16:01:33 <rakhmerov> #startmeeting Mistral 16:01:34 <openstack> Meeting started Mon Jun 20 16:01:33 2016 UTC and is due to finish in 60 minutes. The chair is rakhmerov. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:01:35 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:01:37 <openstack> The meeting name has been set to 'mistral' 16:01:40 <rakhmerov> hi all 16:01:42 <d0ugal> Hello! 16:01:46 <ddeja> rakhmerov: o/ 16:01:48 <jpeeler> hi 16:02:07 <rakhmerov> hi 16:02:13 <mgershen> hi 16:02:22 <rakhmerov> d0ugal: just saw your conversation about meeting time 16:02:27 <d0ugal> :) 16:02:36 <rakhmerov> :) 16:02:51 <rakhmerov> there was an idea to have 2 meetings, bi-weekly 16:03:13 <mgershen> 2 meetings is a lot 16:03:14 <rakhmerov> but honestly I didn't get a lot of feedback on convenient time for different locations 16:03:43 <rakhmerov> no, I mean one week - for Asia/Europe and another week for North America 16:04:03 <rakhmerov> dunno, maybe we will make it 16:04:04 <mgershen> sorry 16:04:08 <rakhmerov> np 16:04:16 <d0ugal> This is fine, I might just need to leave a bit early today. 16:04:40 <rakhmerov> as we now have more people I would actually ask you to vote for convenient time slots 16:04:41 <rakhmerov> soon 16:04:43 <rakhmerov> d0ugal: np 16:04:49 <rakhmerov> ok, let's start then 16:05:21 <d0ugal> rakhmerov: k, maybe we can do a vote on the mailing list 16:05:25 <rakhmerov> first of all, I have to apologize again for not sending you a notice last time that we wouldn't have a meeting 16:06:04 <rakhmerov> d0ugal: yes, there's already a thread but it's kind of obsolete now, I'll need to check if the slots that I suggested ~ 2 months ago are still available 16:06:10 <rakhmerov> I'll check that 16:06:39 <rakhmerov> #action rakhmerov: check time slots for bi-weekly meetings again 16:06:44 <d0ugal> Thanks 16:07:28 <rakhmerov> #topic Current status (progress, issues, roadblocks, further plans) 16:08:37 <rakhmerov> my status: I spent last week mostly on reviews, IRC and doing some internal stuff, also researched a lot of Keystone stuff including Federation API and auth plugins 16:09:08 <rakhmerov> also finished Custom Actions API spec: https://review.openstack.org/#/c/325769/ 16:09:12 <rakhmerov> so please review 16:10:01 <rakhmerov> as far as that spec, I don't think it's possible to fill it with more details at this stage so I guess particular method signatures etc. is out of its scope 16:10:01 <ddeja> my status: Write unit tests for kombu driver for alternative RPC API and fix some bugs while doing so - they are ready for review I guess + done some reviews (not a lot of unfortunately) 16:10:25 <rakhmerov> ddeja: finally, I'm gonna have time tomorrow to review these patches 16:10:58 <ddeja> rakhmerov: thanks 16:11:23 <mgershen> I have some bugs and blueprints on me, but I haven't started them (working on internal stuff). If anyone needs them and want to work on them now, just let me know. 16:11:40 <ddeja> I'd like to apologise that it took me this long... 16:12:04 <rakhmerov> ddeja: that's totally fine, the task is pretty hairy IMO 16:12:25 <rakhmerov> mgershen: do you expect to start with them any time soon? 16:12:52 <rakhmerov> if not too soon we may want to give part of your tasks to someone else 16:13:42 <rakhmerov> please let me know once you know 16:14:05 <mgershen> rakhmerov: it might take 2 weeks 16:14:15 <rakhmerov> ooh, ok 16:14:34 <rakhmerov> mgershen: let's talk offline tomorrow about that 16:14:41 <rakhmerov> jpeeler, d0ugal: any updates from your side guys? 16:14:48 <mgershen> rakhmerov: sure 16:14:56 <rakhmerov> d0ugal: I saw your patch today, left my comments 16:15:09 <d0ugal> I've mostly been working on the TripleO side of things, but I did manage to start on https://bugs.launchpad.net/mistral/+bug/1568541 16:15:09 <openstack> Launchpad bug 1568541 in Mistral "action-update API on non existing action - creates action instead of returning 404" [Medium,In progress] - Assigned to Dougal Matthews (d0ugal) 16:15:13 <rakhmerov> I guess we discussed that already 16:15:17 <d0ugal> I'll continue that and expect to have a new version soon 16:15:27 <rakhmerov> ok, awesome 16:15:27 <d0ugal> Yup, nothing else to say about my patch yet :) 16:15:30 <jpeeler> i haven't started working on mistral fully yet, so no direct status yet 16:15:48 <rakhmerov> ok 16:15:57 <d0ugal> jpeeler: You have been working on the TripleO workflows which is somewhat related :) 16:15:58 <rakhmerov> jpeeler: just for my records, what time zone are you in? 16:16:09 <rakhmerov> yeah :) 16:16:46 <jpeeler> d0ugal: true... it's sort of hard to know what's relevant to report 16:16:54 <d0ugal> indeed. 