You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Commit ff48dfa
Browse filesBrowse the repository at this point in the historyBrowse files
fix(hibernation): report WaitingForHealthy condition for non-healthy clusters (#10193)
When the hibernation annotation is set on a non-healthy cluster (e.g.
CrashLoopBackOff), the operator correctly defers hibernation until the
cluster recovers. However, this decision was previously silent — no
condition was set, leaving `cnpg status` unable to report why
hibernation had not started.
This PR addresses the missing reporting aspect of the issue. Non-healthy
clusters with the hibernation annotation now receive a
`WaitingForHealthy` condition (`Status=False`), providing clear feedback
through `cnpg status` where previously there was none. The underlying
behavior — deferring hibernation until the cluster is healthy — remains
unchanged.
No pre-existing behavior is altered:
- Healthy clusters requesting hibernation still proceed immediately
- Already-ongoing hibernation (DeletingPods, WaitingPodsDeletion,
Hibernated) still bypasses the healthy-phase guard
- Clusters without the annotation remain unaffected
Note: this does not solve the broader problem of hibernating a cluster
that will never become healthy on its own (e.g. persistent
CrashLoopBackOff). That would require a separate discussion about
allowing forced hibernation when PVCs are already fully provisioned.
Ref #10024
Signed-off-by: Armando Ruocco <[email protected]>
Signed-off-by: Marco Nenciarini <[email protected]>
Co-authored-by: Marco Nenciarini <[email protected]>
0 commit comments