}",
+					"hidden": false,
+					"id": "text3208210256",
+					"max": 15,
+					"min": 15,
+					"name": "id",
+					"pattern": "^[a-z0-9]+$",
+					"presentable": false,
+					"primaryKey": true,
+					"required": true,
+					"system": true,
+					"type": "text"
+				},
+				{
+					"cascadeDelete": true,
+					"collectionId": "_pb_users_auth_",
+					"hidden": false,
+					"id": "relation2375276105",
+					"maxSelect": 1,
+					"minSelect": 0,
+					"name": "user",
+					"presentable": false,
+					"required": true,
+					"system": false,
+					"type": "relation"
+				},
+				{
+					"cascadeDelete": true,
+					"collectionId": "2hz5ncl8tizk5nx",
+					"hidden": false,
+					"id": "relation3377271179",
+					"maxSelect": 1,
+					"minSelect": 0,
+					"name": "system",
+					"presentable": false,
+					"required": true,
+					"system": false,
+					"type": "relation"
+				},
+				{
+					"autogeneratePattern": "",
+					"hidden": false,
+					"id": "text2466471794",
+					"max": 0,
+					"min": 0,
+					"name": "alert_id",
+					"pattern": "",
+					"presentable": false,
+					"primaryKey": false,
+					"required": false,
+					"system": false,
+					"type": "text"
+				},
+				{
+					"autogeneratePattern": "",
+					"hidden": false,
+					"id": "text1579384326",
+					"max": 0,
+					"min": 0,
+					"name": "name",
+					"pattern": "",
+					"presentable": false,
+					"primaryKey": false,
+					"required": true,
+					"system": false,
+					"type": "text"
+				},
+				{
+					"hidden": false,
+					"id": "number494360628",
+					"max": null,
+					"min": null,
+					"name": "value",
+					"onlyInt": false,
+					"presentable": false,
+					"required": false,
+					"system": false,
+					"type": "number"
+				},
+				{
+					"hidden": false,
+					"id": "autodate2990389176",
+					"name": "created",
+					"onCreate": true,
+					"onUpdate": false,
+					"presentable": false,
+					"system": false,
+					"type": "autodate"
+				},
+				{
+					"hidden": false,
+					"id": "date2276568630",
+					"max": "",
+					"min": "",
+					"name": "resolved",
+					"presentable": false,
+					"required": false,
+					"system": false,
+					"type": "date"
+				}
+		],
+		"indexes": [
+			"CREATE INDEX ` + "`" + `idx_YdGnup5aqB` + "`" + ` ON ` + "`" + `alerts_history` + "`" + ` (` + "`" + `user` + "`" + `)",
+			"CREATE INDEX ` + "`" + `idx_taLet9VdME` + "`" + ` ON ` + "`" + `alerts_history` + "`" + ` (` + "`" + `created` + "`" + `)"
+		],
+		"system": false
+	},
 	{
 		"id": "juohu4jipgc13v7",
 		"listRule": "@request.auth.id != \"\"",
@@ -236,6 +348,88 @@ func init() {
 		],
 		"system": false
 	},
+	{
+		"id": "pbc_3663931638",
+		"listRule": "@request.auth.id != \"\" && system.users.id ?= @request.auth.id",
+		"viewRule": "@request.auth.id != \"\" && system.users.id ?= @request.auth.id",
+		"createRule": "@request.auth.id != \"\" && system.users.id ?= @request.auth.id && @request.auth.role != \"readonly\"",
+		"updateRule": "@request.auth.id != \"\" && system.users.id ?= @request.auth.id && @request.auth.role != \"readonly\"",
+		"deleteRule": null,
+		"name": "fingerprints",
+		"type": "base",
+		"fields": [
+			{
+				"autogeneratePattern": "[a-z0-9]{9}",
+				"hidden": false,
+				"id": "text3208210256",
+				"max": 15,
+				"min": 9,
+				"name": "id",
+				"pattern": "^[a-z0-9]+$",
+				"presentable": false,
+				"primaryKey": true,
+				"required": true,
+				"system": true,
+				"type": "text"
+			},
+			{
+				"cascadeDelete": true,
+				"collectionId": "2hz5ncl8tizk5nx",
+				"hidden": false,
+				"id": "relation3377271179",
+				"maxSelect": 1,
+				"minSelect": 0,
+				"name": "system",
+				"presentable": false,
+				"required": true,
+				"system": false,
+				"type": "relation"
+			},
+			{
+				"autogeneratePattern": "[a-zA-Z9-9]{20}",
+				"hidden": false,
+				"id": "text1597481275",
+				"max": 255,
+				"min": 9,
+				"name": "token",
+				"pattern": "",
+				"presentable": false,
+				"primaryKey": false,
+				"required": true,
+				"system": false,
+				"type": "text"
+			},
+			{
+				"autogeneratePattern": "",
+				"hidden": false,
+				"id": "text4228609354",
+				"max": 255,
+				"min": 9,
+				"name": "fingerprint",
+				"pattern": "",
+				"presentable": false,
+				"primaryKey": false,
+				"required": false,
+				"system": false,
+				"type": "text"
+			},
+			{
+				"hidden": false,
+				"id": "autodate3332085495",
+				"name": "updated",
+				"onCreate": true,
+				"onUpdate": true,
+				"presentable": false,
+				"system": false,
+				"type": "autodate"
+			}
+		],
+		"indexes": [
+			"CREATE INDEX ` + "`" + `idx_p9qZlu26po` + "`" + ` ON ` + "`" + `fingerprints` + "`" + ` (` + "`" + `token` + "`" + `)",
+			"CREATE UNIQUE INDEX ` + "`" + `idx_ngboulGMYw` + "`" + ` ON ` + "`" + `fingerprints` + "`" + ` (` + "`" + `system` + "`" + `)"
+		],
+		"system": false
+	},
 	{
 		"id": "ej9oowivz8b2mht",
 		"listRule": "@request.auth.id != \"\"",
@@ -669,7 +863,37 @@ func init() {
 	}
 ]`
 
-		return app.ImportCollectionsByMarshaledJSON([]byte(jsonData), false)
+		err := app.ImportCollectionsByMarshaledJSON([]byte(jsonData), false)
+		if err != nil {
+			return err
+		}
+
+		// Get all systems that don't have fingerprint records
+		var systemIds []string
+		err = app.DB().NewQuery(`
+			SELECT s.id FROM systems s
+			LEFT JOIN fingerprints f ON s.id = f.system
+			WHERE f.system IS NULL
+		`).Column(&systemIds)
+		if err != nil {
+			return err
+		}
+		// Create fingerprint records with unique UUID tokens for each system
+		for _, systemId := range systemIds {
+			token := uuid.New().String()
+			_, err = app.DB().NewQuery(`
+				INSERT INTO fingerprints (system, token)
+				VALUES ({:system}, {:token})
+			`).Bind(map[string]any{
+				"system": systemId,
+				"token":  token,
+			}).Execute()
+			if err != nil {
+				return err
+			}
+		}
+
+		return nil
 	}, func(app core.App) error {
 		return nil
 	})

beszel/migrations/initial-settings.go

@@ -5,7 +5,7 @@ import (
 	m "github.com/pocketbase/pocketbase/migrations"
 )
 