16:17:22 <jpeeler> i've been working on adding zaqar messaging to some of our workflows. since i'm pretty new still, this has been helpful in learning mistral in general. 16:17:29 <jpeeler> rakhmerov: EST 16:17:42 <rakhmerov> well, I'm personally also interested in those Mistral related activities 16:18:07 <rakhmerov> jpeeler: ok, so it's early morning for you now 16:18:09 <rakhmerov> got it 16:18:35 <jpeeler> ok that's good to hear. didn't want to assume anything. and it's 12:18 PM for me now, fwiw 16:18:46 <rakhmerov> yeah, I see 16:18:51 <rakhmerov> not too early 16:19:06 <ddeja> but in a lunch time ;) 16:19:18 <rakhmerov> let's move on to the next topics? 16:19:22 <d0ugal> +1 16:19:34 <rakhmerov> ooh, so is our meeting stealing your lunch? ;) 16:19:49 <rakhmerov> that's pretty bad :) 16:20:15 <rakhmerov> #topic Custom Actions API spec 16:20:34 <rakhmerov> https://review.openstack.org/#/c/325769/ 16:21:14 <rakhmerov> so, again: I hope it's kind of finished although someone may want to see more details to be able to start working on it 16:21:23 <rakhmerov> but I'd like to review it again 16:21:40 <rakhmerov> hoping that we'll agree on it soon 16:21:48 * d0ugal adds to review list for tomorrow 16:22:07 <rakhmerov> once it happens I'd like to get you involved in helping with this 16:22:12 <jpeeler> yeah i'll read over it too 16:22:19 <rakhmerov> ok 16:22:38 <rakhmerov> so basically, I was going to ask who could be volunteers for this 16:22:59 <rakhmerov> I know that rbrady wants to help with this 16:23:08 <rakhmerov> not sure who else of you is directly interested 16:23:34 <rakhmerov> I see a number of work items that we could spread among us 16:23:45 <d0ugal> I would like to help 16:24:00 <ddeja> I *may* be able to help, but not sure how long I will be dealing with RPC API, so I do not want to declare anything right now 16:24:05 <rakhmerov> d0ugal: yeah, I think it's directly related to you 16:24:26 <d0ugal> rakhmerov: Yup :) 16:24:43 <rakhmerov> ddeja: thanks! Let's focus on RPC first, I hope we can knock it down soonish 16:25:01 <rakhmerov> the reason I'm paying that much attention to this topic is its importance 16:25:11 <rakhmerov> no need to repeat probably 16:26:02 <rakhmerov> ok, so rbrady, d0ugal and potentially ddeja are ready to help with Custom Actions API 16:26:08 <jpeeler> if it's a three person job, i could also help 16:26:13 <jpeeler> s/three/four 16:26:16 <rakhmerov> #info rbrady, d0ugal and potentially ddeja are ready to help with Custom Actions API 16:26:26 <rakhmerov> #info and jpeeler 16:26:27 <rakhmerov> :) 16:27:01 <rakhmerov> jpeeler: yeah, I think we could split it 16:27:34 <rakhmerov> even if we couldn't this kind of work requires as much feedback as possible anyway 16:28:07 <rakhmerov> we're working an API, it should be as thoughtfully designed as possible 16:28:15 <d0ugal> +1 16:28:30 <rakhmerov> we need to be posting a lot info on ML as well to get a wider feedback 16:29:06 <rakhmerov> so on one hand we need to be active with that, but not sacrificing quality at the same time 16:29:18 <rakhmerov> (trying to inspire you) 16:29:19 <rakhmerov> :) 16:29:26 <rakhmerov> (and myself, too) 16:29:34 <rakhmerov> ok 16:29:36 * ddeja feels inspired 16:29:40 <rakhmerov> :)) 16:29:45 <rakhmerov> ddeja: thanks man! 16:30:07 <rakhmerov> ok, nothing else on this topic 16:30:27 <rakhmerov> #topic Automated testing for OpenStack actions 16:31:09 <rakhmerov> this is a topic that we wanted to discuss with hparekh but seems like he didn't make it to the meeting 16:31:39 <d0ugal> Sounds like an interesting topic 16:31:39 <rakhmerov> he had some ideas on how we can automate testing of the whole bunch of OpenStack actions that we have 16:31:43 <rakhmerov> yeah 16:32:13 <d0ugal> I would like a way to automate testing workflows, which is sort-of related :) 16:32:22 <rakhmerov> he wrote me an email about possible approaches (~ an hour ago), but I didn't read it yet 16:32:34 <rakhmerov> d0ugal: what workflows? 