Symptoms

In the 4.5 version of PFMP The backup and restore operator is currently not waiting for the db pods to be synchronized at the end. If the sync is interrupted, it corrupts the replica. 

Cause

PostgreSQL replicas can lag behind the primary database instance. If this happens during a restore, it may cause the restore to fail.This is a rare intermittent failure and difficult to fix without major changes from the CMO team that maintains PostgreSQL. It is planned to be fixed in their upcoming release.

Resolution

Re-run the restore operation in MVM node. root@mno-installer# /tmp/k8/PFMP_Installer/scripts/restore_backup.sh ... Performing restore with below payload: ... Checking the execution status of the restore job: 2023-08-09 12:58:56,216 | INFO | Restore successful from the backup.

Support Cases