Open Stack

On Mon, Apr 03, 2017 at 01:08:26PM +0200, Attila Darazs wrote:
> If the topics below interest you and you want to contribute to the
> discussion, feel free to join the next meeting:
> 
> Time: Thursdays, 14:30-15:30 UTC
> Place: https://bluejeans.com/4113567798/
> 
> Full minutes: https://etherpad.openstack.org/p/tripleo-ci-squad-meeting
> 
> We had a meeting full of intense discussion last Thursday. Here's the
> summary.
> 
> = Promotion jobs and HTTP caching =
> 
> The first part of it was centered around trying to improve and mostly speed
> up the promotion process for TripleO, which is an ongoing discussion for the
> last few weeks.
> 
> Image building takes a long time (~30 minutes) for each promotion job, which
> we can be spared by having a separate job build the images. This would
> result in fewer job timeouts. Zuul v3 will be able to handle these kind of
> job dependencies directly, but meanwhile we can probably work around it. Our
> contact on this work is pabelanger.
> 
> A lot of other outside queries can be probably sped up by having an
> infra-wide caching proxy. This might be an Apache server with mod_proxy for
> the short term, and AFS mirror in the long term. This will speed up image
> downloads and docker registry downloads as well, speeding up our jobs.
> 
As an update, we did roll out the proxy late last week and so far we setup a
reverse proxy cache for https://trunk.rdoproject.org [1]. Currently we just
updated puppet jobs to use it, and so far it seems to be working. We are going
to make an announcement to the ML later day for other projects to start using
it.

We have also started mirroring RDO into AFS, which is our long term solution.
This should be ready sometime this week.

As for docker things, clarkb and I are planning to test the reverse proxy with
hub.docker.com.  We should know more in a few days, and if the results are good
we'll likely setup a reverse proxy too.

Currently, our apache2 configuration is caching files under 100MB up to 2GB of
data. Because we have finite resources on our mirrors, this is the defaults for
today.

[1] http://mirror.regionone.infracloud-vanilla.openstack.org:8080/rdo/centos7/

> = Quickstart transition update =
> 
> The big OVB change from last week got merged, now we're checking the
> stability of those jobs before proceeding with the transition. We'll want to
> have more extensive testing before we move the voting jobs over, so probably
> we'll create parallel non-voting jobs this time (ha/non-ha/updates + gate
> job), not just testing through pending tripleo-ci changes.
> 
> 
> 
> We will probably combine the former ha and nonha OVB jobs to save resources
> on rh1. Relevant change and discussion here[1].
> 
> We also briefly talked and discussed how to involve and bring up to speed
> more people for reviewing Quickstart changes. There will be a deep dive
> session on the subject given by one of the current cores probably.
> 
> Best regards,
> Attila
> 
> [1] https://review.openstack.org/449785
> 
> __________________________________________________________________________
> OpenStack Development Mailing List (not for usage questions)
> Unsubscribe: OpenStack-dev-request at lists.openstack.org?subject:unsubscribe
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev