PITR unreliable, frozen

My PITR pod seems to stop working, its doing nothing for last 5 hours.

# kubectl logs -n pxctest pxctest-pitr-6ff5845d8b-pbrjp
2024/03/15 11:26:47 Starting to process binlog with name binlog.009181
2024/03/15 11:26:49 Successfully written binlog file binlog.009181 to s3 with name binlog_1710501946_de549d41a0fda813ffc31018a77b30ff
2024/03/15 11:27:06 Reading binlogs from pxc with hostname= pxctest-pxc-0.pxctest-pxc.pxctest.svc.cluster.local
2024/03/15 11:27:27 Starting to process binlog with name binlog.009182

So for me PITR cannot be trusted for backup, recovery or anything.

No problem with primary view or any other problem was recorded in pmm durring this time

Deleting pod lead to PITR working again with apparent consistency/continuity in saved GTID sets/binlogs. In other words, seems like no logs has been lost.

In latest version 1.14 of the percona operator there is a new field under pitr section named “timeoutSeconds”. I guess it is designed to help with this issue.

Hi @jw12314 thank you for posting to the Percona forums! Can you confirm whether release 1.14 has rectified your issue?