While deploying DKP, you may encounter an issue while Kommander installs various Helm Releases in the cluster where Rook Ceph never becomes ready. To identify if you're hitting this issue, run the following command and look in status of the cephcluster for the following message:
kubectl get cephcluster -n kommander dkp-ceph-cluster -o yaml
- lastHeartbeatTime: "2023-05-02T02:13:07Z"
lastTransitionTime: "2023-05-02T01:55:09Z"
message: 'CephCluster "kommander/dkp-ceph-cluster" will not be deleted until all
dependents are removed: CephObjectStore: [dkp-object-store]'
Confirm that the cephobjectstore object has a status of progressing:
kubectl get cephobjectstore dkp-object-store -o yaml -n kommander
status:
endpoints:
insecure:
- http://rook-ceph-rgw-dkp-object-store.kommander.svc:80
secure: []
info:
endpoint: http://rook-ceph-rgw-dkp-object-store.kommander.svc:80
phase: Progressing
The cephobjectstore will never delete on its own as it has finalizers that prevent this, which causes the Rook Ceph Helm Release to time out. To fix this, remove the finalizers from the cephobjectstore object. It will then be deleted as will the cephcluster. If the helm release has already timed out, delete it and it will resume installation.