[bitnami/postgresql-ha] Pods crashing by "Name or service not known" about charts HOT 3 CLOSED

Hi @BruceDu521, thanks for using bitnami/charts.

With the information provided is not easy to reproduce the failure. It seems you are facing any kind of deadlock. Do you know how you reached that situation? Have you upgraded the images or your configuration? Could you share the values/configuration you are using?

from charts.

Hi @fmulero , thanks for help.

I found the reason last night.
The pg-ha pods we deployed by a program, and the program would assign username and password to postgres cluster. For a stupid bug, program got wrong username and password. So we cleared pvc, fixed program's bug, and restart the whole progress, pg-ha pods was running as normal.

What confused me is, why it was because of wrong username and password, but pg printed log as this:

[2024-02-19 07:57:17] [DETAIL]
could not translate host name "lake-sql-postgres-gis-7bc4d3b9c8-postgresql-2.lake-sql-postgres-gis-7bc4d3b9c8-postgresql-headless.default.svc.cluster.local" to address: Name or service not known

from charts.

Great @BruceDu521! it shouldn't have been easy. Thanks a lot for sharing your findings.

The DNS error appears because the pod never gets ready and it isn't attached to the headless service.

I agree about the error, it should be clearer, with that information is not easy to find the root cause. If it happens again, please enable the debug mode in the image to get more logs.

from charts.