-var (
+const (
 	TempAdminEmail = "_@b.b"
 )
 

beszel/site/index.html

@@ -4,12 +4,13 @@
 		<meta charset="UTF-8" />
 		<link rel="manifest" href="./static/manifest.json" />
 		<link rel="icon" type="image/svg+xml" href="./static/favicon.svg" />
-		<meta name="viewport" content="width=device-width, initial-scale=1.0" />
+		<meta name="viewport" content="width=device-width, initial-scale=1.0,maximum-scale=1.0, user-scalable=no, viewport-fit=cover" />
 		<title>Beszel</title>
 		<script>
 			globalThis.BESZEL = {
 				BASE_PATH: "%BASE_URL%",
-				HUB_VERSION: "{{V}}"
+				HUB_VERSION: "{{V}}",
+				HUB_URL: "{{HUB_URL}}"
 			}
 		</script>
 	</head>

beszel/site/package.json

@@ -1,10 +1,10 @@
 {
 	"name": "beszel",
 	"private": true,
-	"version": "0.11.1",
+	"version": "0.12.6",
 	"type": "module",
 	"scripts": {
-		"dev": "vite",
+		"dev": "vite --host",
 		"build": "lingui extract --overwrite && lingui compile && vite build",
 		"preview": "vite preview",
 		"sync": "lingui extract --overwrite && lingui compile",
@@ -13,54 +13,54 @@
 	"dependencies": {
 		"@henrygd/queue": "^1.0.7",
 		"@henrygd/semaphore": "^0.0.2",
-		"@lingui/detect-locale": "^5.2.0",
+		"@lingui/detect-locale": "^5.4.1",
-		"@lingui/macro": "^5.2.0",
+		"@lingui/macro": "^5.4.1",
-		"@lingui/react": "^5.2.0",
+		"@lingui/react": "^5.4.1",
 		"@nanostores/react": "^0.7.3",
 		"@nanostores/router": "^0.11.0",
-		"@radix-ui/react-alert-dialog": "^1.1.6",
+		"@radix-ui/react-alert-dialog": "^1.1.15",
-		"@radix-ui/react-checkbox": "^1.1.4",
+		"@radix-ui/react-checkbox": "^1.3.3",
-		"@radix-ui/react-dialog": "^1.1.6",
+		"@radix-ui/react-dialog": "^1.1.15",
-		"@radix-ui/react-direction": "^1.1.0",
+		"@radix-ui/react-direction": "^1.1.1",
-		"@radix-ui/react-dropdown-menu": "^2.1.6",
+		"@radix-ui/react-dropdown-menu": "^2.1.16",
-		"@radix-ui/react-label": "^2.1.2",
+		"@radix-ui/react-label": "^2.1.7",
-		"@radix-ui/react-select": "^2.1.6",
+		"@radix-ui/react-select": "^2.2.6",
-		"@radix-ui/react-separator": "^1.1.2",
+		"@radix-ui/react-separator": "^1.1.7",
-		"@radix-ui/react-slider": "^1.2.3",
+		"@radix-ui/react-slider": "^1.3.6",
-		"@radix-ui/react-slot": "^1.1.2",
+		"@radix-ui/react-slot": "^1.2.3",
-		"@radix-ui/react-switch": "^1.1.3",
+		"@radix-ui/react-switch": "^1.2.6",
-		"@radix-ui/react-tabs": "^1.1.3",
+		"@radix-ui/react-tabs": "^1.1.13",
-		"@radix-ui/react-toast": "^1.2.6",
+		"@radix-ui/react-toast": "^1.2.15",
-		"@radix-ui/react-tooltip": "^1.1.8",
+		"@radix-ui/react-tooltip": "^1.2.8",
-		"@tanstack/react-table": "^8.21.2",
+		"@tanstack/react-table": "^8.21.3",
+		"@tanstack/react-virtual": "^3.13.12",
 		"class-variance-authority": "^0.7.1",
 		"clsx": "^2.1.1",
-		"cmdk": "^1.0.4",
+		"cmdk": "^1.1.1",
 		"d3-time": "^3.1.0",
 		"lucide-react": "^0.452.0",
 		"nanostores": "^0.11.4",
-		"pocketbase": "^0.25.2",
+		"pocketbase": "^0.26.2",
-		"react": "^18.3.1",
+		"react": "^19.1.1",
-		"react-dom": "^18.3.1",
+		"react-dom": "^19.1.1",
-		"recharts": "^2.15.1",
+		"recharts": "^2.15.4",
-		"tailwind-merge": "^2.6.0",
+		"tailwind-merge": "^3.3.1",
-		"tailwindcss-animate": "^1.0.7",
+		"valibot": "^0.42.1"
-		"valibot": "^0.42.0"
 	},
 	"devDependencies": {
-		"@lingui/cli": "^5.2.0",
+		"@lingui/cli": "^5.4.1",
-		"@lingui/swc-plugin": "^5.5.0",
+		"@lingui/swc-plugin": "^5.6.1",
-		"@lingui/vite-plugin": "^5.2.0",
+		"@lingui/vite-plugin": "^5.4.1",
-		"@types/bun": "^1.2.4",
+		"@tailwindcss/container-queries": "^0.1.1",
-		"@types/react": "^18.3.1",
+		"@tailwindcss/vite": "^4.1.12",
-		"@types/react-dom": "^18.3.1",
+		"@types/bun": "^1.2.20",
-		"@vitejs/plugin-react-swc": "^3.8.0",
+		"@types/react": "^19.1.11",
-		"autoprefixer": "^10.4.20",
+		"@types/react-dom": "^19.1.7",
-		"postcss": "^8.5.3",
+		"@vitejs/plugin-react-swc": "^4.0.1",
-		"tailwindcss": "^3.4.17",
+		"tailwindcss": "^4.1.12",
-		"tailwindcss-rtl": "^0.9.0",
+		"tw-animate-css": "^1.3.7",
-		"typescript": "^5.8.2",
+		"typescript": "^5.9.2",
-		"vite": "^6.2.0"
+		"vite": "^7.1.3"
 	},
 	"overrides": {
 		"@nanostores/router": {

beszel/site/postcss.config.js

@@ -1,6 +0,0 @@
-export default {
-	plugins: {
-		tailwindcss: {},
-		autoprefixer: {},
-	},
-}

beszel/site/src/components/add-system.tsx

@@ -11,20 +11,29 @@ import {
 	DialogTrigger,
 } from "@/components/ui/dialog"
 import { Tabs, TabsContent, TabsList, TabsTrigger } from "@/components/ui/tabs"
-import { Tooltip, TooltipContent, TooltipProvider, TooltipTrigger } from "@/components/ui/tooltip"
-
 import { Input } from "@/components/ui/input"
 import { Label } from "@/components/ui/label"
-import { $publicKey, pb } from "@/lib/stores"
+import { $publicKey } from "@/lib/stores"
-import { cn, copyToClipboard, isReadOnlyUser, useLocalStorage } from "@/lib/utils"
+import { cn, generateToken, tokenMap, useBrowserStorage } from "@/lib/utils"
-import { i18n } from "@lingui/core"
+import { pb, isReadOnlyUser } from "@/lib/api"
 import { useStore } from "@nanostores/react"
-import { ChevronDownIcon, Copy, ExternalLinkIcon, PlusIcon } from "lucide-react"
+import { ChevronDownIcon, ExternalLinkIcon, PlusIcon } from "lucide-react"
-import { memo, useRef, useState } from "react"
+import { memo, useEffect, useRef, useState } from "react"
-import { basePath, navigate } from "./router"
+import { $router, basePath, Link, navigate } from "./router"
-import { DropdownMenu, DropdownMenuContent, DropdownMenuItem, DropdownMenuTrigger } from "./ui/dropdown-menu"
 import { SystemRecord } from "@/types"
+import { SystemStatus } from "@/lib/enums"
 import { AppleIcon, DockerIcon, TuxIcon, WindowsIcon } from "./ui/icons"
+import { InputCopy } from "./ui/input-copy"
+import { getPagePath } from "@nanostores/router"
+import {
+	copyDockerCompose,
+	copyDockerRun,
+	copyLinuxCommand,
+	copyWindowsCommand,
+	DropdownItem,
+	InstallDropdown,
+} from "./install-dropdowns"
+import { DropdownMenu, DropdownMenuTrigger } from "./ui/dropdown-menu"
 
 export function AddSystemButton({ className }: { className?: string }) {
 	const [open, setOpen] = useState(false)
@@ -51,44 +60,11 @@ export function AddSystemButton({ className }: { className?: string }) {
 	)
 }
 
-function copyDockerCompose(port = "45876", publicKey: string) {
+/**
-	copyToClipboard(`services:
+ * Token to be used for the next system.
-  beszel-agent:
+ * Prevents token changing if user copies config, then closes dialog and opens again.
-    image: "henrygd/beszel-agent"
+ */
-    container_name: "beszel-agent"
+let nextSystemToken: string | null = null
-    restart: unless-stopped
-    network_mode: host
-    volumes:
-      - /var/run/docker.sock:/var/run/docker.sock:ro
-      # monitor other disks / partitions by mounting a folder in /extra-filesystems
-      # - /mnt/disk/.beszel:/extra-filesystems/sda1:ro
-    environment:
-      LISTEN: ${port}
-      KEY: "${publicKey}"`)
-}
-
-function copyDockerRun(port = "45876", publicKey: string) {
-	copyToClipboard(
-		`docker run -d --name beszel-agent --network host --restart unless-stopped -v /var/run/docker.sock:/var/run/docker.sock:ro -e KEY="${publicKey}" -e LISTEN=${port} henrygd/beszel-agent:latest`
-	)
-}
-
-function copyLinuxCommand(port = "45876", publicKey: string, brew = false) {
-	let cmd = `curl -sL https://get.beszel.dev${
-		brew ? "/brew" : ""
-	} -o /tmp/install-agent.sh && chmod +x /tmp/install-agent.sh && /tmp/install-agent.sh -p ${port} -k "${publicKey}"`
-	// brew script does not support --china-mirrors
-	if (!brew && (i18n.locale + navigator.language).includes("zh-CN")) {
-		cmd += ` --china-mirrors`
-	}
-	copyToClipboard(cmd)
-}
-
-function copyWindowsCommand(port = "45876", publicKey: string) {
-	copyToClipboard(
-		`Set-ExecutionPolicy -ExecutionPolicy RemoteSigned -Scope CurrentUser; & iwr -useb https://get.beszel.dev -OutFile "$env:TEMP\\install-agent.ps1"; & "$env:TEMP\\install-agent.ps1" -Key "${publicKey}" -Port ${port}`
-	)
-}
 
