[OpenStack-Infra] [Third-party-announce] Four Cisco CI systems are disabled

Bob Ball bob.ball at citrix.com
Tue Mar 8 17:53:53 UTC 2016


I had a consistent issue with zuul-merger failing to merge due to incorrect host keys set up by the community puppet modules.

See https://review.openstack.org/#/c/266041/ for the fix, but note that we must have a newline at the end of the gerrit_ssh_host_key.

You can easily confirm if you are having the same issue as me by checking ~zuul/.ssh/known_hosts after starting the merger.
If you have two known_hosts entries on one line (check for the pipe (|) character right in the middle of the line) then it's the same issue.

It appears that if the gerrit_ssh_host_key is incorrect then however zuul-merger is adding the new key to the known_hosts file does not check for / add a new line if it's missing.

Bob 

-----Original Message-----
From: Clark Boylan [mailto:cboylan at sapwetik.org] 
Sent: 08 March 2016 17:36
To: Wan, Sam <Sam.Wan at emc.com>; openstack-infra at lists.openstack.org
Subject: Re: [OpenStack-Infra] [Third-party-announce] Four Cisco CI systems are disabled

The best way to determine why zuul is failing to merge changes is to check your zuul-merger debug log. If you are using the community puppet module these should end up in /var/log/zuul/merger-debug.log. When a merge fails the mergers should log the exact error from git in this log file.

We run the latest version of zuul master in our deployment. I don't know of any 2.1.0 issues and would expect that release to function, though the log information as described above should help us dig more.

I assume you mean the local zuul git repo cache? It is safe to delete those repos as long as there are no jobs running that need to fetch the refs from previous merge attempts. Typically this is only needed if the merger becomes slow due to having many stale refs in the repository. It is possible that the repo has somehow corrupted itself and having zuul reclone it would correct merge errors in that case.

To recap I would start by reading your debug logs to narrow down the issue, then from there you can decide what action is necessary.

On Tue, Mar 8, 2016, at 09:27 AM, Anita Kuno wrote:
> Moving this post to the openstack-infra mailing list thread for 
> discussion. The third-party-announce thread is meant to be low traffic 
> for contacting third party ci operators. Discussion of third party 
> tools and systems are to take place on the openstack-infra mailing list.
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-infra
> 
> Thank you,
> Anita.
> 
> On 03/07/2016 09:58 PM, Wan, Sam wrote:
> >  Our CI's ran into this issue too recently.
> >  I've followed your suggestion to change merge-failure to only send SMTP message.
> >  But could you tell me why this issue happened?
> > 
> >  I've upgraded my zuul to latest 2.1.0. Is this OK?
> >  Also someone told me that I might need to clean local zuul cache from time to time, is that correct?

_______________________________________________
OpenStack-Infra mailing list
OpenStack-Infra at lists.openstack.org
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-infra



More information about the OpenStack-Infra mailing list