= Attendees =
Lars Kurth
Anthony Perard
Joao Martin
Jim Fehlig (vacation)
Bob Ball
= Agenda =
== 1 year review of WG / CI loop running ==
It's been a year since the CI loop went live: I want to report back to the AB on
* What went well
BB: Great that it has been voting for a year. There have been a few changes when there were correct negative votes.
BB: Integration generally good
LK: Were fast of ramping this up
LK: WG, when things were stuck, we were all helping each other and get things resolved
JM: Been here less than 50% (we were in group B already)
JM: OpenStack support is fairly good
JM: CI loop maintenance has been handled well (in future, would
AP: Have been able to install everything - we have clear instructions on what to do
AP: Very few issues, which are often closed quite quickly
AP: Tempest is not yet run as part of OSSTEST (patches have been posted)
ACTION: Lars to chase up Ian Jackson (should we try and get this done before the release => may throw up some issues)
JM: Things look well, but I don't know whether we have any issues on the Hypervisor side
* What didn't go well / where we can improve
JM: Didn't see anything specifically negative
JM: Maybe being more responsive to issues that were raised
JM: Maybe we need more coordination, and have more people stepping up
BB: The biggest issue is ongoing maintenance of libvirt and Xen integration
BB: There are a number of failures which we know exist and some patches have been becoming stale
BB: Being faster in turning around integration issues would help. At the moment we don't have a good answer for a solution
LK: Why do you think we don't have a good answer
BB: As far as I am aware, there has not been much active engagement on open issues (e.g. on various patches)
BB: They have not been touched for 3 months
BB: We had some failures (5-10%) that have been there since the CI system was brought up
BB: Don't have fixes for them
BB: We are not confident that proposed fixes may be fixed, but we don't know how many would be fixed
BB: Have not seen external evidence of the momentum we had at the beginning being maintained
JM: Do you have the failure info at hand
ACTION: when we set up new VM, the URL became stale due to DNS issues (104.130.29.226 and
http://104.130.29.226:8080/)
BB: Don't know how many were real failures and how many were false negatives
BB: Openstack 3rdpartysystems
ACTION: Bob to send some links and instructions
LK: Openstack workflow is different from what we are used within the Xen Project
BB: My interpretation is that there is not good integration between OpenStack and Xen Project community and we don't have much overlap (e.g. on the channels)
BB: There was a conversation a couple of weeks ago, that there may be too many false negatives
BB: Seems that the engagement of the communities isn't what it would need to be to identify issues when they come up
JM: Was not aware of any failures - would like to understand how to work with OpenStack better
LK: Bob, would you be able to attend part of the Hackathon (see
http://wiki.xenproject.org/wiki/Hackathon/April2016)
BB: Possibly
ACTION: Action on Bob to get back to the list and Lars
---
Current CI pass / failure rates: Jenkins results – can get history of builds from here: http://104.130.29.226:8080 Ci watch can show you the Jenkins passes and scroll across to the XenProject CI, you can then easily
see tests which the XenProject CI failed where Jenkins passed. http://ci-watch.tintri.com/project?project=nova&time=7+days
---
* Where do we need more engagement from outside the existing stake-holders (Citrix, Suse, Rackspace, Oracle)
LK: Maybe need to better coordinate and be more effective within the existing group
JM: I agree that unless we have more people without affecting the workflow
BB: Not sure whom to bring in?
LK: Possibly Intel, ARM eco-system vendors, Huawei
== Stuck Reviews/Known bugs ==
We seem to have a few stuck reviews, e.g.
https://review.openstack.org/#/c/201257/5
https://review.openstack.org/#/c/199093/4
How important are these. Should they be canned, etc.
AP: These are the same (one is depending on the other)
LK: We believe that these are responsible for intermittent failures
JM: Can you explain the issues
AP: Xen and Openstack are updating a common set of IP tables, which can cause a race condition
AP: Can't change upstream, so this is a work-around
AP: These are not currently tested by the CI loop
LK: Is this something were you need help
AP: This needs to be reviewed and need to chase on IRC channel
LK: There seems to be some controversy around these, which is part of the issue
ACTION: Joao will reach out to Michael to see whether he can help push these ones along
JM: Not sure whether Oracle have any voting people who can help shepherd changes through
== Xen Project CI/Xen Project Openstack Leadership ==
Defer this because Bob could not make (or brainstorm at the Hackathon)
ACTION: Bob happy to do a little bit of thinking and get back to the group
I think we are missing end-to-end ownership/leadership rather than “work item” level activities like fixing bugs or maintaining infrastructure. Stefano was sort of doing some of it at the beginning, but never quite
fully. I have been trying to fill the gap, but am not closely enough engaged in engineering for me to be able to do this well in the long run. With Stefano gone now, we should probably look at this topic.
Perhaps we – the project - need a sort of “Openstack maintainer” or “Openstack coordinator” in a similar manner to how we have a release coordinator/maintainer for point releases to make sure all the individual
things (like bug fixes and CI loop breakages) come together to make a release successful.
* If so, what would be involved in such a community role (activities, time requirement)? It sort of sounds similar to what Jan does for maintenance releases of Xen.
* Who would be willing (vendor or individual) to step up or be suitable?
* What can we use Credativ for (they could work under the supervision of the “Openstack maintainer/coordinator”) and what cannot be done via Credativ?
I think this is related to, but distinct from, the WG leadership. If we found someone who is willing to step up, I would be willing to hand over WG leadership to that person also. Am also willing to mentor.
