giantswarm/aws-operator

VPCs not being deleted on cluster deletion

rossf7 opened this issue · 3 comments

On gauss we have a lot of VPCs that haven't been deleted and have a route table remaining. We may need more retry logic when deleting network resources as the VPC won't be deleted if these still exist.

Retry logic for VPC deletion was added in /pull/449 will check if this helps.

This is part of the postmortem in #1953. I believe the bug was fixed by the extra retry logic. But we need to clean up the orphan VPCs and have an automated check to verify its gone.

See https://github.com/giantswarm/giantswarm/issues/1953#issuecomment-339248254

Closing. VPC deletion has been reliable since the guest cluster lockdown changes. The postmortem in giantswarm/giantswarm#1953 is also closed and the orphan VPC check is deployed.