Hey openstack-discuss, we use Ceph RadosGW (Quincy release) to provide Swift as part of an OpenStack cloud. The endpoints are configured as such:
# openstack endpoint list +----------------------------------+--------+--------------+----------------+---------+-----------+----------------------------------------------------------------------+ | ID | Region | Service Name | Service Type | Enabled | Interface | URL | +----------------------------------+--------+--------------+----------------+---------+-----------+----------------------------------------------------------------------+ | 1234567890 | region | swift | object-store | True | public | https://object-store.region.cloud.example.com/swift/v1/AUTH_%(tenant_id)s | | 2345678901 | region | swift | object-store | True | internal | https://object-store.region.cloud.example.com/swift/v1/AUTH_%(tenant_id)s | | 3456789012 | region | swift | object-store | True | admin | https://object-store.region.cloud.example.com/swift/v1/AUTH_%(tenant_id)s |
so according to https://docs.ceph.com/en/latest/radosgw/keystone/#cross-project-tenant-acces..., if I am not mistaken. I can also use the openstack client to access containers:
# openstack container list +---------------+ | Name | +---------------+ | containera | | containerb | +---------------+
and there is no warnings or other apparent issues. But when using "project cleanup" via e.g. "openstack project cleanup --dry-run --project $PROJECT" I see multiple warnings like
Failed to contact the endpoint at https://object-store.region.cloud.example.com/swift/v1/AUTH_f2bc4bd34567ddc3... for discovery. Fallback to using that endpoint as the base url.
but only mentioning the swift endpoint. Following the warning string this originates from the discovery method of keystoneauth1, see https://github.com/openstack/keystoneauth/blob/28048af9593740315df8d9027c3bd.... Could anybody help me understand what is issue might be? Why does this warning not appear for every use of object storage / swift via openstackclient? Is there any way for me to "fix" this? Could this even be a bug? Thanks and with kind regards Christian