Ceph » rgw

Rook is creating zones and users using :

radosgw-admin --version
ceph version 14.2.0 (3a54b2b6d167d4a2a19e003a705696d4fe619afc) nautilus (stable)

while running radosgw version 13.2.5:

grep "image:" * 
cluster.yaml:23:    image: ceph/ceph:v13.2.5-20190319

radosgw-admin v14.x is adding a slash from the storage class to the default placement even when the storage class is 'standard'

oc -n rook-ceph exec -it $(oc -n rook-ceph get pod -l "app=rook-ceph-rgw" -o jsonpath='{.items[0].metadata.name}') -- bash
radosgw-admin --keyring=/etc/ceph/keyring-store/keyring --mon-host=$ROOK_CEPH_MON_HOST --name=client.radosgw.gateway  metadata get user:cosbench
"default_placement": "/" 
-------------------->^^^

radosgw-admin --keyring=/etc/ceph/keyring-store/keyring --mon-host=$ROOK_CEPH_MON_HOST --name=client.radosgw.gateway  zonegroup get
{
    "placement_targets": [
        {
            "name": "default-placement",
            "tags": []
        }
    ],
    "default_placement": "default-placement/",
------------------------------------------>^
    "realm_id": "52539210-99bc-449b-9bae-00e60bd5064a" 
}

Which is causing the s3cmd error:

s3cmd mb s3://bkt
ERROR: S3 error: 400 (InvalidLocationConstraint)

and the rgw log shows:

2019-04-10 14:35:37.201 7fc747b96700  0 could not find user default placement id / within zonegroup

Reference:
[1] rook tracker issue: https://github.com/rook/rook/issues/2945