Cluster bootstrap & upgrade nuances 🔴
onedr0p opened this issue · comments
Devin Buhl commented
There are concerns that need to addressed when re-provisioning my Talos cluster...
1. Pre bootstrap
- Sync Talos config files to Matchbox
- CloudnativePG needs to have the cluster spec updated with the new
.spec.backup.barmanObjectStore.serverName
and.spec.bootstrap.recovery.source
2. Post bootstrap
-
Sometimes the rook operator pod needs to be deleted so it can re-provision the OSDs after the first set of OSD prepare jobs appear (and fail). There is some testing to be done with
nsenter
in the wipe-disk jobs that might clear this up.kubectl -n rook-ceph delete pod --selector app=rook-ceph-operator
-
Volsync doesn't clean up temp volumes after the volume populator has ran
task volsync:cleanup cluster=main