Retry on `TimeLimitExceeded`

[adi-]

Sometimes worker is stuck on executing email tasks (probably connection problems). To fix this, there is an Celery option TASK_TIME_LIMIT where you can limit task execution time (p.s. default value is infinite). However, when task hits time limit, worker is killed with SIGKILL and what I see from logs, the task is not retried for the second time.

[2020-05-12 11:04:51,352: INFO/MainProcess] Received task: djcelery_email_send_multiple[44dfa3fd-f801-4d8d-b15f-eec05c1e04b4]                                 │
│[2020-05-12 11:05:51,371: ERROR/MainProcess] Task handler raised error: TimeLimitExceeded(60,)                                                                │
│Traceback (most recent call last):                                                                                                                            │
│  File "/opt/venvs/www/lib/python3.5/site-packages/billiard/pool.py", line 658, in on_hard_timeout                                                      │
│    raise TimeLimitExceeded(job._timeout)                                                                                                                     │
│billiard.exceptions.TimeLimitExceeded: TimeLimitExceeded(60,)                                                                                                 │
│[2020-05-12 11:05:51,373: ERROR/MainProcess] Hard time limit (60s) exceeded for djcelery_email_send_multiple[44dfa3fd-f801-4d8d-b15f-eec05c1e04b4]            │
│[2020-05-12 11:05:51,522: ERROR/MainProcess] Process 'ForkPoolWorker-1' pid:17908 exited with 'signal 9 (SIGKILL)'                                            │
│[2020-05-12 11:09:45,841: INFO/MainProcess] Received task: djcelery_email_send_multiple[399a06d3-629f-42df-a926-9afb76afc98e]                                 │
│[2020-05-12 11:09:47,142: INFO/ForkPoolWorker-2] Task djcelery_email_send_multiple[399a06d3-629f-42df-a926-9afb76afc98e] succeeded in 1.2928511229692958s: 1

I am considering to use https://docs.celeryproject.org/en/latest/userguide/configuration.html#task-reject-on-worker-lost, but as docs states, it is probably not the best way to handle this error.

I guess the easiest way would be to handle such error in django-celery-email internally. Or maybe I am missing something?

[adi-]

Meanwhile, I'll try to use TASK_SOFT_TIME_LIMIT.

[adi-]

Meanwhile, I'll try to use TASK_SOFT_TIME_LIMIT.

Ok, nice. Above is working like a charm. Btw, my error with celery was Cannot reach CELERY_EMAIL_BACKEND django. I guess it would be wise to mention in django-celery-email's docs, that it is recommended to set TASK_SOFT_TIME_LIMIT.