 /**
  * SystemDialog component for adding or editing a system.
@@ -96,12 +72,32 @@ function copyWindowsCommand(port = "45876", publicKey: string) {
  * @param {function} props.setOpen - Function to set the open state of the dialog.
  * @param {SystemRecord} [props.system] - Optional system record for editing an existing system.
  */
-export const SystemDialog = memo(({ setOpen, system }: { setOpen: (open: boolean) => void; system?: SystemRecord }) => {
+export const SystemDialog = ({ setOpen, system }: { setOpen: (open: boolean) => void; system?: SystemRecord }) => {
 	const publicKey = useStore($publicKey)
 	const port = useRef<HTMLInputElement>(null)
 	const [hostValue, setHostValue] = useState(system?.host ?? "")
 	const isUnixSocket = hostValue.startsWith("/")
-	const [tab, setTab] = useLocalStorage("as-tab", "docker")
+	const [tab, setTab] = useBrowserStorage("as-tab", "docker")
+	const [token, setToken] = useState(system?.token ?? "")
+
+	useEffect(() => {
+		;(async () => {
+			// if no system, generate a new token
+			if (!system) {
+				nextSystemToken ||= generateToken()
+				return setToken(nextSystemToken)
+			}
+			// if system exists,get the token from the fingerprint record
+			if (tokenMap.has(system.id)) {
+				return setToken(tokenMap.get(system.id)!)
+			}
+			const { token } = await pb.collection("fingerprints").getFirstListItem(`system = "${system.id}"`, {
+				fields: "token",
+			})
+			tokenMap.set(system.id, token)
+			setToken(token)
+		})()
+	}, [system?.id])
 
 	async function handleSubmit(e: SubmitEvent) {
 		e.preventDefault()
@@ -111,14 +107,20 @@ export const SystemDialog = memo(({ setOpen, system }: { setOpen: (open: boolean
 		try {
 			setOpen(false)
 			if (system) {
-				await pb.collection("systems").update(system.id, { ...data, status: "pending" })
+				await pb.collection("systems").update(system.id, { ...data, status: SystemStatus.Pending })
 			} else {
-				await pb.collection("systems").create(data)
+				const createdSystem = await pb.collection("systems").create(data)
+				await pb.collection("fingerprints").create({
+					system: createdSystem.id,
+					token,
+				})
+				// Reset the current token after successful system
+				// creation so next system gets a new token
+				nextSystemToken = null
 			}
 			navigate(basePath)
-			// console.log(record)
 		} catch (e) {
-			console.log(e)
+			console.error(e)
 		}
 	}
 
@@ -131,7 +133,7 @@ export const SystemDialog = memo(({ setOpen, system }: { setOpen: (open: boolean
 		>
 			<Tabs defaultValue={tab} onValueChange={setTab}>
 				<DialogHeader>
-					<DialogTitle className="mb-2">
+					<DialogTitle className="mb-2 max-w-100 truncate pr-8">
 						{system ? `${t`Edit`} ${system?.name}` : <Trans>Add New System</Trans>}
 					</DialogTitle>
 					<TabsList className="grid w-full grid-cols-2">
@@ -143,18 +145,35 @@ export const SystemDialog = memo(({ setOpen, system }: { setOpen: (open: boolean
 				</DialogHeader>
 				{/* Docker (set tab index to prevent auto focusing content in edit system dialog) */}
 				<TabsContent value="docker" tabIndex={-1}>
-					<DialogDescription className="mb-4 leading-normal w-0 min-w-full">
+					<DialogDescription className="mb-3 leading-relaxed w-0 min-w-full">
 						<Trans>
-							The agent must be running on the system to connect. Copy the
+							Copy the
-							<code className="bg-muted px-1 rounded-sm leading-3">docker-compose.yml</code> for the agent below.
+							<code className="bg-muted px-1 rounded-sm leading-3">docker-compose.yml</code> content for the agent
+							below, or register agents automatically with a{" "}
+							<Link
+								onClick={() => setOpen(false)}
+								href={getPagePath($router, "settings", { name: "tokens" })}
+								className="link"
+							>
+								universal token
+							</Link>
+							.
 						</Trans>
 					</DialogDescription>
 				</TabsContent>
 				{/* Binary */}
 				<TabsContent value="binary" tabIndex={-1}>
-					<DialogDescription className="mb-4 leading-normal w-0 min-w-full">
+					<DialogDescription className="mb-3 leading-relaxed w-0 min-w-full">
 						<Trans>
-							The agent must be running on the system to connect. Copy the installation command for the agent below.
+							Copy the installation command for the agent below, or register agents automatically with a{" "}
+							<Link
+								onClick={() => setOpen(false)}
+								href={getPagePath($router, "settings", { name: "tokens" })}
+								className="link"
+							>
+								universal token
+							</Link>
+							.
 						</Trans>
 					</DialogDescription>
 				</TabsContent>
@@ -190,46 +209,27 @@ export const SystemDialog = memo(({ setOpen, system }: { setOpen: (open: boolean
 						<Label htmlFor="pkey" className="xs:text-end whitespace-pre">
 							<Trans comment="Use 'Key' if your language requires many more characters">Public Key</Trans>
 						</Label>
-						<div className="relative">
+						<InputCopy value={publicKey} id="pkey" name="pkey" />
-							<Input readOnly id="pkey" value={publicKey} required></Input>
+						<Label htmlFor="tkn" className="xs:text-end whitespace-pre">
-							<div
+							<Trans>Token</Trans>
-								className={
+						</Label>
-									"h-6 w-24 bg-gradient-to-r rtl:bg-gradient-to-l from-transparent to-background to-65% absolute top-2 end-1 pointer-events-none"
+						<InputCopy value={token} id="tkn" name="tkn" />
-								}
-							></div>
-							<TooltipProvider delayDuration={100}>
-								<Tooltip>
-									<TooltipTrigger asChild>
-										<Button
-											type="button"
-											variant={"link"}
-											className="absolute end-0 top-0"
-											onClick={() => copyToClipboard(publicKey)}
-										>
-											<Copy className="size-4" />
-										</Button>
-									</TooltipTrigger>
-									<TooltipContent>
-										<p>
-											<Trans>Click to copy</Trans>
-										</p>
-									</TooltipContent>
-								</Tooltip>
-							</TooltipProvider>
-						</div>
 					</div>
 					<DialogFooter className="flex justify-end gap-x-2 gap-y-3 flex-col mt-5">
 						{/* Docker */}
 						<TabsContent value="docker" className="contents">
 							<CopyButton
 								text={t({ message: "Copy docker compose", context: "Button to copy docker compose file content" })}
-								onClick={() => copyDockerCompose(isUnixSocket ? hostValue : port.current?.value, publicKey)}
+								onClick={async () =>
+									copyDockerCompose(isUnixSocket ? hostValue : port.current?.value, publicKey, token)
+								}
 								icon={<DockerIcon className="size-4 -me-0.5" />}
 								dropdownItems={[
 									{
 										text: t({ message: "Copy docker run", context: "Button to copy docker run command" }),
-										onClick: () => copyDockerRun(isUnixSocket ? hostValue : port.current?.value, publicKey),
+										onClick: async () =>
-										icons: [<DockerIcon className="size-4" />],
+											copyDockerRun(isUnixSocket ? hostValue : port.current?.value, publicKey, token),
+										icons: [DockerIcon],
 									},
 								]}
 							/>
@@ -239,22 +239,24 @@ export const SystemDialog = memo(({ setOpen, system }: { setOpen: (open: boolean
 							<CopyButton
 								text={t`Copy Linux command`}
 								icon={<TuxIcon className="size-4" />}
-								onClick={() => copyLinuxCommand(isUnixSocket ? hostValue : port.current?.value, publicKey)}
+								onClick={async () => copyLinuxCommand(isUnixSocket ? hostValue : port.current?.value, publicKey, token)}
 								dropdownItems={[
 									{
 										text: t({ message: "Homebrew command", context: "Button to copy install command" }),
-										onClick: () => copyLinuxCommand(isUnixSocket ? hostValue : port.current?.value, publicKey, true),
+										onClick: async () =>
-										icons: [<AppleIcon className="size-4" />, <TuxIcon className="w-4 h-4" />],
+											copyLinuxCommand(isUnixSocket ? hostValue : port.current?.value, publicKey, token, true),
+										icons: [AppleIcon, TuxIcon],
 									},
 									{
 										text: t({ message: "Windows command", context: "Button to copy install command" }),
-										onClick: () => copyWindowsCommand(isUnixSocket ? hostValue : port.current?.value, publicKey),
+										onClick: async () =>
-										icons: [<WindowsIcon className="size-4" />],
+											copyWindowsCommand(isUnixSocket ? hostValue : port.current?.value, publicKey, token),
+										icons: [WindowsIcon],
 									},
 									{
 										text: t`Manual setup instructions`,
 										url: "https://beszel.dev/guide/agent-installation#binary",
-										icons: [<ExternalLinkIcon className="size-4" />],
+										icons: [ExternalLinkIcon],
 									},
 								]}
 							/>
@@ -266,20 +268,13 @@ export const SystemDialog = memo(({ setOpen, system }: { setOpen: (open: boolean
 			</Tabs>
 		</DialogContent>
 	)
-})
-
-interface DropdownItem {
-	text: string
-	onClick?: () => void
-	url?: string
-	icons?: React.ReactNode[]
 }
 
