The cleanup job will need to run wherever the ceph daemons were running. The scheduling is done by rook on the specific nodes where the daemons were running, so there is no need for other placement such as node or pod affinity. Signed-off-by: Travis Nielsen <tnielsen@redhat.com>