Created attachment 847290 [details] log Description of problem: there's a problem launching an instance with create a volume (there is a time out on create volume which causes the instance to moves to error before the volume finish to create). If I selected "delete on terminate" option, the volumes (yes, a second bug creating two volumes) are not deleted if I destroy the instance. Version-Release number of selected component (if applicable): openstack-cinder-2014.1-0.2.b1.el6.noarch openstack-nova-compute-2014.1-0.5.b1.el6.noarch python-django-horizon-2014.1-0.1b1.el6.noarch How reproducible: 100% Setup: stand alone cinder with gluster backend stand alone glance with gluster backend nova + all other component + one more compute all physical servers Steps to Reproduce: 1. launch an instance with option create new volume + select the delete on terminate check box 2. after the instance moves to error state delete the instance 3. Actual results: there are two volume deleted (bug which is opened) and none of them are deleted when the instance is destroyed Expected results: we should be able to follow the tasks sent so that even if the instance is moved to error thinking that the volume was not created that a delete volume will be sent in any case the check box is selected. Additional info: logs
delete on terminate does not work when using launch from volume (create new volume) option.
Dafna, thanks for the comment, still a couple of points: - It'd be useful if you could post relevant contextual log messages/stack trace here instead of attaching a tar ball. This just saves time to debug/analysze. - "Steps to Reproduce" is a bit sparse. Would be useful to also provide a CLI-based reproducer. (I could infer you tried this via - Gentle reminder - https://wiki.openstack.org/wiki/BugFilingRecommendations
Steps to Reproduce: (already existed in the bug). 1. launch an instance with option create new volume + select the delete on terminate check box 2. after the instance moves to error state delete the instance 3. This is done from horizon - I attached all the logs
The volume log shows numerous qpid Heartbeat Timeouts. It's possible the volume service never got the request to delete the volume.
I am not able to reproduce this behavior with current (RDO Juno) packages.