Done
Details
Assignee
Pavel TankovPavel Tankov(Deactivated)Reporter
Oksana GrishchenkoOksana GrishchenkoNeeds QA
YesStory Points
1Sprint
NoneFix versions
Priority
Medium
Details
Details
Assignee
Pavel Tankov
Pavel Tankov(Deactivated)Reporter
Oksana Grishchenko
Oksana GrishchenkoNeeds QA
Yes
Story Points
1
Sprint
None
Fix versions
Priority
Smart Checklist
Smart Checklist
Smart Checklist
Created May 15, 2024 at 8:16 AM
Updated September 9, 2024 at 1:31 PM
Resolved August 30, 2024 at 8:38 AM
If there is a psmbdb cluster in an error state (in particular bc of a wrong configuration) and the cluster has finalizers it's going to stuck in the deletion.
The reason seems to be that the reconciliation never got to the checkFinalizers step if there is an error appeared previously in the
Reconcile()
method, that's why the finalizers are still there and the CR never gets deleted.An example of psmdb configuration for which the problem is reproducible:
(it has finalizers and an error in the configuration - the
max_allowed_packet=128M
can’t be parsed to a map).Steps to reproduce:
Apply the above config. Expected result: a psmdb cluster in error state.
Delete the cluster. Expected result: the cluster CR is deleted. Actual result: the cluster is not deleted.
It’s quite easy to get to this situation currently using Everest and it is confusing for the end user why they can’t delete a failed cluster.