kubernikus
kubernikus copied to clipboard
sapcc
Waits for Load Balancer Deletions via Finalizers
This commit changes the wait behaviour when deleteing load balancers. Previously service deleteion didn't use finalizers and no feedback was given. The service disapeared immediately. The deorbiter used a fixed timeout of 2min to wait for load balancer cleanup.
Since the loadbalancer driver in CCEE was changed to Octavia, that grace period is not enough anymore. Deleteion takes longer and in case of 1.16 never completes if the LoadBalancer is not completely provisioned, e.g. due to quota limitations. Especially a FIP is expected to be attached and causes the deletion to never complete.
This fix depends on Kubernetes 1.17 to be fully functional. For
Kubernetes < 1.17 the Deorbiter will wait 24h for the service to be
deleted. For 1.16 it is possible to set the use-octavia flag as
a workaround.
This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.
![stale[bot] avatar](https://avatars.githubusercontent.com/in/1724?v=4 "Published by a pub.dev verified publisher"){kind=small}
@BugRoger Whats this this little nugget I found in the PR list? Is it ready to be merged?
Should we maybe keep the current behaviour for <1.16 clusters (we do set use-octavia) for a while now. I just would like to avoid customer inquiries about the old cluster stuck in terminating.
This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.