kubedirector
kubedirector copied to clipboard
"stop KDCluster" feature
I'm seeing some usecases for this in various discussion.
One question is how to model a "stop" and "restart" request through edits to the KDCluster spec.
For implementation, we'd need to make sure that any PVs are preserved when we scale the statefulsets down to 0 replicas. On restart we'd need to scale the statefulsets back up to the specified replicas count and re-attach the PVs. (I am also curious about whether the PVs would be vulnerable to being claimed by someone else during this period.)