14:00:46 <ricolin> #startmeeting heat 14:00:47 <openstack> Meeting started Wed Oct 17 14:00:46 2018 UTC and is due to finish in 60 minutes. The chair is ricolin. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:00:48 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:00:50 <openstack> The meeting name has been set to 'heat' 14:00:52 <ricolin> #topic roll call 14:00:55 <zaneb> o/ 14:00:57 <ramishra> hi 14:01:03 <ricolin> o/ 14:01:43 <ricolin> Anyone from Heat-dashboard around? 14:02:03 <ricolin> #topic adding items to agenda 14:02:09 <ricolin> #link https://wiki.openstack.org/wiki/Meetings/HeatAgenda#Agenda_.282018-10-17_1400_UTC.29 14:02:12 <ramishra> it seems gate is completely borked http://logs.openstack.org/95/611295/2/check/openstack-tox-pep8/3495343/job-output.txt.gz#_2018-10-17_13_27_51_465665 14:02:34 <ricolin> #topic amqp1 job - it broke 14:05:44 <ramishra> ricolin: I saw it green 2 days back, what's broken in that job? though everything is broken now, some issue with the images I suspect 14:06:23 <zaneb> oh? I haven't seen the amqp1 job green in a while 14:07:20 <ricolin> ramishra, don't know, finding... 14:07:38 <ramishra> http://logs.openstack.org/62/609962/2/check/heat-functional-convg-mysql-lbaasv2-amqp1/65e5aff/job-output.txt.gz 14:08:03 <ramishra> zaneb: that's 2 days back I guess 14:08:24 <zaneb> hmm, ok, maybe I just had bad luck then 14:08:36 <zaneb> (I added this to the agenda) 14:09:08 <zaneb> ramishra: are you sure the whole gate is broken and not just https://review.openstack.org/611295 ? 14:10:20 <ramishra> zaneb: from the errors it seems like it's all broken http://logs.openstack.org/95/611295/2/check/heat-functional-convg-mysql-lbaasv2/2cddd65/logs/devstacklog.txt.gz#_2018-10-17_13_46_20_516 14:10:20 <ricolin> zaneb, I have put some recheck to see if them works 14:10:54 <zaneb> ah, yep, that looks bad 14:11:58 <zaneb> bad package in the mirror? I imagine it affects everyone and will get fixed pretty quick 14:12:18 <ramishra> yeah, so I did not spend time on it:) 14:13:07 <zaneb> let's move on then :) 14:13:47 <ricolin> sure 14:13:48 <ricolin> #topic Validation improvements 14:14:31 <zaneb> #link https://storyboard.openstack.org/#!/story/2003915 14:14:32 <ricolin> which validate you mean here 14:14:52 <ricolin> oh, okay 14:14:57 <zaneb> so I started working on the placeholder stuff that we've been talking about for years 14:15:11 <zaneb> most of it is not ready for review yet 14:15:29 <zaneb> but I just wanted to let y'all know that it's happening 14:15:51 <zaneb> so that if anyone thinks it's a bad idea/isn't going to get reviewed/whatever can speak up 14:16:42 <zaneb> I also think that splitting the validation into synchronous and asynchronous phases might also happen as part of this 14:17:17 <zaneb> there's some tricks to work out to do with when we validate which parts of nested stacks 14:17:26 <therve> zaneb: How much breakage will happen? :) 14:17:42 <zaneb> therve: that's the question, isn't it :) 14:17:54 <therve> Kinda 14:17:58 <zaneb> the goal is to only break earlier, not more 14:18:28 <zaneb> so we'll catch stuff that would have broken when you went to create the resource, before you start creating the stack 14:19:14 <zaneb> goals and reality are not always the same thing, however 14:19:20 <therve> Wouldn't that go into making more synchronous stuff? 14:19:27 <therve> Validation is somewhat slow already 14:20:03 <zaneb> it's mostly fast stuff 14:20:46 <zaneb> e.g. if you do get_resource on OS::Nova::Keypair and pass it as the image to a server, then it'll see that the custom constraints don't match up and fail 14:20:59 <zaneb> without needing to make any external calls 14:21:28 <zaneb> it's the external calls that really slow down validation. I'm hoping to be able to move _all_ of those to the async phase 14:23:38 <zaneb> in fact ideally if we see that the custom constraints of outputs and inputs _do_ match, we should be able to avoid an external call altogether. I have no idea how to implement that though :) 14:23:58 <therve> OK 14:24:30 <zaneb> I was thinking about writing a spec for this, but I really need to experiment more to even know what's possible 14:25:13 <zaneb> the way things are validated now is extremely poorly documented 14:26:40 <ricolin> zaneb, this definitely need some doc after other tasks 14:28:52 <ricolin> I try to find some good way to improve constraint validate while ago, but I guess we can only do one check at a time:/ 14:30:15 <zaneb> I'm trying to do a bit of the renaming and separating phases we talked about in https://etherpad.openstack.org/p/heat-ocata-validation-improvements 14:30:29 <zaneb> hopefully that will make it easier to document 14:34:08 <zaneb> that's about all I had to say about that for now unless anybody has more questions 14:34:29 <ricolin> zaneb, are you plan to this in S release? 14:35:13 <zaneb> I'm gonna say yes :) 14:35:20 <ricolin> zaneb, :) 14:35:34 <zaneb> it's still early in the cycle, so in theory we have time 14:35:55 <ricolin> Okay, let's keep trace on that 14:36:03 <zaneb> there's nothing like the mox stuff in the pipeline to suck up all of our review bandwidth 14:36:14 <ramishra> may be we do some measurements with a simple poc or something before we decide to go for it this cycle, I'm not sure we've enough time though 14:36:14 <zaneb> that was painful 14:36:16 <ricolin> zaneb, that's for sure 14:37:01 <zaneb> would anybody be interested in working on adding more validation tests? 