04:02:58 <ekcs> #startmeeting congressteammeeting
04:02:59 <openstack> Meeting started Fri Apr 19 04:02:58 2019 UTC and is due to finish in 60 minutes.  The chair is ekcs. Information about MeetBot at http://wiki.debian.org/MeetBot.
04:03:00 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
04:03:02 <openstack> The meeting name has been set to 'congressteammeeting'
04:03:18 <akhil_jain> ekcs: Hi
04:03:50 <ekcs> hi akhil_jain !
04:04:14 <akhil_jain> How's everything?
04:05:13 <ekcs> not too bad. always interesting when preparing presentations. helps point out how much I don’t know about a particular application like nfv.
04:05:17 <ekcs> how are you?
04:06:02 <ekcs> topics list here as usual: https://etherpad.openstack.org/p/congress-meeting-topics
04:06:08 <akhil_jain> I am good. Yes seen your work for presentation, thats really cool
04:07:56 <ekcs> great!
04:08:01 <ekcs> well let’s get going then.
04:08:07 <ekcs> #topic lower constraints
04:08:47 <ekcs> so it turns out for some reason the lower constraints job wasn’t set up quite right. and congress unit tests fail on some of the lower constraints. https://review.openstack.org/#/c/650441/
04:09:08 <ekcs> it’s going to take some effort to figure out whats failing and probbaly to raise to lower constraints appropriately.
04:10:22 <ekcs> just an update. don’t think there’s much to talk about here.
04:10:32 <ekcs> so moving on to next topic.
04:10:52 <ekcs> #topic monasca-tacker-congress talk
04:11:37 <ekcs> so akhil_jain I know you haven’t had much of a chance to look deeply yet. but if there’s any questions or comments to discuss now we could take the opportunity
04:14:09 <akhil_jain> yes I have made one comment. actually the main scenario is tacker respawn on detecting the fault. but as of now what tacker does is launches new instances with all new IPs assigned so from user perspective that wont be much useful. so i used nova evacuate. it is something missing from tacker end. we can add this point in session as well
04:15:07 <ekcs> ok I’ll make sure to mention that as a fault management action.
04:15:10 <akhil_jain> also for demo i have started created new env. old one is not working
04:16:01 <akhil_jain> also i will test respawn api once env is ready and let you know
04:16:02 <ekcs> oh got it any idea why? just something not right any more or maybe evidence of possible congress bug?
04:16:31 <akhil_jain> no it is from tacker side only
04:16:43 <ekcs> ok cool.
04:17:25 <ekcs> by the way, what’s wrong with a new instance with a different IP? I imagine usually there are many VNF instances behind a load balancer. so respawning a new instance with new IP works ok?
04:19:42 <akhil_jain> amm, i am not sure about it but while discussing with Ghanshyam mann he stated if instance is being used anywhere and it need to be recreated with all new network and resources. it wont be fulling its old use
04:20:07 <akhil_jain> like its whole identity will be changed
04:20:51 <ekcs> ok. I guess it all depends on the networking setup. there are cases when we want to keep that specific instance.
04:22:10 <akhil_jain> yes, that is case, but if you think ip doesnt matter. we can display respawn as well from tacker side
04:23:01 <ekcs> I’ll just mention it as a possible response along with the others.
04:23:21 <ekcs> well that’s helpful. do please leave more comments if you have any later. especially helpful would be any insight into the particular requirements of the nfv eco system that i am not clear on.
04:23:30 <ekcs> moving on then if nothing else on this topic?
04:23:48 <akhil_jain> yes nothing more as of now
04:24:19 <ekcs> ok!
04:24:23 <ekcs> #topic Train cycle feature targets
04:25:55 <ekcs> thanks for putting up the topic. good time to start thinking and discussing train targets.
04:27:21 <akhil_jain> yes I want to finalize my targets for this cycle and share with my client
04:28:53 <akhil_jain> need your help to get some features to work on
04:29:32 <ekcs> ok do you already have thoughts on what’s important for the client? maybe start with collecting that info.
04:31:26 <akhil_jain> its more on self-healing side. but nothing fixed as of now. so I want to continue the same till then
04:32:36 <akhil_jain> Do you have any feature on your mind pending to be implemented in congress. I would like to work on that
04:33:18 <ekcs> I’m in the process of updating this backlog: https://etherpad.openstack.org/p/congress-task-priority
04:34:14 <ekcs> the ones that have most to do with self healing seem like these:
04:34:16 <ekcs> feedback on action executions
04:34:21 <ekcs> results
04:35:01 <ekcs> execute-repeat semantic
04:35:32 <ekcs> and other execution semantics to be determined by that disucssion depending on use case.
04:36:14 <ekcs> I think priority really depends on client use case.
04:38:16 <akhil_jain> yes, but i have option to propose usecases and related development work as well
04:39:05 <akhil_jain> i will look into the features you shared, and thanks again for help as always
04:41:18 <ekcs> ok got it. why don’t we schedule another time to talk about this specifically?
04:41:34 <akhil_jain> yes that would be great
04:42:44 <ekcs> ok cool. should we stop here for today then?
04:42:50 <akhil_jain> also there is one more that to be discussed
04:42:53 <akhil_jain> https://review.openstack.org/#/c/652294/
04:43:12 <akhil_jain> i think it is failing because of psycopg==2.5.1
04:43:37 <akhil_jain> which is not installable on bionic (i think)
04:44:16 <akhil_jain> we can discuss it on gerrit later on
04:44:34 <ekcs> hmm. that is a problem, but I don’t htink it’s the problem here.
04:44:55 <ekcs> because the psycopg==2.5.1 lower costraint should not affect the other jobs.
04:45:28 <ekcs> also we haven’t actually merged that patch on master branch. https://review.openstack.org/#/c/650095/
04:46:20 <akhil_jain> hmm, i will look into it again
04:47:41 <ekcs> seems it DOES have to do with psycopg, but not the low version.
04:47:46 <ekcs> look at the end of this file: http://logs.openstack.org/94/652294/1/check/openstack-tox-py27/aa62bb4/tox/py27-1.log
04:48:50 <ekcs> I’m guessing this patch caused the failure: https://github.com/openstack/requirements/commit/a96ee0e258aafb2880149b3e25dd5959f7b37c09
04:49:56 <akhil_jain> yes that can be the reason
04:52:09 <ekcs> surprised I don’t see a discussion on ML already or a bug filed. should be breaking others as well. maybe not enough projects depend on it so it’ll take a little longer.
04:52:23 <ekcs> guess we’ll have to figure it out / start that conversation.
04:55:36 <ekcs> same problem discussed here, but not clear what the solution is yet: https://github.com/Cloud-CV/EvalAI/issues/1480
04:57:21 <ekcs> here’s a better discussion: https://github.com/psycopg/psycopg2/issues/699
04:59:20 <ekcs> ok time to end meeting =)
04:59:32 <ekcs> have a great weekend!
04:59:33 <ekcs> #endmeeting