 interface CopyButtonProps {
 	text: string
 	onClick: () => void
 	dropdownItems: DropdownItem[]
-	icon?: React.ReactNode
+	icon?: React.ReactElement<any>
 }
 
 const CopyButton = memo((props: CopyButtonProps) => {
@@ -300,26 +295,7 @@ const CopyButton = memo((props: CopyButtonProps) => {
 						<ChevronDownIcon />
 					</Button>
 				</DropdownMenuTrigger>
-				<DropdownMenuContent align="end">
+				<InstallDropdown items={props.dropdownItems} />
-					{props.dropdownItems.map((item, index) => (
-						<DropdownMenuItem key={index} asChild={!!item.url}>
-							{item.url ? (
-								<a
-									href={item.url}
-									className="cursor-pointer flex items-center gap-1.5"
-									target="_blank"
-									rel="noopener noreferrer"
-								>
-									{item.text} {item.icons?.map((icon) => icon)}
-								</a>
-							) : (
-								<div onClick={item.onClick} className="cursor-pointer flex items-center gap-1.5">
-									{item.text} {item.icons?.map((icon) => icon)}
-								</div>
-							)}
-						</DropdownMenuItem>
-					))}
-				</DropdownMenuContent>
 			</DropdownMenu>
 		</div>
 	)

beszel/site/src/components/alerts-history-columns.tsx

@@ -0,0 +1,167 @@
+import { ColumnDef } from "@tanstack/react-table"
+import { AlertsHistoryRecord } from "@/types"
+import { Button } from "@/components/ui/button"
+import { Badge } from "@/components/ui/badge"
+import { formatShortDate, toFixedFloat, formatDuration, cn } from "@/lib/utils"
+import { alertInfo } from "@/lib/alerts"
+import { Trans } from "@lingui/react/macro"
+import { t } from "@lingui/core/macro"
+
+export const alertsHistoryColumns: ColumnDef<AlertsHistoryRecord>[] = [
+	{
+		accessorKey: "system",
+		enableSorting: true,
+		header: ({ column }) => (
+			<Button variant="ghost" onClick={() => column.toggleSorting(column.getIsSorted() === "asc")}>
+				<Trans>System</Trans>
+			</Button>
+		),
+		cell: ({ row }) => (
+			<div className="ps-2 max-w-60 truncate">{row.original.expand?.system?.name || row.original.system}</div>
+		),
+		filterFn: (row, _, filterValue) => {
+			const display = row.original.expand?.system?.name || row.original.system || ""
+			return display.toLowerCase().includes(filterValue.toLowerCase())
+		},
+	},
+	{
+		// accessorKey: "name",
+		id: "name",
+		accessorFn: (record) => {
+			const name = record.name
+			const info = alertInfo[name]
+			return info?.name().replace("cpu", "CPU") || name
+		},
+		header: ({ column }) => (
+			<Button variant="ghost" onClick={() => column.toggleSorting(column.getIsSorted() === "asc")}>
+				<Trans>Name</Trans>
+			</Button>
+		),
+		cell: ({ getValue, row }) => {
+			let name = getValue() as string
+			const info = alertInfo[row.original.name]
+			const Icon = info?.icon
+
+			return (
+				<span className="flex items-center gap-2 ps-1 min-w-40">
+					{Icon && <Icon className="size-3.5" />}
+					{name}
+				</span>
+			)
+		},
+	},
+	{
+		accessorKey: "value",
+		enableSorting: false,
+		header: () => (
+			<Button variant="ghost">
+				<Trans>Value</Trans>
+			</Button>
+		),
+		cell({ row, getValue }) {
+			const name = row.original.name
+			if (name === "Status") {
+				return <span className="ps-2">{t`Down`}</span>
+			}
+			const value = getValue() as number
+			const unit = alertInfo[name]?.unit
+			return (
+				<span className="tabular-nums ps-2.5">
+					{toFixedFloat(value, value < 10 ? 2 : 1)}
+					{unit}
+				</span>
+			)
+		},
+	},
+	{
+		accessorKey: "state",
+		enableSorting: true,
+		sortingFn: (rowA, rowB) => (rowA.original.resolved ? 1 : 0) - (rowB.original.resolved ? 1 : 0),
+		header: ({ column }) => (
+			<Button variant="ghost" onClick={() => column.toggleSorting(column.getIsSorted() === "asc")}>
+				<Trans comment="Context: alert state (active or resolved)">State</Trans>
+			</Button>
+		),
+		cell: ({ row }) => {
+			const resolved = row.original.resolved
+			return (
+				<Badge
+					className={cn(
+						"capitalize pointer-events-none",
+						resolved
+							? "bg-green-100 text-green-800 border-green-200 dark:opacity-80"
+							: "bg-yellow-100 text-yellow-800 border-yellow-200"
+					)}
+				>
+					{/* {resolved ? <CircleCheckIcon className="size-3 me-0.5" /> : <CircleAlertIcon className="size-3 me-0.5" />} */}
+					{resolved ? <Trans>Resolved</Trans> : <Trans>Active</Trans>}
+				</Badge>
+			)
+		},
+	},
+	{
+		accessorKey: "created",
+		accessorFn: (record) => formatShortDate(record.created),
+		enableSorting: true,
+		invertSorting: true,
+		header: ({ column }) => (
+			<Button variant="ghost" onClick={() => column.toggleSorting(column.getIsSorted() === "asc")}>
+				<Trans comment="Context: date created">Created</Trans>
+			</Button>
+		),
+		cell: ({ getValue, row }) => (
+			<span className="ps-1 tabular-nums tracking-tight" title={`${row.original.created} UTC`}>
+				{getValue() as string}
+			</span>
+		),
+	},
+	{
+		accessorKey: "resolved",
+		enableSorting: true,
+		invertSorting: true,
+		header: ({ column }) => (
+			<Button variant="ghost" onClick={() => column.toggleSorting(column.getIsSorted() === "asc")}>
+				<Trans>Resolved</Trans>
+			</Button>
+		),
+		cell: ({ row, getValue }) => {
+			const resolved = getValue() as string | null
+			if (!resolved) {
+				return null
+			}
+			return (
+				<span className="ps-1 tabular-nums tracking-tight" title={`${row.original.resolved} UTC`}>
+					{formatShortDate(resolved)}
+				</span>
+			)
+		},
+	},
+	{
+		accessorKey: "duration",
+		invertSorting: true,
+		enableSorting: true,
+		sortingFn: (rowA, rowB) => {
+			const aCreated = new Date(rowA.original.created)
+			const bCreated = new Date(rowB.original.created)
+			const aResolved = rowA.original.resolved ? new Date(rowA.original.resolved) : null
+			const bResolved = rowB.original.resolved ? new Date(rowB.original.resolved) : null
+			const aDuration = aResolved ? aResolved.getTime() - aCreated.getTime() : null
+			const bDuration = bResolved ? bResolved.getTime() - bCreated.getTime() : null
+			if (!aDuration && bDuration) return -1
+			if (aDuration && !bDuration) return 1
+			return (aDuration || 0) - (bDuration || 0)
+		},
+		header: ({ column }) => (
+			<Button variant="ghost" onClick={() => column.toggleSorting(column.getIsSorted() === "asc")}>
+				<Trans>Duration</Trans>
+			</Button>
+		),
+		cell: ({ row }) => {
+			const duration = formatDuration(row.original.created, row.original.resolved)
+			if (!duration) {
+				return null
+			}
+			return <span className="ps-2">{duration}</span>
+		},
+	},
+]

