fix(node): drop online clients and sub-nodes of nodes no longer synced

What the master derives from a node's reports (online clients, active
inbounds, learned sub-nodes) must live only while that node is still
synced; ClearNodeOnlineClients states it: a downed node must not keep its
clients listed as online.

Only a failed snapshot fetch cleared the online set, and only a failed
probe cleared sub-nodes. A disabled node (both jobs skip it), a node marked
offline before the sync tick reached it, a deleted node, and a node whose
snapshot fetched but failed to merge all kept their clients online in
onlineClients, onlineByGuid and activeInbounds, which the dashboard and a
parent master's /clients/onlines read. Disabled and deleted nodes also kept
their sub-nodes on the Nodes page until the panel restarted.

The traffic sync now keeps online sets only for enabled, online nodes in
its list, the heartbeat keeps sub-nodes only for enabled listed nodes, both
before the empty-list return, and a failed merge clears like a failed
fetch. The sync job's one-line call has no job-level test: that package
cannot install the xray process, so RetainSyncedNodeOnlineClients carries
the tested rule.

internal/web/job/node_heartbeat_descendants_test.go

@@ -0,0 +1,98 @@
+package job
+
+import (
+	"net/http"
+	"net/http/httptest"
+	"path/filepath"
+	"strconv"
+	"strings"
+	"testing"
+
+	"github.com/op/go-logging"
+
+	"github.com/mhsanaei/3x-ui/v3/internal/database"
+	"github.com/mhsanaei/3x-ui/v3/internal/database/model"
+	xuilogger "github.com/mhsanaei/3x-ui/v3/internal/logger"
+	"github.com/mhsanaei/3x-ui/v3/internal/web/runtime"
+	"github.com/mhsanaei/3x-ui/v3/internal/web/service"
+)
+
+func transitiveGuids(t *testing.T) []string {
+	t.Helper()
+	tree, err := (&service.NodeService{}).GetNodeTree()
+	if err != nil {
+		t.Fatalf("GetNodeTree: %v", err)
+	}
+	var out []string
+	for _, n := range tree {
+		if n.Transitive {
+			out = append(out, n.Guid)
+		}
+	}
+	return out
+}
+
+// The heartbeat skips a disabled node and never sees a deleted one, so the
+// sub-nodes it had learned from them stayed on the Nodes page for good.
+func TestHeartbeatDropsSubNodesOfNodesItNoLongerProbes(t *testing.T) {
+	cases := []struct {
+		name   string
+		retire func(t *testing.T, nodeID int)
+	}{
+		{"disabled", func(t *testing.T, nodeID int) {
+			if err := (&service.NodeService{}).SetEnable(nodeID, false); err != nil {
+				t.Fatalf("SetEnable: %v", err)
+			}
+		}},
+		{"deleted", func(t *testing.T, nodeID int) {
+			if err := (&service.NodeService{}).Delete(nodeID); err != nil {
+				t.Fatalf("Delete: %v", err)
+			}
+		}},
+	}
+	for _, tc := range cases {
+		t.Run(tc.name, func(t *testing.T) {
+			xuilogger.InitLogger(logging.ERROR)
+			if err := database.InitDB(filepath.Join(t.TempDir(), "x-ui.db")); err != nil {
+				t.Fatalf("InitDB: %v", err)
+			}
+			t.Cleanup(func() { _ = database.CloseDB() })
+			runtime.SetManager(runtime.NewManager(runtime.LocalDeps{APIPort: func() int { return 0 }, SetNeedRestart: func() {}}))
+			t.Cleanup(func() { runtime.SetManager(nil) })
+
+			srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+				w.Header().Set("Content-Type", "application/json")
+				switch {
+				case strings.HasSuffix(r.URL.Path, "server/status"):
+					_, _ = w.Write([]byte(`{"success":true,"obj":{"panelGuid":"direct-guid","xray":{"state":"running"}}}`))
+				case strings.HasSuffix(r.URL.Path, "server/descendants"):
+					_, _ = w.Write([]byte(`{"success":true,"obj":[{"guid":"sub-guid","parentGuid":"direct-guid","name":"sub","status":"online"}]}`))
+				default:
+					_, _ = w.Write([]byte(`{"success":true}`))
+				}
+			}))
+			t.Cleanup(srv.Close)
+			host, port, _ := strings.Cut(strings.TrimPrefix(srv.URL, "http://"), ":")
+			portNum, _ := strconv.Atoi(port)
+			node := &model.Node{
+				Name: "direct", Scheme: "http", Address: host, Port: portNum, BasePath: "/", ApiToken: "tok",
+				Enable: true, Status: "unknown", AllowPrivateAddress: true, TlsVerifyMode: "verify",
+			}
+			if err := database.GetDB().Create(node).Error; err != nil {
+				t.Fatalf("create node: %v", err)
+			}
+
+			hb := NewNodeHeartbeatJob()
+			hb.Run()
+			if got := transitiveGuids(t); len(got) != 1 || got[0] != "sub-guid" {
+				t.Fatalf("sub-nodes after first heartbeat = %v, want [sub-guid]", got)
+			}
+
+			tc.retire(t, node.Id)
+			hb.Run()
+			if got := transitiveGuids(t); len(got) != 0 {
+				t.Fatalf("sub-nodes after the node was %s = %v, want none", tc.name, got)
+			}
+		})
+	}
+}

