stabilize worker

iaai_scraper/worker/celery_app.py

@@ -1,11 +1,15 @@
 # Инициализация Celery-приложения и периодических задач.
 
+import logging
+
 from celery import Celery
-from celery.signals import worker_process_init, setup_logging as celery_setup_logging
+from celery.signals import worker_process_init, worker_ready, setup_logging as celery_setup_logging
 
 from ..core.config import settings
 from ..core.logs import setup_logging
 
+logger = logging.getLogger("iaai_scraper.worker.celery_app")
+
 
 @celery_setup_logging.connect
 def _configure_logging(loglevel=None, **kwargs):
@@ -36,14 +40,27 @@ celery_app = Celery(
     backend=_result_backend(),
 )
 
+# Auto-clamp: если hard limit слишком далёк от soft (> soft + 120),
+# ограничиваем, чтобы зависший worker не жил вечно.
+_soft = settings.celery.task_soft_time_limit
+_hard = settings.celery.task_time_limit
+_max_hard = _soft + 120 if _soft else _hard
+if _hard > _max_hard:
+    logger.warning(
+        "CELERY_TASK_TIME_LIMIT=%d too far from CELERY_TASK_SOFT_TIME_LIMIT=%d; "
+        "clamping hard limit to %d",
+        _hard, _soft, _max_hard,
+    )
+    _hard = _max_hard
+
 celery_app.conf.update(
     task_serializer="json",
     accept_content=["json"],
     result_serializer="json",
     timezone="UTC",
     enable_utc=True,
-    task_soft_time_limit=settings.celery.task_soft_time_limit,
-    task_time_limit=settings.celery.task_time_limit,
+    task_soft_time_limit=_soft,
+    task_time_limit=_hard,
     task_acks_late=True,
     task_reject_on_worker_lost=True,
     task_track_started=True,
@@ -72,4 +89,16 @@ celery_app.conf.update(
     },
 )
 
-celery_app.autodiscover_tasks(["iaai_scraper.worker"])
+celery_app.autodiscover_tasks(["iaai_scraper.worker"])
+
+
+@worker_ready.connect
+def _on_worker_ready(**kwargs):
+    """Сразу при старте worker отправляем первую задачу sync_listing,
+    чтобы не ждать час до первого beat-цикла."""
+    logger.info("Worker ready — dispatching initial sync_listing task")
+    celery_app.send_task(
+        "iaai_scraper.worker.tasks.sync_listing_task",
+        kwargs={"limit": settings.celery.beat_sync_limit},
+        queue="scraping",
+    )