fix: prevent WebSocket reconnect loops on slow agent collections (#2294)

The hub waited only 5s (the request manager default) for stats over
WebSocket and closed the connection on any error. On hosts where
`zpool list` stalls (seen on Proxmox, up to its 10s timeout), collection
exceeded that limit, so the hub sent a close (code 1000) and the agent
reconnected. The refresh ran every other cycle, which caused a
disconnect roughly every 2 minutes.

Hub:
- Wait up to 30s for WebSocket stats responses.
- Keep the connection open when a stats request times out; only close
  it (and fall back to SSH) for other errors.

Agent:
- After the first collection, refresh `zpool list` pool stats and
  `zfs list` dataset usage in the background and serve cached values
  meanwhile, so a hung utility cannot delay the stats response.

internal/hub/systems/ws_data_timeout_test.go

@@ -0,0 +1,82 @@
+//go:build testing
+
+package systems
+
+import (
+	"context"
+	"net/http"
+	"net/http/httptest"
+	"strings"
+	"testing"
+	"time"
+
+	"github.com/blang/semver"
+	"github.com/fxamacker/cbor/v2"
+	"github.com/henrygd/beszel/internal/common"
+	esystem "github.com/henrygd/beszel/internal/entities/system"
+	"github.com/henrygd/beszel/internal/hub/ws"
+	"github.com/lxzan/gws"
+	"github.com/stretchr/testify/require"
+)
+
+// slowDataClient answers GetData only after release is closed, simulating an
+// agent whose collection outlasts the hub's request timeout.
+type slowDataClient struct {
+	gws.BuiltinEventHandler
+	release chan struct{}
+}
+
+func (c *slowDataClient) OnMessage(conn *gws.Conn, message *gws.Message) {
+	defer message.Close()
+	var req common.HubRequest[cbor.RawMessage]
+	if err := cbor.Unmarshal(message.Bytes(), &req); err != nil || req.Action != common.GetData {
+		return
+	}
+	<-c.release
+	response, _ := cbor.Marshal(common.AgentResponse{Id: req.Id, SystemData: &esystem.CombinedData{}})
+	_ = conn.WriteMessage(gws.OpcodeBinary, response)
+}
+
+func TestFetchDataTimeoutKeepsWebSocketOpen(t *testing.T) {
+	originalTimeout := wsDataRequestTimeout
+	wsDataRequestTimeout = 50 * time.Millisecond
+	t.Cleanup(func() { wsDataRequestTimeout = originalTimeout })
+
+	connections := make(chan *ws.WsConn, 1)
+	upgrader := gws.NewUpgrader(&monitorSyncServer{}, nil)
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		conn, err := upgrader.Upgrade(w, r)
+		if err != nil {
+			t.Error(err)
+			return
+		}
+		wsConn := ws.NewWsConnection(conn, semver.MustParse("0.20.0"))
+		conn.Session().Store("wsConn", wsConn)
+		connections <- wsConn
+		conn.ReadLoop()
+	}))
+	t.Cleanup(server.Close)
+
+	client := &slowDataClient{release: make(chan struct{})}
+	conn, _, err := gws.NewClient(client, &gws.ClientOption{Addr: "ws" + strings.TrimPrefix(server.URL, "http")})
+	require.NoError(t, err)
+	t.Cleanup(func() { _ = conn.NetConn().Close() })
+	go conn.ReadLoop()
+
+	var sys *System
+	select {
+	case wsConn := <-connections:
+		sys = &System{WsConn: wsConn}
+	case <-time.After(3 * time.Second):
+		t.Fatal("websocket connection was not established")
+	}
+
+	_, err = sys.fetchDataFromAgent(common.DataRequestOptions{})
+	require.ErrorIs(t, err, context.DeadlineExceeded)
+	require.True(t, sys.WsConn.IsConnected(), "a slow collection must not close the connection")
+
+	// The late response is discarded and the next request still succeeds.
+	close(client.release)
+	_, err = sys.fetchDataFromAgent(common.DataRequestOptions{})
+	require.NoError(t, err)
+}