internal/web/job/node_heartbeat_job.go

@@ -39,6 +39,7 @@ func (j *NodeHeartbeatJob) Run() {
 		logger.Warning("node heartbeat: load nodes failed:", err)
 		return
 	}
+	j.nodeService.RetainEnabledNodeDescendants(nodes)
 	if len(nodes) == 0 {
 		return
 	}

internal/web/job/node_traffic_sync_job.go

@@ -94,6 +94,7 @@ func (j *NodeTrafficSyncJob) Run() {
 		logger.Warning("node traffic sync: load nodes failed:", err)
 		return
 	}
+	j.inboundService.RetainSyncedNodeOnlineClients(nodes)
 	if len(nodes) == 0 {
 		return
 	}

internal/web/service/inbound_node.go

@@ -355,6 +355,10 @@ func (s *InboundService) SetRemoteTraffic(nodeID int, snap *runtime.TrafficSnaps
 		structuralChange, inner = s.setRemoteTrafficLocked(nodeID, snap, dirty, justPushed)
 		return inner
 	})
+	if err != nil {
+		// As on a failed fetch: a node whose snapshot did not merge keeps no online set.
+		s.ClearNodeOnlineClients(nodeID)
+	}
 	return structuralChange, err
 }
 
@@ -1454,6 +1458,20 @@ func (s *InboundService) ClearNodeOnlineClients(nodeID int) {
 	}
 }
 
+// RetainSyncedNodeOnlineClients keeps online clients only for nodes the traffic
+// sync still fetches; a node missing from nodes was deleted.
+func (s *InboundService) RetainSyncedNodeOnlineClients(nodes []*model.Node) {
+	process := currentXrayProcess()
+	if process == nil {
+		return
+	}
+	synced := make(map[int]bool, len(nodes))
+	for _, n := range nodes {
+		synced[n.Id] = n.Enable && n.Status == "online"
+	}
+	process.RetainNodeOnlineClients(func(nodeID int) bool { return synced[nodeID] })
+}
+
 // panelGuid returns this panel's stable self-identifier, used to key the local
 // panel's own clients in the per-node online maps (#4983).
 func (s *InboundService) panelGuid() string {

internal/web/service/node_tree.go

@@ -86,6 +86,22 @@ func (s *NodeService) ClearDescendants(nodeID int) {
 	nodeDescendantsMu.Unlock()
 }
 
