Hi,
Sorry for not being totally clear. The cluster is managed by TripleO. After each deployment/update, the cinder configuration is updated with incorrect names. Currently we correct it manually in the cinder configuration. So it is not only aesthetic.
Sinan
Op 11 nov. 2019 om 14:48 heeft Gorka Eguileor geguileo@redhat.com het volgende geschreven:
On 07/11, Sinan Polat wrote: Hi,
I am using Ceph as the backend for Cinder. Within Ceph we have defined 2 RBD pools (ssdvolumes, sasvolumes). In OpenStack I created 2 Volume Types (ssd, sas). Each Volume Type has property "volume_backend_name='tripleo_ceph_<ssd|sas>'".
In the Cinder configuration I have the following backends configured:
[tripleo_ceph_ssd] backend_host=hostgroup volume_backend_name=tripleo_ceph_ssd volume_driver=cinder.volume.drivers.rbd.RBDDriver rbd_ceph_conf=/etc/ceph/ceph.conf rbd_user=openstack rbd_pool=ssdvolumes
[tripleo_ceph_sas] backend_host=hostgroup volume_backend_name=tripleo_ceph_sas volume_driver=cinder.volume.drivers.rbd.RBDDriver rbd_ceph_conf=/etc/ceph/ceph.conf rbd_user=openstack rbd_pool=sasvolumes
As you might have noticed, the backend name (tripleo_ceph_ssd) and the RBD pool name (ssdvolumes, not ssd) does not match. So far, we do not have any problems. But I want to correct the names and I do not want to have the mismatch anymore.
So I want to change the value of key volume_backend_name for both Volume Types (tripleo_ceph_ssd => tripleo_ceph_ssdvolumes).
Hi,
I agree with Sean, I wouldn't change it since this is only aesthetic.
Having said that, there's always a way to do most things, even if it's NOT RECOMMENDED:
- Update cinder.conf
- Get the volume type id for the 2 volume types to change
- Stop cinder services
- Go into the DB and manually update the volume types changes in the
"volume_type_extra_specs" table filtering by the volume_type_id and the key "volume_backend_name" and setting the new "value".
- Use the "cinder-manage volume update_host" to update existing volumes
to the new backend (you could also do this directly in the DB).
- Start cinder services
- Remove the old service from the DB (they appear as down now) using the
"cinder-manage service remove" command.
Cinder-manage docs: https://docs.openstack.org/cinder/latest/cli/cinder-manage.html
Regards, Gorka.
I tried the following: $ openstack volume type show 80cb25ff-376a-4483-b4f7-d8c75839e0ce +--------------------+----------------------------------------+ | Field | Value | +--------------------+----------------------------------------+ | access_project_ids | None | | description | | | id | 80cb25ff-376a-4483-b4f7-d8c75839e0ce | | is_public | True | | name | ssd | | properties | volume_backend_name='tripleo_ceph_ssd' | | qos_specs_id | None | +--------------------+----------------------------------------+ $
$ openstack volume type set --property volume_backend_name='tripleo_ceph_ssdvolumes' 80cb25ff-376a-4483-b4f7-d8c75839e0ce Failed to set volume type property: Volume Type is currently in use. (HTTP 400) (Request-ID: req-5efaa5b7-910f-4802-8494-3115cfc4ab93) Command Failed: One or more of the operations failed $
How to solve my problem?
Thanks!
Sinan