Open Stack

Roger, Sayali,

Are there any more races that we might need to fix in the existing cluster
for Juno? Apart from the ones mentioned above and metadata service being
flaky for Juno?

Regards,
Pranav

On Mon, Mar 30, 2015 at 8:15 PM, Roger Luethi <rl at patchworkscience.org>
wrote:

> On Fri, 27 Mar 2015 23:52:52 +0530, Sayali Lunkad wrote:
> >    - *Host not found while launching an instance.* (happens very
> frequently
> >    when they try  to launch the instance for the first time). It
> normally gets
> >    resolved by restarting the compute node or increasing the RAM for
> compute
> >    and then restarting it. I am not sure why this should happen only on
> some
> >    machines though even thought the configuration of  the machines was
> the
> >    same.
>
> This sounds like the controller not realizing that the space for an
> instance has been freed (our default config has enough space for only one
> instance). If so, just wait a minute or so after deleting all instances,
> then it should work.
>
> >    - *HTTP 500 errors*. This is notices frequently after a using the CLI
> >    for sometime. After running a few CLI commands this error shows up.
>
> There might be some useful information in the logs when that happens.
>
> >    - *Cluster breaks at horizon repeatedly about 4 times. *As spotted
> >    earlier there is a race while osbash tried to setup horizon. It is
> found
> >    that this could be a problem with apache2 server when you start it
> for the
> >    first time. But the root cause is unknown. We should take a look at
> this as
> >    it is happening frequently.
>
> Hopefully fixed by https://review.openstack.org/#/c/168711/.
>
> >    - *Cluster breaks while setting up Nova compute. *The second time the
> >    cluster breaks is while setting up nova compute. On checking the logs
> >    it is shows that it is unable to fetch some archives. But again I
> >    don't see why it would only break at nova compute if it was a problem
> >    with the connection.
>
> We pre-download everything into the basedisk so we don't need a network
> connection when building the cluster. Only I failed to add a new package
> to pre-download when updating the scripts for Juno. Patch forthcoming.
>
> >    - *VMs do not poweroff automatically on windows host. *Bug has been
> >    filed for this by one of the attendees.
> >    https://bugs.launchpad.net/openstack-training-guides/+bug/1435932
>
> Not sure I understand that bug report. On all operating systems, the
> cluster is up and running after the cluster build is finished. You can
> log in to shut it down, or you can just power off the node VMs and
> restore them to the last snapshot. What am I missing?
>
> Roger
>
> _______________________________________________
> OpenStack-docs mailing list
> OpenStack-docs at lists.openstack.org
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-docs
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-docs/attachments/20150413/ea735586/attachment-0001.html>