Open slothyrulez opened 1 month ago
That's an interesting one. It's possible that the VACUUM VERBOSE ANALYZE crashes itself (e.g. due to another extension), though I wonder if the amount of notices produced by VERBOSE is causing an issue with the background worker mode used by RDS. Can you try to run without VERBOSE to see whether the issue goes away? (which is a strong indication that pg_cron is indeed the problem)
Ey!, thanks for your time.
Yes, I think is possible that the cause is not pg_cron itself, but VACUUM. Ok, I'll reconfigure the job to run without the VERBOSE, and let you know.
Thanks again.
@marcoslot update here: we have been running VACUUM ANALYZE 5 days without any issue for the moment.
Checked again and running as expected, I don't know if i can help you somehow.
Hi, Me too getting this issue in aurora postgreSQL RDS database. while executing a materialized view via pg_cron.
2024-08-21 06:47:00 UTC::@:[554]:LOG: background worker "pg_cron" (PID 22225) was terminated by signal 11: Segmentation fault
2024-08-21 06:47:00 UTC::@:[554]:DETAIL: Failed process was running: SELECT ingress.refresh_interactions_materialized_view();
2024-08-21 06:47:00 UTC::@:[554]:LOG: terminating any other active server processes
2024-08-21 06:47:00 UTC::@:[554]:FATAL: Aurora Runtime process unexpectedly exited
2024-08-21 06:47:00 UTC::@:[554]:LOG: database system is shut down
When I manually executes this, it will take around 4 minutes to finish.
Thanks.
@bdrouvotAWS do you happen to have any way of getting a stack trace?
@rinshadka @slothyrulez could you please ping me at bdrouvot@amazon.com?
Hi!
We are running postgresql 14.10 on AWS RDS, we have installed recently the pg_cron 1.6 extension, following this guide: https://docs.aws.amazon.com/AmazonRDS/latest/UserGuide/PostgreSQL_pg_cron.html
We only have one job configured, to launch every day out of peak hours, a VACUUM VERBOSE ANALYZE on the whole database.
I have found we have 2 crashes related to pg_cron:
and
Do you have any hints ?