17:20:43 <imaximets> #startmeeting ovn_community_development_discussion
17:20:44 <openstack> Meeting started Thu Sep 17 17:20:43 2020 UTC and is due to finish in 60 minutes.  The chair is imaximets. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:20:45 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
17:20:47 <openstack> The meeting name has been set to 'ovn_community_development_discussion'
17:21:05 <imaximets> I do not have much updates.
17:21:05 <numans> Hi
17:21:07 <_lore_> hi all
17:21:12 <imaximets> Who wants to go first?
17:21:19 <numans> I can go
17:21:28 <imaximets> numans, all yours.
17:21:33 <numans> #info branch-20.09 is created
17:21:54 <numans> Yesterday I created the branch-20.09.
17:22:07 <numans> Mark is on leave, so I did it.
17:22:29 <numans> I spent much of the time on the code reviews.
17:22:46 <numans> Reviewed Anton's refactor patches and Han's VxLAN feature patch.
17:22:58 <numans> Included these 2 patch sets before creating the branch.
17:23:16 <zhouhan> numans: VXLAN patch is not mine. It is from Ihar :)
17:23:25 <numans> zhouhan, sorry.
17:23:32 <numans> I got confused :)
17:23:48 <zhouhan> np
17:23:49 <numans> Ihar has requested for a couple of patches to be included before the hard freeze.
17:24:02 <numans> I found VxLAN to be fine.
17:24:21 <numans> Apart from that, I also reviewed zhouhan's crash fix patches.
17:24:30 <zhouhan> numans: I reviewed Ihar's "multiple controllers" patch and posted some commments.
17:24:35 <numans> And I submitted a small bug fix patch.
17:24:50 <numans> zhouhan, I saw the email. But didn't look into the comments.
17:24:55 <numans> zhouhan, thanks for it.
17:24:59 <zhouhan> numans: He requested that to be included in 20.09 as well. How would we handle it?
17:25:09 <numans> I think its better not to include it.
17:25:20 <numans> It seems a bit risky.
17:25:23 <zhouhan> numans: agree.
17:25:49 <numans> With Vxlan, I found the risk to be less and the code seemed fine to me.
17:26:06 <numans> and I had reviewed earlier patch versions too.
17:26:13 <numans> Ok. that's it from me.
17:26:38 <zhouhan> thanks numans. May I go next
17:26:39 <dceara> numans: re branch-20.09, I'd like to try to have this patch merged there too if possible:
17:26:41 <dceara> #link http://patchwork.ozlabs.org/project/ovn/patch/20200917125050.19729.41529.stgit@dceara.remote.csb/
17:26:54 <numans> zhouhan, sure.
17:27:09 <dceara> zhouhan: sorry, I interrupted, please go ahead
17:27:09 <numans> dceara, I couldn't look into it. I'll take a look tomorrow.
17:27:35 <zhouhan> I was mainly working on the bug reported by dceara. Big thanks to dceara for finding a simple way to reproduce it.
17:27:38 <numans> and if zhouhan can look into that patch, it would be great.
17:28:04 <zhouhan> numans: I will take a look
17:28:13 <numans> zhouhan, thanks.
17:28:27 <numans> thanks to dceara.
17:29:04 <zhouhan> Now the bug fix is merged. I am going to merge to 20.09 and 20.06, but leave 20.03 as is. Thanks for the reviews from dceara and numans.
17:30:23 <dceara> zhouhan: np
17:30:32 <zhouhan> For Ihar's "multiple controller" patch review, my major concern is when it checks if there is multiple controllers configured, it skips the tunnel and patch port resource cleanup, which I think is a problem.
17:30:44 <zhouhan> We will discuss more in ML.
17:30:51 <zhouhan> That's it from me.
17:31:11 <imaximets> Thanks.  Who wants to go next?
17:32:37 <imaximets> Anyone?
17:33:02 <dceara> I have a quick comment if that's ok.
17:33:10 <imaximets> dceara, sure.
17:34:36 <dceara> Mark started some discussions regarding CI a while ago on the mailing list but there wasn't much conclusion. I was wondering if we should revive that effort and include upstream ovn-k8s/openstack/etc CI to use latest OVN master code. Given the increasing number of regressions we saw recently.
17:35:09 <zhouhan> +1
17:35:12 <numans> I agree to that.
17:35:20 <dceara> It's probably better to discuss this on the mailing list but I thought it might be good to mention it here first :)
17:35:54 <numans> as part of this discussion, I think it would be better if we move to github actions instead of travis ci.
17:36:13 <numans> May be we can discuss more in ML, but I'd like to know if there any quick comments here.
17:36:36 <zhouhan> numans: I am not familiar with git actions. What's the motivation?
17:36:38 <numans> ovn-k8s is already using github actions and we can probably use most of the actions to run ovn-k8s tests
17:37:13 <numans> zhouhan, As an example you can take a look - https://github.com/ovn-org/ovn-kubernetes/blob/master/.github/workflows/test.yml
17:37:30 <zhouhan> numans: ok, thanks
17:37:32 <dceara> numans: that is a good point, I looked a bit at how ovn-k8s build their image for CI testing and I think we can change it quite easily to build latest OVN from source.
17:37:37 <numans> zhouhan, although we don't use PR, I think once the patch is merged, the tests defined in the actions would be executed.
17:37:54 <numans> dceara, they already have a docker file to build ovs/ovn from sources.
17:38:09 <numans> we can probably use that to build the ovn-k8s image
17:38:16 <zhouhan> numans: where would the test run?
17:38:27 <dceara> numans: hmm, I thought they were only consuming fedora rpms, I'll have a better look :)
17:38:40 <numans> zhouhan, I think github runs it.
17:38:50 <zhouhan> Oh, I see
17:39:01 <numans> zhouhan, for public repos, I think it lets you run certain amount of time for free.
17:39:14 <numans> I don't know the details. but onv-k8s uses it,
17:39:24 <zhouhan> numans: thanks to Microsoft
17:39:50 <numans> zhouhan, https://github.com/ovn-org/ovn-kubernetes/pull/1705#partial-pull-merging
17:39:53 <numans> one such example here
17:40:13 <numans> I'll try it out on my personal repo and see.
17:40:30 <numans> and if it is fine, I can propose it to ovn-org/ovn
17:40:56 <numans> dceara, sorry. go ahead
17:41:07 <numans> one more link
17:41:07 <imaximets> numans, zhouhan: I think that most of things that could be done in github workflows could also be done in travis.  What's the point of using github workflows?  Does it have some features that travis has not?
17:41:09 <numans> #link https://github.com/ovn-org/ovn-kubernetes/pull/1705/checks?check_run_id=1121866343
17:41:58 <dceara> numans: right, but those runs are with ovn-k8s using fedora rpms.
17:42:27 <numans> imaximets, not sure about the features, but if we can use much of ovn-k8s work flow code, it would be easier.
17:42:53 <numans> dceara, https://github.com/ovn-org/ovn-kubernetes/blob/master/dist/images/Dockerfile.fedora.dev
17:43:10 <dceara> numans: great, thanks!
17:43:31 <numans> I guess we can discuss that in the ML. I'll just try it out first on my personal ovn repo.
17:44:22 <dceara> In any case, I just wanted to bring this up here, we can probably continue on the mailing list to figure out details like travis vs gh workflows
17:44:42 <dceara> That's it on my side, thanks.
17:45:45 <imaximets> OK. Thanks. Any more updates?
17:46:43 <imaximets> So, I think, we could call it a meeting.
17:46:56 <imaximets> Thanks everyone!
17:47:02 <numans> Thanks everyone!
17:47:10 <imaximets> #endmeeting