Reinstall a Host From a Rook Ceph Cluster

Note

Reinstall the hosts sequentially, completing the full process on each host before moving on to the next.

Reinstall a Host

#. Remove all OSDs from the host via the Rook Ceph cluster configuration, before reinstalling a host that runs Ceph OSDs.

Warning

Removing OSDs from the system can cause data loss. For more information on how to delete OSDs and how to prevent data loss, see Rook Ceph OSD Management.

#. Obtain the UUID of the OSDs, mark them for forced removal, then reapply the Rook Ceph application to remove all OSDs by force.

# Get the UUID of the OSD
$ system host-stor-list <host>

# Mark each OSD for deletion
$ system host-stor-delete <uuid> --force

# Reapply the application
$ system application-apply rook-ceph

# Check OSD removal
$ ceph -s
$ kubectl get pods -n rook-ceph -l app=rook-ceph-osd

  1. Reinstall the host, after removing all OSDs.

# Lock the host
$ system host-lock <host>

# Check if system is locked
$ system host-list

# With the host locked, reinstall the system
$ system host-reinstall <host>

# Check if the host is reinstalling
$ system host-show <host>

  1. Unlock the host.

# Unlock the host
$ system host-unlock <host>

  1. Re-add the OSDs to the system.

# Check disk UUID
$ system host-disk-list <host>

# Add the OSD
$ system host-stor-add <host> <disk_UUID>

# Reapply the Rook Ceph application
$ system application-apply rook-ceph

Note

For more information on how to add OSDs, see Rook Ceph OSD Management.