beszel/site/src/components/alerts/alert-button.tsx

@@ -1,132 +1,36 @@
-import { t } from "@lingui/core/macro";
+import { t } from "@lingui/core/macro"
-import { Trans } from "@lingui/react/macro";
 import { memo, useMemo, useState } from "react"
 import { useStore } from "@nanostores/react"
 import { $alerts } from "@/lib/stores"
-import {
+import { BellIcon } from "lucide-react"
-	Dialog,
+import { cn } from "@/lib/utils"
-	DialogTrigger,
-	DialogContent,
-	DialogDescription,
-	DialogHeader,
-	DialogTitle,
-} from "@/components/ui/dialog"
-import { BellIcon, GlobeIcon, ServerIcon } from "lucide-react"
-import { alertInfo, cn } from "@/lib/utils"
 import { Button } from "@/components/ui/button"
-import { AlertRecord, SystemRecord } from "@/types"
+import { SystemRecord } from "@/types"
-import { Link } from "../router"
+import { AlertDialogContent } from "./alerts-sheet"
-import { Tabs, TabsContent, TabsList, TabsTrigger } from "@/components/ui/tabs"
+import { Sheet, SheetContent, SheetTrigger } from "@/components/ui/sheet"
-import { Checkbox } from "../ui/checkbox"
-import { SystemAlert, SystemAlertGlobal } from "./alerts-system"
 
 export default memo(function AlertsButton({ system }: { system: SystemRecord }) {
-	const alerts = useStore($alerts)
 	const [opened, setOpened] = useState(false)
+	const alerts = useStore($alerts)
 
-	const hasAlert = alerts.some((alert) => alert.system === system.id)
+	const hasSystemAlert = alerts[system.id]?.size > 0
-
 	return useMemo(
 		() => (
-			<Dialog>
+			<Sheet>
-				<DialogTrigger asChild>
+				<SheetTrigger asChild>
 					<Button variant="ghost" size="icon" aria-label={t`Alerts`} data-nolink onClick={() => setOpened(true)}>
 						<BellIcon
 							className={cn("h-[1.2em] w-[1.2em] pointer-events-none", {
-								"fill-primary": hasAlert,
+								"fill-primary": hasSystemAlert,
 							})}
 						/>
 					</Button>
-				</DialogTrigger>
+				</SheetTrigger>
-				<DialogContent className="max-h-full overflow-auto max-w-[35rem]">
+				<SheetContent className="max-h-full overflow-auto w-145 !max-w-full p-4 sm:p-6">
 					{opened && <AlertDialogContent system={system} />}
-				</DialogContent>
+				</SheetContent>
-			</Dialog>
+			</Sheet>
 		),
-		[opened, hasAlert]
+		[opened, hasSystemAlert]
 	)
 })
-
-function AlertDialogContent({ system }: { system: SystemRecord }) {
-	const alerts = useStore($alerts)
-	const [overwriteExisting, setOverwriteExisting] = useState<boolean | "indeterminate">(false)
-
-	// alertsSignature changes only when alerts for this system change
-	let alertsSignature = ""
-	const systemAlerts = alerts.filter((alert) => {
-		if (alert.system === system.id) {
-			alertsSignature += alert.name + alert.min + alert.value
-			return true
-		}
-		return false
-	}) as AlertRecord[]
-
-	return useMemo(() => {
-		// console.log("render modal", system.name, alertsSignature)
-		const data = Object.keys(alertInfo).map((name) => {
-			const alert = alertInfo[name as keyof typeof alertInfo]
-			return {
-				name: name as keyof typeof alertInfo,
-				alert,
-				system,
-			}
-		})
-
-		return (
-			<>
-				<DialogHeader>
-					<DialogTitle className="text-xl">
-						<Trans>Alerts</Trans>
-					</DialogTitle>
-					<DialogDescription>
-						<Trans>
-							See{" "}
-							<Link href="/settings/notifications" className="link">
-								notification settings
-							</Link>{" "}
-							to configure how you receive alerts.
-						</Trans>
-					</DialogDescription>
-				</DialogHeader>
-				<Tabs defaultValue="system">
-					<TabsList className="mb-1 -mt-0.5">
-						<TabsTrigger value="system">
-							<ServerIcon className="me-2 h-3.5 w-3.5" />
-							{system.name}
-						</TabsTrigger>
-						<TabsTrigger value="global">
-							<GlobeIcon className="me-1.5 h-3.5 w-3.5" />
-							<Trans>All Systems</Trans>
-						</TabsTrigger>
-					</TabsList>
-					<TabsContent value="system">
-						<div className="grid gap-3">
-							{data.map((d) => (
-								<SystemAlert key={d.name} system={system} data={d} systemAlerts={systemAlerts} />
-							))}
-						</div>
-					</TabsContent>
-					<TabsContent value="global">
-						<label
-							htmlFor="ovw"
-							className="mb-3 flex gap-2 items-center justify-center cursor-pointer border rounded-sm py-3 px-4 border-destructive text-destructive font-semibold text-sm"
-						>
-							<Checkbox
-								id="ovw"
-								className="text-destructive border-destructive data-[state=checked]:bg-destructive"
-								checked={overwriteExisting}
-								onCheckedChange={setOverwriteExisting}
-							/>
-							<Trans>Overwrite existing alerts</Trans>
-						</label>
-						<div className="grid gap-3">
-							{data.map((d) => (
-								<SystemAlertGlobal key={d.name} data={d} overwrite={overwriteExisting} />
-							))}
-						</div>
-					</TabsContent>
-				</Tabs>
-			</>
-		)
-	}, [alertsSignature, overwriteExisting])
-}

