messenger: clear the state problem once a write succeeds again, so the status stops saying a failure that passed (issue 1507)

modules/messenger/cmd/messenger/holder.go

@@ -1130,6 +1130,8 @@ func (h *Holder) record(s Sent) {
 	if h.Store != nil {
 		if err := h.Store.PutRecent(recent); err != nil {
 			h.noteStore(err)
+		} else {
+			h.storeWritten()
 		}
 	}
 }
@@ -1143,6 +1145,20 @@ func (h *Holder) persist(r *Record) {
 	h.mu.Unlock()
 	if err := h.Store.Put(c); err != nil {
 		h.noteStore(err)
+	} else {
+		h.storeWritten()
+	}
+}
+
+// storeWritten clears a state problem once a write succeeds again: the status says what holds now, never a
+// failure that passed (issue 1507's reading of 2026-10-28: a verdict kept a single failure of the day before).
+func (h *Holder) storeWritten() {
+	h.mu.Lock()
+	was := h.storeErr
+	h.storeErr = ""
+	h.mu.Unlock()
+	if was != "" {
+		h.Logf("[messenger] writes its state again (it could not: %s)", was)
 	}
 }
 

modules/messenger/cmd/messenger/holder_test.go

@@ -71,9 +71,14 @@ type memStore struct {
 	passes map[string]Pass
 	// puts counts the writes, so a write of bytes identical to what is stored is still seen (issue 422's review).
 	puts int
+	// fail, when set, refuses every write of a record or of the sends, as the bus does with no responders.
+	fail error
 }
 
 func (m *memStore) Put(r Record) error {
+	if m.fail != nil {
+		return m.fail
+	}
 	if m.recs == nil {
 		m.recs = map[string]Record{}
 	}
@@ -89,7 +94,13 @@ func (m *memStore) All() ([]Record, error) {
 	}
 	return out, nil
 }
-func (m *memStore) PutRecent(s []Sent) error { m.recent = s; return nil }
+func (m *memStore) PutRecent(s []Sent) error {
+	if m.fail != nil {
+		return m.fail
+	}
+	m.recent = s
+	return nil
+}
 func (m *memStore) PutWalk(w Walk) error {
 	if m.walks == nil {
 		m.walks = map[string]Walk{}
@@ -431,15 +442,21 @@ func TestEscalationIsSaidOnce(t *testing.T) {
 }
 
 func TestAChannelThatCannotSendSaysSoAndIsTriedAgain(t *testing.T) {
-	h, tg, _, c, _ := newHolder(t)
+	h, tg, _, c, store := newHolder(t)
 	h.Desk = nil
 	tg.fail = errors.New("telegram sendMessage: cannot connect (dial)")
+	// The state could not be written either, once: said while it lasts, and no longer once a write succeeds
+	// (issue 1507's reading of 2026-10-28: a verdict kept a day-old "no responders").
+	store.fail = errors.New("nats: no responders available for request")
 	h.Condition(EventRaised, cond("machine.ace.silent", Urgent, "silent"))
 	st := h.Status("listening")
 	if !strings.HasPrefix(st.Verdict, "CANNOT SEND") || len(st.Unsent) != 1 || st.Channels[0].LastError == "" {
 		t.Fatalf("status: %+v", st)
 	}
-	tg.fail = nil
+	if st.StateProblem == "" {
+		t.Fatalf("a failed state write is not said: %+v", st)
+	}
+	tg.fail, store.fail = nil, nil
 	h.Flush()
 	if len(tg.sends) != 0 {
 		t.Fatalf("tried again before a minute")