Replacing ceph by ceph-mon leaves old ceph IP in /etc/ceph/ceph.conf

Bug #1820041 reported by Alvaro Uria
This bug affects 1 person
Affects Status Importance Assigned to Milestone
OpenStack Cinder Charm

Bug Description

* trusty-mitaka using latest charmstore revs
* single cinder unit, with "vip" set and related to hacluster
* single ceph unit, with no osd devices

When I run the following steps, taken from [1]:
juju deploy --config no-bootstrap=True cs:trusty/ceph-mon
juju add-unit -n2 ceph-mon
# wait until ceph-mon units settle
juju add-relation ceph ceph-mon
# wait until units settle
juju add-relation ceph-mon ceph-osd
# wait until units settle
juju add-relation ceph-mon cinder
juju add-relation ceph-mon glance
juju add-relation ceph-mon nova-compute

# wait until units settle
juju remove-application ceph

When I check glance's /etc/ceph/ceph.conf, new mon IPs are configured. However, cinder unit has the old IP and I needed to fix it by running:
juju run --unit cinder/0 hooks/config-changed


Revision history for this message
Alvaro Uria (aluria) wrote :

Sorry, the above hooks/config-changed run fixed ceph.conf, but the rbd driver could not be initialized. I had to remove/add the relation between cinder and ceph-mon to fix it.

Frode Nordahl (fnordahl)
Changed in charm-cinder:
status: New → Triaged
importance: Undecided → High
tags: added: charm-upgrade
tags: removed: charm-upgrade
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.