Test instance manager cleanup during uninstall
- Deploy Longhorn v1.1.2
- Launch some running volumes.
- Upgrade to v1.2.0. ==> All old unused engine managers should be cleaned up automatically.
- Make sure all running volumes keep state
running
.
- Upgrade all volumes. ==> All old replica managers should be cleaned up automatically.
- Detach all running volumes. ==> All old engine managers should be cleaned up automatically.
- do offline upgrade then reattach these volumes.
- Directly uninstall the Longhorn system.
And use
kubectl -n longhorn-system get lhim -w
to verify that the system doesn’t loop in instance manager cleanup-recreation.
==> The uninstaller should work fine.
[Edit]