[openstack-dev] [tripleo] ha & upgrade jobs are broken
emilien at redhat.com
Tue Jun 28 19:39:34 UTC 2016
On Mon, Jun 27, 2016 at 10:47 PM, Emilien Macchi <emilien at redhat.com> wrote:
> CI is still in bad shape, even if we fixed non-ha job .
> See https://bugs.launchpad.net/tripleo/+bug/1596758 for the new issue
> (ha & upgrade job look broken), it seems related to Pacemaker.
> I investigated a little bit (late here) and it seems like a problem
> with Pacemaker (/var/log/host_info.txt on controller-1 shows a timeout
> when trying to contact the cluster).
> If anyone can look during the morning, otherwise I'll continue
> tomorrow. Any help is warmly welcome!
>  https://review.openstack.org/#/c/334555/
We managed to have the 3 jobs (non-ha, ha and upgrade) green again.
Here's what we did:
- increase openstackclient calls timeout:
https://review.openstack.org/334996 (so we should not have timeouts in
our CI anymore when creating Keystone_user[admin] resource.
Note about this one: this issue will disappear as soon as we
implement the new HA architecture where httpd won't be manage by
Pacemaker anymore iiuc.
- fix a regression (ssl port removal logic in postconfig) in
Kudos to those who helped to track this down.
You can now rebase/recheck your patches so get CI runs again.
Please let us know here if you still see some errors.
Don't hesitate to use tripleo.org/cistatus.html as a reference for
general CI forecast.
More information about the OpenStack-dev