Instructions to forcibly cleanup OCS

ocs_cleanup.bash

#!/bin/bash
set -x

#Delete any apps consuming PVCs based on OCS storageclasses, and then delete the PVs
#Delete the StorageCluster object without deleting its dependents.
oc delete -n openshift-storage storagecluster --all --wait=true --cascade=false
#Delete the Noobaa resource.
oc delete -n openshift-storage noobaa noobaa --wait=true 
#Wait for the Noobaa PVC to be automatically deleted.
oc wait -n openshift-storage --for delete pvc -l noobaa-core=noobaa --timeout=5m
#Delete the CephCluster resource and wait for it to finish deleting.
oc delete -n openshift-storage cephcluster --all --wait=true  --timeout=5m
#Delete the Namespaces and wait for them to finish deleting.
oc delete project openshift-storage --wait=true  --timeout=5m
# remove the taint from storage nodes
oc adm taint nodes --all node.ocs.openshift.io/storage-
# unlabel the nodes the storage nodes
oc label node  --all cluster.ocs.openshift.io/openshift-storage-

First list PVCs List pvcs that must be thoughtfully deleted:

# for the Ceph RBD StorageClass
oc get pvc -o=jsonpath='{range .items[?(@.spec.storageClassName=="ocs-storagecluster-ceph-rbd")]}{"Name: "}{@.metadata.name}{" Namespace: "}{@.metadata.namespace}{"\n"}{end}' --all-namespaces|grep -v db-noobaa-core-0

# and for the CephFS StorageClass
oc get pvc -o=jsonpath='{range .items[?(@.spec.storageClassName=="ocs-storagecluster-cephfs")]}{"Name: "}{@.metadata.name}{" Namespace: "}{@.metadata.namespace}{"\n"}{end}' --all-namespaces

Changleog:

• Remove taints as well
• Handle noobaa pvc deadlock.
• Don't refer to the cephcluster and storage cluster by name.
• Don't delete local-storage project.

Use this to check that everything is gone:

# wait for the namespace to terminate
while oc get ns|grep "openshift-storage"
do
    sleep 1
done