14:03:22 <lujinluo> #startmeeting neutron_upgrades
14:03:23 <openstack> Meeting started Thu Apr  5 14:03:22 2018 UTC and is due to finish in 60 minutes.  The chair is lujinluo. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:03:24 <TuanVu> Hi Luo
14:03:25 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:03:27 <TuanVu> Hi Ihar
14:03:28 <openstack> The meeting name has been set to 'neutron_upgrades'
14:03:35 <lujinluo> hi ihrachys and TuanVu
14:04:07 <lujinluo> we do not have any action items from last week. we are basically still waiting for https://review.openstack.org/#/c/556667/ first
14:04:47 <ihrachys> yeah this patch gets out of hand. takes so long (though arguably because of me being on vacation)
14:05:42 <ihrachys> I am working with zzzeek on it. I think we are quite close but there are still some issues to solve that keep CI in red.
14:06:13 <lujinluo> ihrachys: yeah, i tested it today locally. i saw network and rbac related tests failing
14:06:36 <ihrachys> one complexity being that zzzeek prefers code isolated from neutron to reproduce any issues we may have, so every time I hit a unit test failure, I need to "port" it to neutron-less script
14:07:00 <ihrachys> and I don't really grasp internals of sqlalchemy to go on my own, so I need to accommodate
14:07:08 * mlavalle slides in silently to the back of the room hoping nobody will notice he is late
14:07:30 <ihrachys> at one point zzzeek suggested the result of this work should go as a new feature for sqlalchemy
14:07:46 <ihrachys> because he never got it that far trying to handle all the different cases
14:07:47 <lujinluo> i totally understand. educate myself a lot to understand what the patch means..
14:08:00 <TuanVu> @ mlavalle: haha :))
14:08:04 <lujinluo> nice to hear that haha
14:08:37 <ihrachys> that's all I have for now. I will continue working on it today.
14:08:45 <lujinluo> ok, thanks
14:08:57 <TuanVu> thank you very much, Ihar :)
14:09:07 <lujinluo> #topic OVO
14:09:23 <lujinluo> I do not think we have any updates on those OVO patches though
14:09:25 <lujinluo> https://review.openstack.org/#/q/status:open+project:openstack/neutron+branch:master+topic:bp/adopt-oslo-versioned-objects-for-db
14:09:42 <lujinluo> TuanVu: i saw you pushed network patch. any updates?
14:10:04 <lujinluo> https://review.openstack.org/#/c/507772/
14:10:29 <lujinluo> ah, my bad. it seems to be CI
14:10:30 <TuanVu> I've already solved all the concerns from Ihar
14:11:15 <ihrachys> the failure suggests a legit issue though no? http://logs.openstack.org/72/507772/41/check/openstack-tox-py27/b7bc0e1/testr_results.html.gz
14:11:20 <TuanVu> that patch is waiting for Ihar's patch to pass all the tests
14:12:21 <TuanVu> Ihar's patch will help to solve remaining failed tests
14:12:22 <ihrachys> I probably just forgot something, but why do we think it's related to my patch?
14:12:53 <lujinluo> yeah, i do not think the constant query failures can be fixed by Ihar's patch either
14:13:05 <TuanVu> https://review.openstack.org/#/c/507772/41/neutron/objects/base.py
14:13:07 <TuanVu> line 320
14:13:20 <ihrachys> right. there was the InvalidRequestError that I think should be solved by my patch but not queries number
14:13:48 <ihrachys> TuanVu, right. is the number of queries related to the error?
14:14:46 <TuanVu> also this one
14:14:47 <TuanVu> https://review.openstack.org/#/c/507772/41/neutron/objects/network.py
14:14:58 <TuanVu> line 291
14:15:19 <TuanVu> after Ihar's patch, I think that it's not needed to reload share
14:16:18 <ihrachys> yeah this will probably be affected. I may need to worry about the test case myself though, not sure if it will pass just fine with the automatic fk invalidation.
