'change-queue' is now in full production for the 'master' flow

[Barak Korren]

Yesterday we've applied a patch with JJB that made the 'master' change
queue push updates into 'tested' instead of the 'experimental' flow.
This essentially means that the change queue is now the main
production flow for 'master'.

A quick refresher on how to find and handle issues with CQ
jobs:

1. You don't really need to monitor the jobs, when the CQ
   suspects it  found a real code issue it will send
   infra at ovirt.org an email with the suspected patch.
2. The *-tester job is what actually runs the tests. Don't
   worry if it fails occasionally, the CQ will run bisection
   and tell you exactly which change caused it.
3. The output of the *-tester job is very similar to the old
   'experimental' job.
4. When you get an email, you need to  check the output of the
   job linked to it to see if this is a race-condition kind of
   failure, a build failure or a real regression. Unless its a
   build failure, you probably need to report it and document
   in the table.
5. If any other CQ job besides the *-tester job fails, this is
   a real infra issue. When you see such an issue you need to
   investigate it and fix the infra. Real infra issues typically
   affect many jobs, so once you are done fixing it, you
   probably need to go back and re-trigger Gerrit merge events
   to get packages rebuilt and submitted to the queue.
6. One very simple kind of an infra issue, is trying to submit
   to a queue that does not exist. You should nag maintainers
   to remove jobs for versions of oVirt that we don't have queues
   for.

Good luck!

-- 
Barak Korren
RHV DevOps team , RHCE, RHCi
Red Hat EMEA
redhat.com | TRIED. TESTED. TRUSTED. | redhat.com/trusted