I have manually triggered update of worker node pool in GKE but after update one of our database persisent volumes was recreated instead of just reattaching to pod on new node.
We have backup for GKE enabled so I tried to restore affected namespace but unfortunately volume is not created. As this is postgresql statefulset it just creates new PVC with no data. There are 2 volumes in this namespace one for application and one for database. The application volume is restored without problem but database volume does not seems to be restored at all. I have checked backups and they are successful. Also restore process is showed as successful but shows only application volume not database.
completeTime: '2023-04-19T05:01:36.014234036Z'
createTime: '2023-04-19T05:00:13.899181954Z'
diskSizeBytes: '17179869184'
etag: '6'
format: GCE_PERSISTENT_DISK
name: ********
sourcePvc:
name: data-postgresql-0
namespace: namespace
state: SUCCEEDED
storageBytes: '27198848'
updateTime: '2023-04-19T05:01:36.026476611Z'
Whats also interesting is that application volume is created from backup with storage class refering to gkebackup but database is not.
data-postgresql-0 Bound pvc-9dcfd2c8-58aa-410b-8b03-46bd232baed6 16Gi RWO standard-rwo 27s
data-application Bound pvc-65883a2b94bf4b44 10Gi RWO gce-pd-gkebackup-ie 29s
Has anyone faced similar issue ? First of all I do not understand how this volume could be recreated during the update instead of just reattached to new node/pod.
Thank you very much