beszel/site/src/components/alerts/alerts-sheet.tsx

@@ -0,0 +1,299 @@
+import { t } from "@lingui/core/macro"
+import { Trans, Plural } from "@lingui/react/macro"
+import { $alerts, $systems } from "@/lib/stores"
+import { cn, debounce } from "@/lib/utils"
+import { alertInfo } from "@/lib/alerts"
+import { Switch } from "@/components/ui/switch"
+import { AlertInfo, AlertRecord, SystemRecord } from "@/types"
+import { lazy, memo, Suspense, useMemo, useState } from "react"
+import { toast } from "@/components/ui/use-toast"
+import { useStore } from "@nanostores/react"
+import { getPagePath } from "@nanostores/router"
+import { Checkbox } from "@/components/ui/checkbox"
+import { DialogTitle, DialogDescription } from "@/components/ui/dialog"
+import { Tabs, TabsList, TabsTrigger, TabsContent } from "@/components/ui/tabs"
+import { ServerIcon, GlobeIcon } from "lucide-react"
+import { $router, Link } from "@/components/router"
+import { DialogHeader } from "@/components/ui/dialog"
+import { pb } from "@/lib/api"
+
+const Slider = lazy(() => import("@/components/ui/slider"))
+
+const endpoint = "/api/beszel/user-alerts"
+
+const alertDebounce = 100
+
+const alertKeys = Object.keys(alertInfo) as (keyof typeof alertInfo)[]
+
+const failedUpdateToast = (error: unknown) => {
+	console.error(error)
+	toast({
+		title: t`Failed to update alert`,
+		description: t`Please check logs for more details.`,
+		variant: "destructive",
+	})
+}
+
+/** Create or update alerts for a given name and systems */
+const upsertAlerts = debounce(
+	async ({ name, value, min, systems }: { name: string; value: number; min: number; systems: string[] }) => {
+		try {
+			await pb.send<{ success: boolean }>(endpoint, {
+				method: "POST",
+				// overwrite is always true because we've done filtering client side
+				body: { name, value, min, systems, overwrite: true },
+			})
+		} catch (error) {
+			failedUpdateToast(error)
+		}
+	},
+	alertDebounce
+)
+
+/** Delete alerts for a given name and systems */
+const deleteAlerts = debounce(async ({ name, systems }: { name: string; systems: string[] }) => {
+	try {
+		await pb.send<{ success: boolean }>(endpoint, {
+			method: "DELETE",
+			body: { name, systems },
+		})
+	} catch (error) {
+		failedUpdateToast(error)
+	}
+}, alertDebounce)
+
+export const AlertDialogContent = memo(function AlertDialogContent({ system }: { system: SystemRecord }) {
+	const alerts = useStore($alerts)
+	const [overwriteExisting, setOverwriteExisting] = useState<boolean | "indeterminate">(false)
+	const [currentTab, setCurrentTab] = useState("system")
+
+	const systemAlerts = alerts[system.id] ?? new Map()
+
+	// We need to keep a copy of alerts when we switch to global tab. If we always compare to
+	// current alerts, it will only be updated when first checked, then won't be updated because
+	// after that it exists.
+	const alertsWhenGlobalSelected = useMemo(() => {
+		return currentTab === "global" ? structuredClone(alerts) : alerts
+	}, [currentTab])
+
+	return (
+		<>
+			<DialogHeader>
+				<DialogTitle className="text-xl">
+					<Trans>Alerts</Trans>
+				</DialogTitle>
+				<DialogDescription>
+					<Trans>
+						See{" "}
+						<Link href={getPagePath($router, "settings", { name: "notifications" })} className="link">
+							notification settings
+						</Link>{" "}
+						to configure how you receive alerts.
+					</Trans>
+				</DialogDescription>
+			</DialogHeader>
+			<Tabs defaultValue="system" onValueChange={setCurrentTab}>
+				<TabsList className="mb-1 -mt-0.5">
+					<TabsTrigger value="system">
+						<ServerIcon className="me-2 h-3.5 w-3.5" />
+						<span className="truncate max-w-60">{system.name}</span>
+					</TabsTrigger>
+					<TabsTrigger value="global">
+						<GlobeIcon className="me-1.5 h-3.5 w-3.5" />
+						<Trans>All Systems</Trans>
+					</TabsTrigger>
+				</TabsList>
+				<TabsContent value="system">
+					<div className="grid gap-3">
+						{alertKeys.map((name) => (
+							<AlertContent
+								key={name}
+								alertKey={name}
+								data={alertInfo[name as keyof typeof alertInfo]}
+								alert={systemAlerts.get(name)}
+								system={system}
+							/>
+						))}
+					</div>
+				</TabsContent>
+				<TabsContent value="global">
+					<label
+						htmlFor="ovw"
+						className="mb-3 flex gap-2 items-center justify-center cursor-pointer border rounded-sm py-3 px-4 border-destructive text-destructive font-semibold text-sm"
+					>
+						<Checkbox
+							id="ovw"
+							className="text-destructive border-destructive data-[state=checked]:bg-destructive"
+							checked={overwriteExisting}
+							onCheckedChange={setOverwriteExisting}
+						/>
+						<Trans>Overwrite existing alerts</Trans>
+					</label>
+					<div className="grid gap-3">
+						{alertKeys.map((name) => (
+							<AlertContent
+								key={name}
+								alertKey={name}
+								system={system}
+								alert={systemAlerts.get(name)}
+								data={alertInfo[name as keyof typeof alertInfo]}
+								global={true}
+								overwriteExisting={!!overwriteExisting}
+								initialAlertsState={alertsWhenGlobalSelected}
+							/>
+						))}
+					</div>
+				</TabsContent>
+			</Tabs>
+		</>
+	)
+})
+
+export function AlertContent({
+	alertKey,
+	data: alertData,
+	system,
+	alert,
+	global = false,
+	overwriteExisting = false,
+	initialAlertsState = {},
+}: {
+	alertKey: string
+	data: AlertInfo
+	system: SystemRecord
+	alert?: AlertRecord
+	global?: boolean
+	overwriteExisting?: boolean
+	initialAlertsState?: Record<string, Map<string, AlertRecord>>
+}) {
+	const { name } = alertData
+
+	const singleDescription = alertData.singleDesc?.()
+
+	const [checked, setChecked] = useState(global ? false : !!alert)
+	const [min, setMin] = useState(alert?.min || 10)
+	const [value, setValue] = useState(alert?.value || (singleDescription ? 0 : alertData.start ?? 80))
+
+	const Icon = alertData.icon
+
+	/** Get system ids to update */
+	function getSystemIds(): string[] {
+		// if not global, update only the current system
+		if (!global) {
+			return [system.id]
+		}
+		// if global, update all systems when overwriteExisting is true
+		// update only systems without an existing alert when overwriteExisting is false
+		const allSystems = $systems.get()
+		const systemIds: string[] = []
+		for (const system of allSystems) {
+			if (overwriteExisting || !initialAlertsState[system.id]?.has(alertKey)) {
+				systemIds.push(system.id)
+			}
+		}
+		return systemIds
+	}
+
+	function sendUpsert(min: number, value: number) {
+		const systems = getSystemIds()
+		systems.length &&
+			upsertAlerts({
+				name: alertKey,
+				value,
+				min,
+				systems,
+			})
+	}
+
+	return (
+		<div className="rounded-lg border border-muted-foreground/15 hover:border-muted-foreground/20 transition-colors duration-100 group">
+			<label
+				htmlFor={`s${name}`}
+				className={cn("flex flex-row items-center justify-between gap-4 cursor-pointer p-4", {
+					"pb-0": checked,
+				})}
+			>
+				<div className="grid gap-1 select-none">
+					<p className="font-semibold flex gap-3 items-center">
+						<Icon className="h-4 w-4 opacity-85" /> {alertData.name()}
+					</p>
+					{!checked && <span className="block text-sm text-muted-foreground">{alertData.desc()}</span>}
+				</div>
+				<Switch
+					id={`s${name}`}
+					checked={checked}
+					onCheckedChange={(newChecked) => {
+						setChecked(newChecked)
+						if (newChecked) {
+							// if alert checked, create or update alert
+							sendUpsert(min, value)
+						} else {
+							// if unchecked, delete alert (unless global and overwriteExisting is false)
+							deleteAlerts({ name: alertKey, systems: getSystemIds() })
+							// when force deleting all alerts of a type, also remove them from initialAlertsState
+							if (overwriteExisting) {
+								for (const curAlerts of Object.values(initialAlertsState)) {
+									curAlerts.delete(alertKey)
+								}
+							}
+						}
+					}}
+				/>
+			</label>
+			{checked && (
+				<div className="grid sm:grid-cols-2 mt-1.5 gap-5 px-4 pb-5 tabular-nums text-muted-foreground">
+					<Suspense fallback={<div className="h-10" />}>
+						{!singleDescription && (
+							<div>
+								<p id={`v${name}`} className="text-sm block h-8">
+									<Trans>
+										Average exceeds{" "}
+										<strong className="text-foreground">
+											{value}
+											{alertData.unit}
+										</strong>
+									</Trans>
+								</p>
+								<div className="flex gap-3">
+									<Slider
+										aria-labelledby={`v${name}`}
+										defaultValue={[value]}
+										onValueCommit={(val) => sendUpsert(min, val[0])}
+										onValueChange={(val) => setValue(val[0])}
+										step={alertData.step ?? 1}
+										min={alertData.min ?? 1}
+										max={alertData.max ?? 99}
+									/>
+								</div>
+							</div>
+						)}
+						<div className={cn(singleDescription && "col-span-full lowercase")}>
+							<p id={`t${name}`} className="text-sm block h-8 first-letter:uppercase">
+								{singleDescription && (
+									<>
+										{singleDescription}
+										{` `}
+									</>
+								)}
+								<Trans>
+									For <strong className="text-foreground">{min}</strong>{" "}
+									<Plural value={min} one="minute" other="minutes" />
+								</Trans>
+							</p>
+							<div className="flex gap-3">
+								<Slider
+									aria-labelledby={`v${name}`}
+									defaultValue={[min]}
+									onValueCommit={(minVal) => sendUpsert(minVal[0], value)}
+									onValueChange={(val) => setMin(val[0])}
+									min={1}
+									max={60}
+								/>
+							</div>
+						</div>
+					</Suspense>
+				</div>
+			)}
+		</div>
+	)
+}

