You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
When Postgres CR is deleted it triggers deletion routine, which might not end completely due to crash, misconfiguration or intermittent Kubernetes errors. To be able to finish cleanup on restart, Postgres CR should remain present up until deletion is completed.
Kubernetes solutions for that is to use finalizers where resources is still present and queryable until all finalizers are completed. This would allow operator to continue cleanup routine on restart
Reacted by Sergey Dudoladov, Stephane T., Christian Rohmann, Björn Fischer and Ricardo Trevizo
Activity
[-]use finalizers avoid loosing deletions[/-][+]use finalizers avoid losing deletions[/+]FxKu commentedon Apr 5, 2019
related to #387 and #551
frittentheke commentedon Jan 9, 2020
@redbaron thanks for opening this ... I was just about to suggest this myself ;-)
Another use case for finalizers would be a final pg_basebackup or logical backup before the cluster is annihilated by the operator for good - Just like AWS does with their final snapshot upon deletion of and RDS instance: https://docs.aws.amazon.com/AmazonRDS/latest/UserGuide/USER_DeleteInstance.html