Open Stack

Hi nico,

This may be related to another thread currently on the ML about multi-node
devstack.

There seems to be a bug in multi-node devstack that if you do not specify
"rabbit" as one of the enabled services on each compute node, the
nova-compute on the compute node tries to contact the rabbit server on
localhost, rather than on the IP specified by RABBIT_HOST.

If the compute nodes are not correctly connecting to rabbit, I believe the
scheduler would not be able to notify those servers to create VMs, which
*may* explain why your instances are getting stuck in scheduling. Looking
at the nova-compute log on your compute hosts could probably help determine
if this is the issue, as there would be errors due to the failure to
connect to rabbit. Putting "rabbit" in the set of ENABLED_SERVICES on all
nodes seems to have worked for me as a work around.

I posted a question to the devstack team about this:
https://answers.launchpad.net/devstack/+question/197749

Dan



On Wed, May 23, 2012 at 11:05 AM, Nicolae Paladi <n.paladi at gmail.com> wrote:

> Hi, a very similar question has been asked in
> https://lists.launchpad.net/openstack/msg10125.html,
> however I could not find a follow-up.
>
> I have followed the instructions for
> http://devstack.org/guides/multinode-lab.html,
> however I am running Ubuntu 12.04 on both hosts.
>
> There are 2 issues:
>
> 1. The 'services' panel in the dashboard only displays the services
> running on
> the controller, while nova-manage service list on both hosts displays:
>
> nova-compute     duff                                 nova
> enabled    :-)   2012-05-23 18:01:14
> nova-cert        duff                                 nova
> enabled    :-)   2012-05-23 18:01:18
> nova-volume      duff                                 nova
> enabled    :-)   2012-05-23 18:01:24
> nova-scheduler   duff                                 nova
> enabled    :-)   2012-05-23 18:01:18
> nova-consoleauth duff                                 nova
> enabled    :-)   2012-05-23 18:01:19
> nova-network     duff                                 nova
> enabled    :-)   2012-05-23 18:01:16
> nova-compute     cirrus                               nova
> enabled    :-)   2012-05-23 18:01:17
> nova-volume      cirrus                               nova
> enabled    :-)   2012-05-23 18:01:16
> nova-network     cirrus                               nova
> enabled    :-)   2012-05-23 18:01:18
>
>
> 2. Once an instance is launched, it remains in the "scheduling" state with
> no progress.
>
> The logs look fine, no errors there.
>
> Thanks!
> /nico
>
> _______________________________________________
> Mailing list: https://launchpad.net/~openstack
> Post to     : openstack at lists.launchpad.net
> Unsubscribe : https://launchpad.net/~openstack
> More help   : https://help.launchpad.net/ListHelp
>
>


-- 
~~~~~~~~~~~~~~~~~~~~~~~~~~~
Dan Wendlandt
Nicira, Inc: www.nicira.com
twitter: danwendlandt
~~~~~~~~~~~~~~~~~~~~~~~~~~~
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack/attachments/20120523/e7a1530e/attachment.html>