beszel/site/src/components/alerts/alerts-system.tsx

@@ -1,310 +0,0 @@
-import { t } from "@lingui/core/macro"
-import { Trans, Plural } from "@lingui/react/macro"
-import { $alerts, $systems, pb } from "@/lib/stores"
-import { alertInfo, cn } from "@/lib/utils"
-import { Switch } from "@/components/ui/switch"
-import { AlertInfo, AlertRecord, SystemRecord } from "@/types"
-import { lazy, Suspense, useMemo, useState } from "react"
-import { toast } from "../ui/use-toast"
-import { BatchService } from "pocketbase"
-import { getSemaphore } from "@henrygd/semaphore"
-
-interface AlertData {
-	checked?: boolean
-	val?: number
-	min?: number
-	updateAlert?: (checked: boolean, value: number, min: number) => void
-	name: keyof typeof alertInfo
-	alert: AlertInfo
-	system: SystemRecord
-}
-
-const Slider = lazy(() => import("@/components/ui/slider"))
-
-const failedUpdateToast = () =>
-	toast({
-		title: t`Failed to update alert`,
-		description: t`Please check logs for more details.`,
-		variant: "destructive",
-	})
-
-export function SystemAlert({
-	system,
-	systemAlerts,
-	data,
-}: {
-	system: SystemRecord
-	systemAlerts: AlertRecord[]
-	data: AlertData
-}) {
-	const alert = systemAlerts.find((alert) => alert.name === data.name)
-
-	data.updateAlert = async (checked: boolean, value: number, min: number) => {
-		try {
-			if (alert && !checked) {
-				await pb.collection("alerts").delete(alert.id)
-			} else if (alert && checked) {
-				await pb.collection("alerts").update(alert.id, { value, min, triggered: false })
-			} else if (checked) {
-				pb.collection("alerts").create({
-					system: system.id,
-					user: pb.authStore.record!.id,
-					name: data.name,
-					value: value,
-					min: min,
-				})
-			}
-		} catch (e) {
-			failedUpdateToast()
-		}
-	}
-
-	if (alert) {
-		data.checked = true
-		data.val = alert.value
-		data.min = alert.min || 1
-	}
-
-	return <AlertContent data={data} />
-}
-
-export const SystemAlertGlobal = ({ data, overwrite }: { data: AlertData; overwrite: boolean | "indeterminate" }) => {
-	data.checked = false
-	data.val = data.min = 0
-
-	// set of system ids that have an alert for this name when the component is mounted
-	const existingAlertsSystems = useMemo(() => {
-		const map = new Set<string>()
-		const alerts = $alerts.get()
-		for (const alert of alerts) {
-			if (alert.name === data.name) {
-				map.add(alert.system)
-			}
-		}
-		return map
-	}, [])
-
-	data.updateAlert = async (checked: boolean, value: number, min: number) => {
-		const sem = getSemaphore("alerts")
-		await sem.acquire()
-		try {
-			// if another update is waiting behind, don't start this one
-			if (sem.size() > 1) {
-				return
-			}
-
-			const recordData: Partial<AlertRecord> = {
-				value,
-				min,
-				triggered: false,
-			}
-
-			const batch = batchWrapper("alerts", 25)
-			const systems = $systems.get()
-			const currentAlerts = $alerts.get()
-
-			// map of current alerts with this name right now by system id
-			const currentAlertsSystems = new Map<string, AlertRecord>()
-			for (const alert of currentAlerts) {
-				if (alert.name === data.name) {
-					currentAlertsSystems.set(alert.system, alert)
-				}
-			}
-
-			if (overwrite) {
-				existingAlertsSystems.clear()
-			}
-
-			const processSystem = async (system: SystemRecord): Promise<void> => {
-				const existingAlert = existingAlertsSystems.has(system.id)
-
-				if (!overwrite && existingAlert) {
-					return
-				}
-
-				const currentAlert = currentAlertsSystems.get(system.id)
-
-				// delete existing alert if unchecked
-				if (!checked && currentAlert) {
-					return batch.remove(currentAlert.id)
-				}
-				if (checked && currentAlert) {
-					// update existing alert if checked
-					return batch.update(currentAlert.id, recordData)
-				}
-				if (checked) {
-					// create new alert if checked and not existing
-					return batch.create({
-						system: system.id,
-						user: pb.authStore.record!.id,
-						name: data.name,
-						...recordData,
-					})
-				}
-			}
-
-			// make sure current system is updated in the first batch
-			await processSystem(data.system)
-			for (const system of systems) {
-				if (system.id === data.system.id) {
-					continue
-				}
-				if (sem.size() > 1) {
-					return
-				}
-				await processSystem(system)
-			}
-			await batch.send()
-		} finally {
-			sem.release()
-		}
-	}
-
-	return <AlertContent data={data} />
-}
-
-/**
- * Creates a wrapper for performing batch operations on a specified collection.
- */
-function batchWrapper(collection: string, batchSize: number) {
-	let batch: BatchService | undefined
-	let count = 0
-
-	const create = async <T extends Record<string, any>>(options: T) => {
-		batch ||= pb.createBatch()
-		batch.collection(collection).create(options)
-		if (++count >= batchSize) {
-			await send()
-		}
-	}
-
-	const update = async <T extends Record<string, any>>(id: string, data: T) => {
-		batch ||= pb.createBatch()
-		batch.collection(collection).update(id, data)
-		if (++count >= batchSize) {
-			await send()
-		}
-	}
-
-	const remove = async (id: string) => {
-		batch ||= pb.createBatch()
-		batch.collection(collection).delete(id)
-		if (++count >= batchSize) {
-			await send()
-		}
-	}
-
-	const send = async () => {
-		if (count) {
-			await batch?.send({ requestKey: null })
-			batch = undefined
-			count = 0
-		}
-	}
-
-	return {
-		update,
-		remove,
-		send,
-		create,
-	}
-}
-
-function AlertContent({ data }: { data: AlertData }) {
-	const { name } = data
-
-	const singleDescription = data.alert.singleDesc?.()
-
-	const [checked, setChecked] = useState(data.checked || false)
-	const [min, setMin] = useState(data.min || 10)
-	const [value, setValue] = useState(data.val || (singleDescription ? 0 : 80))
-
-	const Icon = alertInfo[name].icon
-
-	return (
-		<div className="rounded-lg border border-muted-foreground/15 hover:border-muted-foreground/20 transition-colors duration-100 group">
-			<label
-				htmlFor={`s${name}`}
-				className={cn("flex flex-row items-center justify-between gap-4 cursor-pointer p-4", {
-					"pb-0": checked,
-				})}
-			>
-				<div className="grid gap-1 select-none">
-					<p className="font-semibold flex gap-3 items-center">
-						<Icon className="h-4 w-4 opacity-85" /> {data.alert.name()}
-					</p>
-					{!checked && <span className="block text-sm text-muted-foreground">{data.alert.desc()}</span>}
-				</div>
-				<Switch
-					id={`s${name}`}
-					checked={checked}
-					onCheckedChange={(newChecked) => {
-						setChecked(newChecked)
-						data.updateAlert?.(newChecked, value, min)
-					}}
-				/>
-			</label>
-			{checked && (
-				<div className="grid sm:grid-cols-2 mt-1.5 gap-5 px-4 pb-5 tabular-nums text-muted-foreground">
-					<Suspense fallback={<div className="h-10" />}>
-						{!singleDescription && (
-							<div>
-								<p id={`v${name}`} className="text-sm block h-8">
-									<Trans>
-										Average exceeds{" "}
-										<strong className="text-foreground">
-											{value}
-											{data.alert.unit}
-										</strong>
-									</Trans>
-								</p>
-								<div className="flex gap-3">
-									<Slider
-										aria-labelledby={`v${name}`}
-										defaultValue={[value]}
-										onValueCommit={(val) => {
-											data.updateAlert?.(true, val[0], min)
-										}}
-										onValueChange={(val) => {
-											setValue(val[0])
-										}}
-										min={1}
-										max={alertInfo[name].max ?? 99}
-									/>
-								</div>
-							</div>
-						)}
-						<div className={cn(singleDescription && "col-span-full lowercase")}>
-							<p id={`t${name}`} className="text-sm block h-8 first-letter:uppercase">
-								{singleDescription && (
-									<>
-										{singleDescription}
-										{` `}
-									</>
-								)}
-								<Trans>
-									For <strong className="text-foreground">{min}</strong>{" "}
-									<Plural value={min} one="minute" other="minutes" />
-								</Trans>
-							</p>
-							<div className="flex gap-3">
-								<Slider
-									aria-labelledby={`v${name}`}
-									defaultValue={[min]}
-									onValueCommit={(min) => {
-										data.updateAlert?.(true, value, min[0])
-									}}
-									onValueChange={(val) => {
-										setMin(val[0])
-									}}
-									min={1}
-									max={60}
-								/>
-							</div>
-						</div>
-					</Suspense>
-				</div>
-			)}
-		</div>
-	)
-}