+// RetainEnabledNodeDescendants drops sub-nodes learned from nodes the heartbeat no
+// longer probes: disabled ones it skips, deleted ones missing from nodes.
+func (s *NodeService) RetainEnabledNodeDescendants(nodes []*model.Node) {
+	enabled := make(map[int]bool, len(nodes))
+	for _, n := range nodes {
+		enabled[n.Id] = n.Enable
+	}
+	nodeDescendantsMu.Lock()
+	for nodeID := range nodeDescendantsCache {
+		if !enabled[nodeID] {
+			delete(nodeDescendantsCache, nodeID)
+		}
+	}
+	nodeDescendantsMu.Unlock()
+}
+
 func cachedDescendants() []model.NodeSummary {
 	nodeDescendantsMu.RLock()
 	defer nodeDescendantsMu.RUnlock()

internal/web/service/node_unsynced_online_test.go

@@ -0,0 +1,69 @@
+package service
+
+import (
+	"fmt"
+	"reflect"
+	"testing"
+
+	"github.com/mhsanaei/3x-ui/v3/internal/database"
+	"github.com/mhsanaei/3x-ui/v3/internal/database/model"
+	"github.com/mhsanaei/3x-ui/v3/internal/web/runtime"
+	"github.com/mhsanaei/3x-ui/v3/internal/xray"
+)
+
+func useOnlineTestProcess(t *testing.T) *xray.Process {
+	t.Helper()
+	previousProcess, previousResult := xrayState.snapshot()
+	process := xray.NewTestProcess(nil, "")
+	xrayState.replace(process)
+	t.Cleanup(func() {
+		xrayState.mu.Lock()
+		xrayState.process = previousProcess
+		xrayState.result = previousResult
+		xrayState.mu.Unlock()
+	})
+	return process
+}
+
+// Only a failed snapshot fetch used to clear a node's online set, so a node the
+// sync stopped reaching (disabled, marked offline, deleted) kept its clients online.
+func TestRetainSyncedNodeOnlineClientsDropsUnsyncedNodes(t *testing.T) {
+	setupConflictDB(t)
+	process := useOnlineTestProcess(t)
+	svc := InboundService{}
+	for id := 1; id <= 4; id++ {
+		guid := fmt.Sprintf("g%d", id)
+		svc.SetNodeOnlineTree(id, map[string][]string{guid: {guid + "@x"}})
+		process.SetNodeActiveInboundTree(id, map[string][]string{guid: {"in-" + guid}})
+	}
+
+	svc.RetainSyncedNodeOnlineClients([]*model.Node{
+		{Id: 1, Enable: true, Status: "online"},
+		{Id: 2, Enable: false, Status: "online"},
+		{Id: 3, Enable: true, Status: "offline"},
+	})
+
+	if got, want := svc.GetOnlineClientsByGuid(), map[string][]string{"g1": {"g1@x"}}; !reflect.DeepEqual(got, want) {
+		t.Errorf("online by guid = %v, want %v", got, want)
+	}
+	if got, want := svc.GetActiveInboundsByGuid(), map[string][]string{"g1": {"in-g1"}}; !reflect.DeepEqual(got, want) {
+		t.Errorf("active inbounds by guid = %v, want %v", got, want)
+	}
+}
+
+func TestSetRemoteTrafficFailureClearsNodeOnlineClients(t *testing.T) {
+	setupConflictDB(t)
+	useOnlineTestProcess(t)
+	svc := InboundService{}
+	svc.SetNodeOnlineTree(7, map[string][]string{"g7": {"a@x"}})
+	if err := database.GetDB().Exec("DROP TABLE inbounds").Error; err != nil {
+		t.Fatalf("drop inbounds: %v", err)
+	}
+
+	if _, err := svc.SetRemoteTraffic(7, &runtime.TrafficSnapshot{}, false, false); err == nil {
+		t.Fatal("SetRemoteTraffic succeeded without an inbounds table")
+	}
+	if got := svc.GetOnlineClientsByGuid(); len(got) != 0 {
+		t.Errorf("online by guid after a failed merge = %v, want none", got)
+	}
+}