From d5db50540a89999c07e57688f5c47a8a3f9ba20c Mon Sep 17 00:00:00 2001
From: wxg0103 <727495428@qq.com>
Date: Tue, 13 Oct 2026 16:55:28 +0800
Subject: [PATCH] chore: implement SafeDjangoJobStore to handle JobLookupError
 gracefully in job updates and removals

---
 apps/common/job/scheduler.py | 33 ++++++++++++++++++++++++++++++---
 1 file changed, 30 insertions(+), 3 deletions(-)

diff --git a/apps/common/job/scheduler.py b/apps/common/job/scheduler.py
index 33ccd27829a..b73f12f0d08 100644
--- a/apps/common/job/scheduler.py
+++ b/apps/common/job/scheduler.py
@@ -1,12 +1,39 @@
+from apscheduler.jobstores.base import JobLookupError
 from apscheduler.schedulers.background import BackgroundScheduler
 from django_apscheduler.jobstores import DjangoJobStore
 
+from common.utils.logger import maxkb_logger
+
+
+class SafeDjangoJobStore(DjangoJobStore):
+    """
+    DjangoJobStore 在目标 job 已被删除时会抛出 JobLookupError。
+
+    多个 celery worker(进程/线程)共享同一张 django_apscheduler_djangojob 表,
+    任一进程删除 job 后, 其它进程在调度该 job 的过程中就会命中这两个异常:
+    update_job 找不到记录, remove_job 找不到记录。
+    该异常会从 APScheduler 的调度主循环(apscheduler.schedulers.base._process_jobs)
+    直接冒泡出去并终止 'APScheduler' 线程, 导致该进程内所有定时任务永久停摆,
+    因此这里把 "job 已不存在" 视为正常情况处理。
+    """
+
+    def update_job(self, job):
+        try:
+            super().update_job(job)
+        except JobLookupError:
+            maxkb_logger.warning(f"Skip update of job [{job.id}], it has been removed already")
+
+    def remove_job(self, job_id):
+        try:
+            super().remove_job(job_id)
+        except JobLookupError:
+            pass
+
+
 scheduler = BackgroundScheduler()
-scheduler.add_jobstore(DjangoJobStore(), "default")
+scheduler.add_jobstore(SafeDjangoJobStore(), "default")
 
 try:
     scheduler.start()
 except Exception as e:
-    from common.utils.logger import maxkb_logger
-
     maxkb_logger.error(f"Failed to start scheduler: {e}")