beszel/site/src/components/charts/area-chart.tsx

@@ -1,150 +1,94 @@
-import { t } from "@lingui/core/macro"
-
 import { Area, AreaChart, CartesianGrid, YAxis } from "recharts"
 import { ChartContainer, ChartTooltip, ChartTooltipContent, xAxis } from "@/components/ui/chart"
-import {
+import { cn, formatShortDate, chartMargin } from "@/lib/utils"
-	useYAxisWidth,
+import { useYAxisWidth } from "./hooks"
-	cn,
+import { ChartData, SystemStatsRecord } from "@/types"
-	formatShortDate,
+import { useMemo } from "react"
-	toFixedWithoutTrailingZeros,
-	decimalString,
-	chartMargin,
-} from "@/lib/utils"
-// import Spinner from '../spinner'
-import { ChartData } from "@/types"
-import { memo, useMemo } from "react"
-import { useLingui } from "@lingui/react/macro"
 
-/** [label, key, color, opacity] */
+export type DataPoint = {
-type DataKeys = [string, string, number, number]
+	label: string
-
+	dataKey: (data: SystemStatsRecord) => number | undefined
-const getNestedValue = (path: string, max = false, data: any): number | null => {
+	color: number | string
-	// fallback value (obj?.stats?.cpum ? 0 : null) should only come into play when viewing
+	opacity: number
-	// a max value which doesn't exist, or the value was zero and omitted from the stats object.
-	// so we check if cpum is present. if so, return 0 to make sure the zero value is displayed.
-	// if not, return null - there is no max data so do not display anything.
-	return `stats.${path}${max ? "m" : ""}`
-		.split(".")
-		.reduce((acc: any, key: string) => acc?.[key] ?? (data.stats?.cpum ? 0 : null), data)
 }
 
-export default memo(function AreaChartDefault({
+export default function AreaChartDefault({
-	maxToggled = false,
-	unit = " MB/s",
-	chartName,
 	chartData,
 	max,
+	maxToggled,
 	tickFormatter,
 	contentFormatter,
-}: {
+	dataPoints,
-	maxToggled?: boolean
+	domain,
-	unit?: string
+}: // logRender = false,
-	chartName: string
+{
 	chartData: ChartData
 	max?: number
-	tickFormatter?: (value: number) => string
+	maxToggled?: boolean
-	contentFormatter?: (value: number) => string
+	tickFormatter: (value: number, index: number) => string
+	contentFormatter: ({ value, payload }: { value: number; payload: SystemStatsRecord }) => string
+	dataPoints?: DataPoint[]
+	domain?: [number, number]
+	// logRender?: boolean
 }) {
 	const { yAxisWidth, updateYAxisWidth } = useYAxisWidth()
-	const { i18n } = useLingui()
 
-	const { chartTime } = chartData
+	return useMemo(() => {
-
+		if (chartData.systemStats.length === 0) {
-	const showMax = chartTime !== "1h" && maxToggled
+			return null
-
-	const dataKeys: DataKeys[] = useMemo(() => {
-		// [label, key, color, opacity]
-		if (chartName === "CPU Usage") {
-			return [[t`CPU Usage`, "cpu", 1, 0.4]]
-		} else if (chartName === "dio") {
-			return [
-				[t({ message: "Write", comment: "Disk write" }), "dw", 3, 0.3],
-				[t({ message: "Read", comment: "Disk read" }), "dr", 1, 0.3],
-			]
-		} else if (chartName === "bw") {
-			return [
-				[t({ message: "Sent", comment: "Network bytes sent (upload)" }), "ns", 5, 0.2],
-				[t({ message: "Received", comment: "Network bytes received (download)" }), "nr", 2, 0.2],
-			]
-		} else if (chartName.startsWith("efs")) {
-			return [
-				[t`Write`, `${chartName}.w`, 3, 0.3],
-				[t`Read`, `${chartName}.r`, 1, 0.3],
-			]
-		} else if (chartName.startsWith("g.")) {
-			return [chartName.includes("mu") ? [t`Used`, chartName, 2, 0.25] : [t`Usage`, chartName, 1, 0.4]]
 		}
-		return []
+		// if (logRender) {
-	}, [chartName, i18n.locale])
+		// 	console.log("Rendered at", new Date())
-
+		// }
-	// console.log('Rendered at', new Date())
+		return (
-
+			<div>
-	if (chartData.systemStats.length === 0) {
+				<ChartContainer
-		return null
+					className={cn("h-full w-full absolute aspect-auto bg-card opacity-0 transition-opacity", {
-	}
+						"opacity-100": yAxisWidth,
-
-	return (
-		<div>
-			<ChartContainer
-				className={cn("h-full w-full absolute aspect-auto bg-card opacity-0 transition-opacity", {
-					"opacity-100": yAxisWidth,
-				})}
-			>
-				<AreaChart accessibilityLayer data={chartData.systemStats} margin={chartMargin}>
-					<CartesianGrid vertical={false} />
-					<YAxis
-						direction="ltr"
-						orientation={chartData.orientation}
-						className="tracking-tighter"
-						width={yAxisWidth}
-						domain={[0, max ?? "auto"]}
-						tickFormatter={(value) => {
-							let val: string
-							if (tickFormatter) {
-								val = tickFormatter(value)
-							} else {
-								val = toFixedWithoutTrailingZeros(value, 2) + unit
-							}
-							return updateYAxisWidth(val)
-						}}
-						tickLine={false}
-						axisLine={false}
-					/>
-					{xAxis(chartData)}
-					<ChartTooltip
-						animationEasing="ease-out"
-						animationDuration={150}
-						content={
-							<ChartTooltipContent
-								labelFormatter={(_, data) => formatShortDate(data[0].payload.created)}
-								contentFormatter={({ value }) => {
-									if (contentFormatter) {
-										return contentFormatter(value)
-									}
-									return decimalString(value) + unit
-								}}
-								// indicator="line"
-							/>
-						}
-					/>
-					{dataKeys.map((key, i) => {
-						const color = `hsl(var(--chart-${key[2]}))`
-						return (
-							<Area
-								key={i}
-								dataKey={getNestedValue.bind(null, key[1], showMax)}
-								name={key[0]}
-								type="monotoneX"
-								fill={color}
-								fillOpacity={key[3]}
-								stroke={color}
-								isAnimationActive={false}
-							/>
-						)
 					})}
-					{/* <ChartLegend content={<ChartLegendContent />} /> */}
+				>
-				</AreaChart>
+					<AreaChart accessibilityLayer data={chartData.systemStats} margin={chartMargin}>
-			</ChartContainer>
+						<CartesianGrid vertical={false} />
-		</div>
+						<YAxis
-	)
+							direction="ltr"
-})
+							orientation={chartData.orientation}
+							className="tracking-tighter"
+							width={yAxisWidth}
+							domain={domain ?? [0, max ?? "auto"]}
+							tickFormatter={(value, index) => updateYAxisWidth(tickFormatter(value, index))}
+							tickLine={false}
+							axisLine={false}
+						/>
+						{xAxis(chartData)}
+						<ChartTooltip
+							animationEasing="ease-out"
+							animationDuration={150}
+							content={
+								<ChartTooltipContent
+									labelFormatter={(_, data) => formatShortDate(data[0].payload.created)}
+									contentFormatter={contentFormatter}
+								/>
+							}
+						/>
+						{dataPoints?.map((dataPoint, i) => {
+							const color = `var(--chart-${dataPoint.color})`
+							return (
+								<Area
+									key={i}
+									dataKey={dataPoint.dataKey}
+									name={dataPoint.label}
+									type="monotoneX"
+									fill={color}
+									fillOpacity={dataPoint.opacity}
+									stroke={color}
+									isAnimationActive={false}
+								/>
+							)
+						})}
+						{/* <ChartLegend content={<ChartLegendContent />} /> */}
+					</AreaChart>
+				</ChartContainer>
+			</div>
+		)
+	}, [chartData.systemStats.at(-1), yAxisWidth, maxToggled])
+}