Description:
I have 2 clusters managed by the same operator: pxc-db and pxc-db-backup-restore. I cannot restore a backup from the pxc-db cluster to the pxc-db-backup-restore cluster using pitr.
Steps to Reproduce:
to restore a PITR backup to a new Percona XtraDB cluster I create the PerconaXtraDBClusterRestore resource as follows:
kind: PerconaXtraDBClusterRestore
apiVersion: pxc.percona.com/v1
metadata:
name: restore-test-2024-11-05
namespace: pxc-db-backup-restore
spec:
pxcCluster: pxc-db-backup-restore
backupSource:
destination: azure://pxc-db/pxc-db-2024-11-05-11:05:20-full
azure:
container: pxc-db
credentialsSecret: azure-secret
pitr:
type: latest
backupSource:
destination: azure://pxc-db/pitr
azure:
container: pxc-db/pitr
credentialsSecret: azure-secret
My Azure storage account has all backups directly in a container named pxc-db and all binlogs are stored in a folder named pitr within the same container. The restoration procedure works fine until the pitr-job pod fails within the first few seconds with the following output:
2024/11/05 14:17:49 ERROR: new recoverer controller: get start GTID: no info files in sst dir
It retries a few times, but all retried pods fail with the same message.
Please note that restoring a backup without any pitr works fine.
Version:
Operator: 1.15.1
Percona XtraDB Cluster: 8.0.36-28.1
Logs:
2024/11/05 14:17:49 ERROR: new recoverer controller: get start GTID: no info files in sst dir
Expected Result:
Success
Actual Result:
Pod Failed
Additional Information:
n/a