[Openstack-operators] Manually nuking a misbehaving instance

Gabe Westmaas gabe.westmaas at RACKSPACE.COM
Wed Sep 19 03:32:02 UTC 2012

Hey Sam,

There's an extension to let you reset the task state to "error" which
often lets you move forward with a delete if its just an invalid task or
vm state that is preventing the issue.  This only works if the services
are all up and functioning properly when you attempt the delete the second
time.  Also this showed up in the folsom release, so if you aren't running
that, this path won't help!  You can do it manually by setting the states
in the DB, and then attempting the delete again.

This strategy doesn't work if the service is still not behaving, or if the
data in the db is just wrong (has the wrong compute node or something


On 9/18/12 11:19 PM, "Samuel Winchenbach" <swinchen at gmail.com> wrote:

>Hi All,
>I was wondering what the correct way to handle a misbehaving instance
>is?  I have, on several occasions, gotten VM instances stuck in an odd
>state such as reboot, deleting, or active when either KVM failed to
>start, or who knows what.   What is the correct procedure for manually
>removing/repairing a misbehaving VM?
>Right now I have been using this SQL script: http://paste2.org/p/2240202
>The problem is, this does not undo all the networking changes, or
>delete the various files in /var/lib/nova/instances and
>Is there a better way to handle this situation than completely nuking
>the database?  If not then what do I need to do to undo the changes to
>the network/iptables?
>OpenStack-operators mailing list
>OpenStack-operators at lists.openstack.org

More information about the OpenStack-operators mailing list