From 1ef1fdeac94c0c6546317be67c721adf088bfb7f Mon Sep 17 00:00:00 2001
From: weichen <weichen@home.local>
Date: Sun, 11 Oct 2026 08:12:03 +0000
Subject: [PATCH] Make notice redelivery iterative instead of self-recursive

Why: deliverNotices re-invoked itself synchronously from a defer after a failed startNoticeRun requeued the batch - a persistently failing delivery (busy conversation, missing row, slow DB) recursed without bound and overflowed the stack (real-PG CI job 2352, internal/runmgr).

Evidence: regression test TestNoticeDeliveryFailureIsBounded (missing conversation row) hangs on the old code and passes with the fix; full backend suite green on SQLite and against real PostgreSQL 18.6 + Redis; clean-HEAD reproduction isolated an unrelated count=2 admin-login artifact.

Impact: failed batches now return after requeue (retried on the next enqueue or restart); notices arriving during delivery are picked up by a bounded loop tail instead of defer recursion; no behavior change on the success path.

Refs: Dev PR #333; plan R8 real-database gate.
---
 internal/runmgr/notice_recursion_test.go | 46 ++++++++++++++++++++++
 internal/runmgr/runmgr.go                | 50 ++++++++++++++----------
 2 files changed, 76 insertions(+), 20 deletions(-)
 create mode 100644 internal/runmgr/notice_recursion_test.go

diff --git a/internal/runmgr/notice_recursion_test.go b/internal/runmgr/notice_recursion_test.go
new file mode 100644
index 0000000..f251e70
--- /dev/null
+++ b/internal/runmgr/notice_recursion_test.go
@@ -0,0 +1,46 @@
+package runmgr
+
+import (
+	"testing"
+	"time"
+
+	"agent-system-backend/internal/store"
+	core "agent-system-core"
+)
+
+// U3 真库门禁回归（job 2352 栈溢出）：投递失败（会话行不存在）时旧实现的
+// defer 同步再入 deliverNotices 无限自递归。修复后：单次有界返回、通知
+// 保留在队列，由下一次触发或重启重投驱动。
+func TestNoticeDeliveryFailureIsBounded(t *testing.T) {
+	runner := &gateRunner{parts: []core.Part{{Type: core.PartText, Text: "结果"}}}
+	c, _ := newSpawnEnv(t, runner)
+	c.mgr.SetWorkspaceRoot(t.TempDir())
+
+	convID := core.ID(store.NewID("cnv"))
+	c.mgr.enqueueNotice(convID, bgNotice{owner: c.run.Owner, name: "检索", payload: core.AgentResultPayload{Status: "completed"}})
+
+	done := make(chan struct{})
+	go func() {
+		c.mgr.deliverNotices(convID)
+		close(done)
+	}()
+	select {
+	case <-done:
+	case <-time.After(5 * time.Second):
+		t.Fatal("deliverNotices 未返回：失败重投仍在递归/自旋")
+	}
+	c.mgr.noticeMu.Lock()
+	queued := len(c.mgr.notices[convID])
+	c.mgr.noticeMu.Unlock()
+	if queued != 1 {
+		t.Fatalf("失败应保留待投通知：queued=%d", queued)
+	}
+	// 第二次触发同样有界（不因重试堆积）。
+	c.mgr.deliverNotices(convID)
+	c.mgr.noticeMu.Lock()
+	queued = len(c.mgr.notices[convID])
+	c.mgr.noticeMu.Unlock()
+	if queued != 1 {
+		t.Fatalf("重复失败不应堆积通知：queued=%d", queued)
+	}
+}
diff --git a/internal/runmgr/runmgr.go b/internal/runmgr/runmgr.go
index dfcaef2..a71eb1f 100644
--- a/internal/runmgr/runmgr.go
+++ b/internal/runmgr/runmgr.go
@@ -1821,34 +1821,44 @@ func (m *Manager) deliverNotices(convID string) {
 	defer func() {
 		m.noticeMu.Lock()
 		delete(m.delivering, convID)
-		more := len(m.notices[convID]) > 0
 		m.noticeMu.Unlock()
-		if more {
-			m.deliverNotices(convID)
-		}
 	}()
 
-	if err := m.startNoticeRun(convID, items); err != nil {
+	// 迭代而非递归（U3 真库门禁实测）：原实现的 defer 在投递失败重新入队后
+	// 同步再入 deliverNotices——持续失败（如会话忙/查库错误）时无限自递归，
+	// test-pg 慢速时序下栈溢出（job 2352）。失败本轮即返回：队列保留，
+	// 由下一次 enqueueNotice 或重启重投驱动；投递期间新到的通知由循环
+	// 尾部再取，不再经 defer 递归。
+	for {
+		if err := m.startNoticeRun(convID, items); err != nil {
+			m.noticeMu.Lock()
+			if m.notices == nil {
+				m.notices = map[string][]bgNotice{}
+			}
+			m.notices[convID] = append(items, m.notices[convID]...)
+			m.noticeMu.Unlock()
+			return
+		}
+		// 投递成功才删持久化行（失败保留下次重投）。
+		ids := make([]string, 0, len(items))
+		for _, n := range items {
+			if n.id != "" {
+				ids = append(ids, n.id)
+			}
+		}
+		if err := m.st.DeleteNotices(ids); err != nil {
+			log.Printf("[runmgr] notice delete: %v", err)
+		}
 		m.noticeMu.Lock()
-		if m.notices == nil {
-			m.notices = map[string][]bgNotice{}
+		items = append([]bgNotice{}, m.notices[convID]...)
+		if len(items) == 0 {
+			m.noticeMu.Unlock()
+			return
 		}
-		m.notices[convID] = append(items, m.notices[convID]...)
+		delete(m.notices, convID)
 		m.noticeMu.Unlock()
-		return
-	}
-	// 投递成功才删持久化行（失败保留下次重投）。
-	ids := make([]string, 0, len(items))
-	for _, n := range items {
-		if n.id != "" {
-			ids = append(ids, n.id)
-		}
-	}
-	if err := m.st.DeleteNotices(ids); err != nil {
-		log.Printf("[runmgr] notice delete: %v", err)
 	}
 }
-
 func (m *Manager) startNoticeRun(convID string, items []bgNotice) error {
 	if len(items) == 0 {
 		return nil