14:16:51 <ihrachys> what I learned from playing with sqlalchemy lately is that its behavior is at times irregular (depends on caching / timing of models in session)
14:17:19 * mlavalle agrees with Ihar
14:17:29 <ihrachys> so it may e.g. issue an event for one action ordering and then miss the same event in another simingly similar case
14:17:34 <lujinluo> ++
14:18:02 <ihrachys> so that's concern if we rely on exact number of queries in a test case. I will need to dig more to understand the impact there better.
14:18:36 <ihrachys> so, tl;dr agreed we may want to wait to see how my patch plays out
14:18:59 <lujinluo> yeah! will keep a close eye on it!
14:19:36 <lujinluo> although I do not think we have any more patches to discuss, I think TuanVu has something to share, right? about the remaining work
14:19:49 <TuanVu> yes
14:20:29 <TuanVu> based on etherpad link from Ihar, I've just made "the Summary for Rolling Upgrade progress of Neutron"
14:20:40 <TuanVu> https://docs.google.com/spreadsheets/d/1Mz7V40GSdcUH_aBoNWjsFaNRp4wf-duz1GFWligiNXc/edit?usp=sharing
14:21:22 <TuanVu> hopefully it'll help us to follow remaining work easier
14:21:44 <ihrachys> nice to see it in this format
14:21:52 <ihrachys> btw one suggestion about making progress
14:21:57 <TuanVu> btw, since the amount of remaining work is huge, it'll be great if we could have more engineer working on it
14:22:48 <ihrachys> I understand that I am blocking most of the work. but in the meantime, what could be solved in parallel is unit tests to port to objects. I think we should be able to safely port tests without consideration we have for production code.
14:23:28 <TuanVu> yeah, great idea, Ihar
14:23:40 <lujinluo> agree. while waiting for your patch to land, the rest of us would better to pick those tests
14:24:28 <ihrachys> great I would appreciate if you add me to those patches to review when they are ready.
14:24:53 <ihrachys> and I think I should make progress on the blocking patch till next meeting (hopefully earlier)
14:25:06 <lujinluo> will do.
14:25:09 <lujinluo> thanks!
14:25:22 <hungpv_> ++
14:25:59 <lujinluo> #topic open discussion
14:26:11 <lujinluo> does anyone has anything to share?
14:26:25 <ihrachys> I don't
14:26:27 <mlavalle> I want to see if it would be possible to get a favor from this team
14:26:52 <lujinluo> mlavalle: what favor if i may ask?
14:26:54 <ihrachys> except that lujinluo is don't a great job chairing the meeting
14:27:02 <ihrachys> *doing!
14:27:13 <lujinluo> ihrachys: thank you!!
14:27:39 <mlavalle> Do you remember that patch that we reverted at the end of the last cycle? The one that added OVO to the ML2 plugin?
14:27:53 <lujinluo> yes, portbinding one
14:28:37 <mlavalle> I want to ask if it is possible for that work to be done (when it is done) on top of this patch: https://review.openstack.org/#/c/414251
14:28:47 <mlavalle> ?
14:29:14 <mlavalle> I have made enough progress with this patch that now I can migrate VIFs from node to node
14:29:27 <mlavalle> and the Nova team is testing their instance migration on top of it
14:29:35 <ihrachys> I think it makes sense, yes. multiple port binding work should not be blocked by internal reworkings.
14:30:18 <lujinluo> i see. i recently had the portbinding patch respinned here https://review.openstack.org/#/c/544206/. i will rebase it upon that one later
14:31:12 <mlavalle> thanks guys!
14:32:09 <lujinluo> ok, if no one else has anything, we can have extra 28 mins back!
14:32:19 <lujinluo> thanks everyone
14:32:37 <TuanVu> thank you Luo, Ihar and mlavalle
14:32:43 <mlavalle> o/
14:32:51 <hungpv_> Thanks you all
14:32:52 <mlavalle> have a good night, most of you!
14:33:06 <hungpv_> I'm having a dinner now, lol
14:33:07 <lujinluo> #endsmeeting
14:33:19 <hungpv_> See you guys
14:33:22 <lujinluo> oops wrong command
14:33:25 <lujinluo> #endmeeting