14:37:19 <ramishra> of course after deciding what to do:) 14:37:30 <zaneb> that would be a useful thing to give us more confidence in any changes 14:38:48 <ricolin> indeed 14:38:54 <zaneb> it's something that could be done in parallel - we can add a bunch of functional tests now and they'll work as regression tests later 14:40:46 <ricolin> I think we can add that to tasks in story 14:41:59 <ricolin> okay, let's move on 14:42:00 <ricolin> #topic Autoscaling integration 14:42:00 <zaneb> might be worth creating a new story, that one is an unreadable mess already 14:42:07 <ricolin> zaneb, +1 14:43:27 <ricolin> The forum for autoscaling improve, is approved I guess 14:43:56 <ricolin> So we're going to discuss in Berlin about #link https://etherpad.openstack.org/p/autoscaling-integration-and-feedback 14:44:39 <ricolin> about what's the benefit, and how can we do it right 14:45:26 <ricolin> there are already some discussion in ML 14:45:53 <ricolin> now I'm trying to put more information together 14:46:46 <ramishra> ricolin: I think I asked about current user issues/bugs in the ML thread and did not hear anything, unless I missed something, other than the fact we feel it's not maintainable atm 14:48:57 <ricolin> ramishra, what I got from users are not compliant for bugs, but more like confuse about the current situation of Heat and Senlin 14:49:39 <ramishra> OK, IMO that's not something I would worry with the review bandwidth we have. If there is bandwidth to improve stuff we should do it though, reminds me of my improvement patch that took 2 years to land 14:51:06 <zaneb> one of the reasons I started working on the validation stuff is that it seemed like an easier win for Stein 14:51:43 <zaneb> nobody seemed to be really excited about autoscaling changes, and there's not much point getting started until after the summit 14:52:03 <zaneb> but it would be really nice to come out of that session with a long-term plan 14:52:29 <ramishra> zaneb: yeah, a very long term plan:) 14:52:33 <ricolin> yeah, IMO we have a lot of duplicate works cross projects, so try to gathering resources might be a good thing to discuss as a long term plan 14:52:37 <zaneb> lol 14:53:55 <ricolin> ramishra, don't worry, there will be actions for Stein:) 14:55:13 <ricolin> btw, we also going to have heat onboarding, heat project update, and heat user/ops feedback in Summit as well 14:55:49 <ricolin> so please give your idea if you would like to see how those session get done 14:57:18 <ricolin> move on 14:57:33 <ricolin> #topic Multi cloud 14:57:42 <ricolin> #link https://review.openstack.org/#/q/status:open+project:openstack/heat+branch:master+topic:bp/multiple-cloud-support 14:59:34 <ricolin> zaneb, do you think we can use tempfile? or move back to use previous for temp file patch 15:00:43 <zaneb> ricolin: did the FIFO thing not work? 15:00:54 <ricolin> zaneb, I think it works as well 15:01:01 <ramishra> got to go, meeting time over;) 15:01:10 <zaneb> ramishra: thanks o/ 15:01:11 <ricolin> ramishra, thanks for join 15:01:49 <zaneb> ricolin: the previous patch is way more complicated than it needs to be, because we now only need to do one file instead of 3 15:02:10 <zaneb> since we decided not to support custom client certs 15:02:49 <ricolin> zaneb, yeah, so we don't need a folder 15:04:04 <ricolin> So the problem now is I don't know how I can properly test CA cert 15:04:37 <ricolin> like have functional/scenario test for it 15:04:37 <zaneb> I liked the FIFO because it can only be read once. but since the CA cert is not secret, it probably doesn't matter 15:05:18 <zaneb> ricolin: yeah, that is a problem. I don't know how to test it either 15:06:19 <ricolin> zaneb, so my plan now is to asking public cloud WG's help once we think we can put these patch under test 15:06:47 <ricolin> they got the actual environment to test 15:07:01 <ricolin> as for writing as functional test, well 15:07:01 <zaneb> ricolin: at the very least you can pass the regular ca_cert for the devstack cloud as a 'custom' ca_cert, and that should exercise the code 15:07:24 <zaneb> no way to verify that it is doing anything at all, but we can test that manually and hope for the best 15:08:01 <ricolin> zaneb, yep 15:08:33 <ricolin> so please review the rest of patches 15:09:04 <ricolin> I will update that patch 15:09:06 <zaneb> yeah, I need to do that 15:09:06 <ricolin> later 15:09:17 <ricolin> zaneb, great, thanks! 15:09:24 <zaneb> ricolin: how confident are you that they're more or less ready to go? 15:10:57 <zaneb> I haven't checked the latest spec yet either I don't think 15:10:57 <ricolin> I already give tempest test for them, so I think it's ready 15:11:05 <zaneb> ok, cool 15:11:44 * zaneb is super busy with TC stuff ahead of the summit :/ 15:11:49 <ricolin> what I still not done yet is to document down 15:12:38 <ricolin> zaneb, don't worry, you will keep seeing these topic on meeting agenda until it's landed 15:12:42 <ricolin> :) 15:12:48 <zaneb> :D 15:13:09 <ricolin> let's put a end meeting since we already over 12 min! 15:13:10 <ricolin> #endmeeting