Skip to content

Cleanup

Cleaning up a Cluster

To tear down the cluster, the following resources need to be cleaned up:

  • The resources created under Rook's namespace (default rook-ceph) such as the Rook operator created by operator.yaml and the cluster CR cluster.yaml.
  • /var/lib/rook/rook-ceph: Path on each host in the cluster where configuration is stored by the ceph mons and osds
  • Devices used by the OSDs

If the default namespaces or paths such as dataDirHostPath are changed in the example yaml files, these namespaces and paths will need to be changed throughout these instructions.

If tearing down a cluster frequently for development purposes, it is instead recommended to use an environment such as Minikube that can easily be reset without worrying about any of these steps.

Delete the Block and File artifacts

First clean up the resources from applications that consume the Rook storage.

These commands will clean up the resources from the example application block and file walkthroughs (unmount volumes, delete volume claims, etc).

1
2
3
4
5
6
kubectl delete -f ../wordpress.yaml
kubectl delete -f ../mysql.yaml
kubectl delete -n rook-ceph cephblockpool replicapool
kubectl delete storageclass rook-ceph-block
kubectl delete -f csi/cephfs/kube-registry.yaml
kubectl delete storageclass csi-cephfs

Important

After applications have been cleaned up, the Rook cluster can be removed. It is important to delete applications before removing the Rook operator and Ceph cluster. Otherwise, volumes may hang and nodes may require a restart.

Delete the CephCluster CRD

Warning

DATA WILL BE PERMANENTLY DELETED AFTER DELETING THE CephCluster

  1. To instruct Rook to wipe the host paths and volumes, edit the CephCluster and add the cleanupPolicy:
kubectl -n rook-ceph patch cephcluster rook-ceph --type merge -p '{"spec":{"cleanupPolicy":{"confirmation":"yes-really-destroy-data"}}}'

Once the cleanup policy is enabled, any new configuration changes in the CephCluster will be blocked. Nothing will happen until the deletion of the CR is requested, so this cleanupPolicy change can still be reverted if needed.

Checkout more details about the cleanupPolicy here

  1. Delete the CephCluster CR.
kubectl -n rook-ceph delete cephcluster rook-ceph
  1. Verify that the cluster CR has been deleted before continuing to the next step.
kubectl -n rook-ceph get cephcluster

If the cleanupPolicy was applied, wait for the rook-ceph-cleanup jobs to be completed on all the nodes. These jobs will perform the following operations:

  • Delete the namespace directory under dataDirHostPath, for example /var/lib/rook/rook-ceph, on all the nodes
  • Wipe the data on the drives on all the nodes where OSDs were running in this cluster

Note

The cleanup jobs might not start if the resources created on top of Rook Cluster are not deleted completely. See deleting block and file artifacts

Delete the Operator Resources

Remove the Rook operator, RBAC, and CRDs, and the rook-ceph namespace.

1
2
3
kubectl delete -f operator.yaml
kubectl delete -f common.yaml
kubectl delete -f crds.yaml

Delete the data on hosts

Attention

The final cleanup step requires deleting files on each host in the cluster. All files under the dataDirHostPath property specified in the cluster CRD will need to be deleted. Otherwise, inconsistent state will remain when a new cluster is started.

If the cleanupPolicy was not added to the CephCluster CR before deleting the cluster, these manual steps are required to tear down the cluster.

Connect to each machine and delete the namespace directory under dataDirHostPath, for example /var/lib/rook/rook-ceph.

Zapping Devices

Disks on nodes used by Rook for OSDs can be reset to a usable state. Note that these scripts are not one-size-fits-all. Please use them with discretion to ensure you are not removing data unrelated to Rook.

A single disk can usually be cleared with some or all of the steps below.

DISK="/dev/sdX"

# Zap the disk to a fresh, usable state (zap-all is important, b/c MBR has to be clean)
sgdisk --zap-all $DISK

# Wipe a large portion of the beginning of the disk to remove more LVM metadata that may be present
dd if=/dev/zero of="$DISK" bs=1M count=100 oflag=direct,dsync

# SSDs may be better cleaned with blkdiscard instead of dd
blkdiscard $DISK

# Inform the OS of partition table changes
partprobe $DISK

Ceph can leave LVM and device mapper data that can lock the disks, preventing the disks from being used again. These steps can help to free up old Ceph disks for reuse. Note that this only needs to be run once on each node. If you have only one Rook cluster and all Ceph disks are being wiped, run the following command.

1
2
3
4
5
6
# This command hangs on some systems: with caution, 'dmsetup remove_all --force' can be used
ls /dev/mapper/ceph-* | xargs -I% -- dmsetup remove %

# ceph-volume setup can leave ceph-<UUID> directories in /dev and /dev/mapper (unnecessary clutter)
rm -rf /dev/ceph-*
rm -rf /dev/mapper/ceph--*

If disks are still reported locked, rebooting the node often helps clear LVM-related holds on disks.

If there are multiple Ceph clusters and some disks are not wiped yet, it is necessary to manually determine which disks map to which device mapper devices.

Troubleshooting

The most common issue cleaning up the cluster is that the rook-ceph namespace or the cluster CRD remain indefinitely in the terminating state. A namespace cannot be removed until all of its resources are removed, so determine which resources are pending termination.

If a pod is still terminating, consider forcefully terminating the pod (kubectl -n rook-ceph delete pod <name>).

kubectl -n rook-ceph get pod

If the cluster CRD still exists even though it has been deleted, see the next section on removing the finalizer.

kubectl -n rook-ceph get cephcluster

Removing the Cluster CRD Finalizer

When a Cluster CRD is created, a finalizer is added automatically by the Rook operator. The finalizer will allow the operator to ensure that before the cluster CRD is deleted, all block and file mounts will be cleaned up. Without proper cleanup, pods consuming the storage will be hung indefinitely until a system reboot.

The operator is responsible for removing the finalizer after the mounts have been cleaned up. If for some reason the operator is not able to remove the finalizer (i.e., the operator is not running anymore), delete the finalizer manually with the following command:

1
2
3
4
for CRD in $(kubectl get crd -n rook-ceph | awk '/ceph.rook.io/ {print $1}'); do
    kubectl get -n rook-ceph "$CRD" -o name | \
    xargs -I {} kubectl patch -n rook-ceph {} --type merge -p '{"metadata":{"finalizers": []}}'
done

If the namespace is still stuck in Terminating state, check which resources are holding up the deletion and remove their finalizers as well:

kubectl api-resources --verbs=list --namespaced -o name \
  | xargs -n 1 kubectl get --show-kind --ignore-not-found -n rook-ceph

Remove critical resource finalizers

Rook adds a finalizer ceph.rook.io/disaster-protection to resources critical to the Ceph cluster so that the resources will not be accidentally deleted.

The operator is responsible for removing the finalizers when a CephCluster is deleted. If the operator is not able to remove the finalizers (i.e., the operator is not running anymore), remove the finalizers manually:

kubectl -n rook-ceph patch configmap rook-ceph-mon-endpoints --type merge -p '{"metadata":{"finalizers": []}}'
kubectl -n rook-ceph patch secrets rook-ceph-mon --type merge -p '{"metadata":{"finalizers": []}}'