cinder services reported as "DOWN" when ceph is used as backend

[Lirim]

cinder services show as being "down" and do not update in the dashboard when ceph is used as a storage backend. However the services (cinder and ceph) seems to be running ok in the background with no problems for creating/attaching/destroying volumes.

[Edward Hope-Morley (hopem)]

Lirim can you please provide the following:
  * output of 'cinder service-list'
  * output of juju status

Also, are you using the cinder-ceph subordinate to configure cinder-volume or are you using native support in the cinder charm?

[Lirim (lirim-osmani)]

Hi Edward,

see below the output of :

 cinder service-list

+------------------+----------------------+------+---------+-------+----------------------------+

| Binary | Host | Zone | Status | State
| Updated_at |

+------------------+----------------------+------+---------+-------+----------------------------+

| cinder-scheduler | cinder | nova | enabled | up |
2016-01-22T17:26:51.000000 |

| cinder-scheduler | juju-machine-1-lxc-2 | nova | enabled | down |
2015-11-27T19:32:33.000000 |

| cinder-scheduler | juju-machine-2-lxc-2 | nova | enabled | down
| None |

| cinder-scheduler | juju-machine-3-lxc-2 | nova | enabled | down
| None |

| cinder-volume | cinder@cinder-ceph | nova | enabled | up |
2016-01-22T17:26:46.000000 |

| cinder-volume | juju-machine-1-lxc-2 | nova | enabled | down |
2015-11-27T19:32:19.000000 |

| cinder-volume | juju-machine-2-lxc-2 | nova | enabled | down |
2015-11-27T19:31:53.000000 |

| cinder-volume | juju-machine-3-lxc-2 | nova | enabled | down |
2015-11-27T19:31:31.000000 |

+------------------+----------------------+------+---------+-------+----------------------------+

and the juju status output is attached.

Let me know if you require anything else.

Lirim

On Thu, Jan 21, 2016 at 9:52 PM, Edward Hope-Morley <
<email address hidden>> wrote:

> Lirim can you please provide the following:
> * output of 'cinder service-list'
> * output of juju status
>
> Also, are you using the cinder-ceph subordinate to configure cinder-
> volume or are you using native support in the cinder charm?
>
> ** Changed in: openstack-dashboard (Juju Charms Collection)
> Status: New => Incomplete
>
> --
> You received this bug notification because you are subscribed to the bug
> report.
> https://bugs.launchpad.net/bugs/1521728
>
> Title:
> cinder services reported as "DOWN" when ceph is used as backend
>
> Status in openstack-dashboard package in Juju Charms Collection:
> Incomplete
>
> Bug description:
> cinder services show as being "down" and do not update in the
> dashboard when ceph is used as a storage backend. However the services
> (cinder and ceph) seems to be running ok in the background with no
> problems for creating/attaching/destroying volumes.
>
> To manage notifications about this bug go to:
>
> https://bugs.launchpad.net/charms/+source/openstack-dashboard/+bug/1521728/+subscriptions
>

[Edward Hope-Morley (hopem)]

Hi Lirim, this is actually a symptom of the way cinder-volume is deployed and not really a bug (although it is something that we are aiming to improve on - see https://bugs.launchpad.net/charms/+source/cinder-ceph/+bug/1492748). Basically the issue here is that when cinder-volume is deployed, if it is using a stateless volume backend driver, its 'host' value will change and thus it will be registered in cinder with that new name. However, the old entries i.e. with the old host name are not removed (since there is no easy way to do this other than manually delete them from the database). So this is why they appear in the output of 'cinder service-list' or in horizon. We are working on a way to improve this as part of bug 1492748.