Delayed job stops, the jobs are stuck in postgres

[Chandananimmu]

@albus522 I'm using delayed job 4.1.9 , delayed_job_active_record 4.1.6 , when I try to send 10,000 emails via delayed job, it get stucks and the delayed job stops working, job will be in postgres, could you please help me to resolve this.

[kaylareopelle]

I think I may be having a similar issue! I'm interested to learn about any solutions folks have found.

[davidkrider]

I get these in my log:

I, [2022-02-10T21:29:22.975757 #3950] INFO -- : 2022-02-10T21:29:22-0500: [Worker(delayed_job host:miner pid:3950)] Error while reserving job: PG::UnableToSend: SSL SYSCALL error: EOF detected I, [2022-02-10T21:29:27.977649 #3950] INFO -- : 2022-02-10T21:29:27-0500: [Worker(delayed_job host:miner pid:3950)] Error while reserving job: PG::UnableToSend: no connection to the server ... x9 F, [2022-02-10T21:30:07.993164 #3950] FATAL -- : PG::UnableToSend: no connection to the server

I get 9 copies of that second message, then it gives up, and the delayed_job daemon falls over.

[brijeshs-atharvasystem]

I am also facing the same issue for specific a queue only.

• delayed_job (4.1.11)
• delayed_job_active_record (4.1.7)
• Postgres 15

In my case, it is not happening with every job. I have a few queues and this issue is happening in only one queue. Even in a single queue, jobs are performed and deleted most of the time but sometimes, jobs are not deleted. Due to this, pending jobs are stuck and not processed and I received this error.

Error: execution expired (Delayed::Worker.max_run_time is only 14400 seconds) (Delayed::WorkerTimeout)

Note: I have been facing this issue since I upgraded the Postgres version from 11 to 15.

Please let me know if anyone found the solution for this issue.
Thanks

[davidkrider]

My app and database run in Azure. In my case, I finally figured out that Microsoft was updating my Postgres instance and restarting it without any notification. (This seems crazy to me.) This was killing my long-running jobs, and leaving me with this error message. I thought the messages were confusing, because they were telling me that the database had dropped, but it seemed perfectly fine. However, they were actually correct. I finally found a setting on the Postgres service to limit updates and restarts to only critical fixes, and this has stopped happening for me.

[brijeshs-atharvasystem]

@davidkrider Thanks for the reply. In my case, I upgraded the Postgres version to 15. I am not sure that is the cause of this issue. If there is any issue with the Postgres version then other queues are also affected which is not the case here.

I observed that the job is performed but the job is not unlocked (locked_at is not updated) so it is not deleted.

[brijeshs-atharvasystem]

It seems like there was some issue with Ruby 2.7.3. I downgraded the version to 2.6.5 and all queues are working fine.