CUB-200: fix event-loss race — register handlers before readLoop starts
Some checks failed
Dev Build & Deploy / test-and-build (pull_request) Failing after 0s
Dev Build & Deploy / docker-build-push (pull_request) Has been skipped

Move registerEventHandlers() call before the readLoop goroutine starts
in connectAndRun(). This eliminates the startup window where live gateway
events were actively read and dropped as 'unhandled' because handler
registration happened only after initialSync completed.

The handlers only depend on c.agents and c.broker, which are wired in the
constructor — they do not require initialSync to have completed.

Also adds TestConnectAndRun_EventNotLostDuringSync regression test that
sends a live presence event during initial sync and asserts it is not lost.

All gateway tests pass with -race.
This commit is contained in:
2026-05-20 21:52:39 +00:00
parent d370d5ec23
commit b7b05bb4e3
2 changed files with 145 additions and 8 deletions

View File

@@ -565,6 +565,138 @@ func TestConnectAndRun_InitialSyncOrdering(t *testing.T) {
}
}
// ── 7. Test: Event not lost during initial sync (regression) ───────────────
// TestConnectAndRun_EventNotLostDuringSync verifies that live gateway events
// arriving during initial sync are NOT dropped. This is a regression test
// for the race where readLoop started before registerEventHandlers(),
// causing events read during that window to be logged as "unhandled" and lost.
//
// The mock server sends a live event (sessions.changed) right after the
// handshake, interleaved with the RPC responses for agents.list and
// sessions.list. The test asserts the event is received by the handler.
func TestConnectAndRun_EventNotLostDuringSync(t *testing.T) {
repo := &mockAgentRepo{agents: make(map[string]models.AgentCardData)}
broker := handler.NewBroker()
capture := newBroadcastCapture(broker)
defer capture.close()
// Pre-seed an agent so the event handler can update it.
repo.agents["otto"] = models.AgentCardData{
ID: "otto",
DisplayName: "Otto",
Status: models.AgentStatusIdle,
}
srv := newTestWSServer(t, func(conn *websocket.Conn) {
// Handshake
handleHandshake(t, conn)
// After handshake, process RPCs and inject a live event.
for {
var req map[string]any
if err := conn.ReadJSON(&req); err != nil {
break
}
reqID, _ := req["id"].(string)
method, _ := req["method"].(string)
// Respond to agents.list RPC
if method == "agents.list" {
// Before responding, inject a live event — simulates
// a gateway pushing a presence update during sync.
evt := map[string]any{
"type": "event",
"event": "presence",
"params": map[string]any{"agentId": "otto", "connected": true, "lastActivityAt": "2025-05-20T12:30:00Z"},
}
if err := conn.WriteJSON(evt); err != nil {
break
}
// Now send the RPC response
res := map[string]any{
"type": "res",
"id": reqID,
"ok": true,
"result": []map[string]any{
{"id": "otto", "name": "Otto", "role": "Orchestrator", "channel": "discord"},
},
}
if err := conn.WriteJSON(res); err != nil {
break
}
continue
}
// Respond to sessions.list RPC
if method == "sessions.list" {
res := map[string]any{
"type": "res",
"id": reqID,
"ok": true,
"result": []map[string]any{},
}
if err := conn.WriteJSON(res); err != nil {
break
}
continue
}
// Default response for other methods
res := map[string]any{
"type": "res",
"id": reqID,
"ok": true,
"result": map[string]any{},
}
if err := conn.WriteJSON(res); err != nil {
break
}
}
})
defer srv.Close()
client := NewWSClient(WSConfig{URL: wsURL(srv), AuthToken: "test-token"}, repo, broker, slog.Default())
ctx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
defer cancel()
done := make(chan struct{})
go func() {
client.Start(ctx)
close(done)
}()
// Wait for the presence event to be processed by checking the repo.
// The presence handler updates the agent, so we check for the
// lastActivityAt change.
deadline := time.Now().Add(5 * time.Second)
var lastActivity string
for time.Now().Before(deadline) {
repo.mu.Lock()
if a, ok := repo.agents["otto"]; ok {
lastActivity = a.LastActivity
}
repo.mu.Unlock()
if lastActivity == "2025-05-20T12:30:00Z" {
break
}
time.Sleep(50 * time.Millisecond)
}
if lastActivity != "2025-05-20T12:30:00Z" {
t.Errorf("presence event during sync was lost: lastActivity = %q, want %q", lastActivity, "2025-05-20T12:30:00Z")
}
cancel()
select {
case <-done:
case <-time.After(3 * time.Second):
t.Fatal("WSClient did not shut down cleanly")
}
}
func TestStrPtr(t *testing.T) {
s := "hello"
p := strPtr(s)