16:32:50 <rakhmerov> well, yes, that's the next level I guess 16:33:28 <d0ugal> rakhmerov: Just any workflows. There is no easy way to automate testing or even validating them 16:33:29 <rakhmerov> the reason it was brought up is because people keep coming in to our IRC and claiming that Mistral doesn't work! 16:33:34 <d0ugal> lol 16:34:02 <ddeja> rakhmerov: I kind of understand them... Remember my first time with mistral ;) 16:34:06 <rakhmerov> because they try some stuff and that really now always works, and in most cases it turns out that it's about not working OpenStack actiosn 16:34:16 <rakhmerov> not Mistral itself 16:34:29 <rakhmerov> ddeja: yep 16:34:36 <rakhmerov> I know 16:34:39 <d0ugal> ddeja: did it not work? :) 16:34:46 <rakhmerov> I clearly realize that's painful 16:35:07 <rakhmerov> you all know how these actions look like from implementation perspective 16:35:12 <ddeja> d0ugal: yup. Took me sometime to make it work despite the fact that I had some OpenStack background 16:35:31 <rakhmerov> they are dynamically generated with using those mapping.json file that we have to support manually 16:35:46 <rakhmerov> and if some APIs change we have to react 16:35:58 <rakhmerov> so 16:36:18 <rakhmerov> it would be really cool if we could test them automatically 16:36:27 <rakhmerov> at some gate 16:36:49 <rakhmerov> may be at least not on every commit, I don't know 16:37:01 <rakhmerov> because I can imagine how many tests that would be 16:37:25 <rakhmerov> or maybe we could create multiple gates, a gate per a service (e.g. for Nova) 16:37:43 <rakhmerov> or maybe we could take a representative subset of actions per service 16:37:48 <d0ugal> They should probably be non-voting gates 16:37:55 <rakhmerov> (just brainstorming now actually) 16:37:57 <ddeja> I guess the last one would be best 16:38:04 <rakhmerov> d0ugal: I think so, yes 16:38:18 <rakhmerov> ddeja: yep, maybe 16:38:25 <rakhmerov> one more important thing here 16:38:37 <ddeja> Just write some workflows that would use core openstack-clients and test them 16:38:45 <rakhmerov> at some point we will move OpenStack actions to a separate subproject 16:39:13 <rakhmerov> and I think it's ok if that subproject's gates will take long to complete 16:39:27 <rakhmerov> because it won't hard core Mistral development 16:39:41 <rakhmerov> ddeja: yes 16:40:16 <rakhmerov> d0ugal: btw, as far as workflow, do you mean some real-life workflows that use some OpenStack stuff? 16:40:26 <d0ugal> rakhmerov: Yup 16:40:33 <rakhmerov> or you're mostly talking about workflow capabilities? 16:41:09 <rakhmerov> if latter, that's covered well in unit tests and we also planned to write a number of functional tests 16:41:12 <d0ugal> rakhmerov: Real life workflows. We don't have a good approach to testing and validating them in TripleO 16:41:24 <rakhmerov> d0ugal: I see 16:41:32 <d0ugal> It is a tricky problem :) 16:41:37 <rakhmerov> I bet 16:42:02 <rakhmerov> well, the first thing that comes to my mind is standing up a production-like environment 16:42:19 <rakhmerov> but that is probably a hard thing to do 16:42:44 <d0ugal> rakhmerov: Yeah, we have that and it works quite well, but it will probably only ever cover the main path, lots of edge cases wont be tested 16:42:50 <rakhmerov> and run tests against it, but making sure we do a proper clean up after tests etc. 16:43:09 <rakhmerov> d0ugal: right, I see what you're saying 16:43:16 <rakhmerov> hm.. yeah 16:43:22 <d0ugal> ... but I don't want to side track the meeting :) 16:43:32 <d0ugal> it is just something I am thinking about 16:43:33 <rakhmerov> d0ugal: that's fine 16:43:55 <rakhmerov> d0ugal: share your ideas pls if you get them 16:44:25 <rakhmerov> one fundamental problem I see here is that most workflows take long 16:44:30 <d0ugal> rakhmerov: Will do. I'll find some time to write them to the list 16:44:55 <d0ugal> Ideally I would really like a way to run a workflow with mocked services 16:45:04 <rakhmerov> so that if we decide to test all various paths it'll be no good ) 16:45:21 <rakhmerov> d0ugal: hm... that is interesting 16:46:50 <rakhmerov> d0ugal: I think it's actually a complex problem meaning that the ultimate goal is to have more confidence in workflows that we deal with and because of that it's not only automated tests but also other means that would help to have that confidence 16:47:13 <rakhmerov> validation tools, graphical tools that provide some assistance when designing them 16:47:14 <rakhmerov> etc. etc 16:47:40 <rakhmerov> more convenient ways to debug workflows and find errors 16:47:49 <rakhmerov> the last one is especially important 16:47:56 <d0ugal> +1 16:48:17 <d0ugal> rakhmerov: I spoke with rbrady about some of this, so maybe we can chat again about it at the meeting next week. 16:48:30 <rakhmerov> absolutely 16:48:40 <rakhmerov> yes, please guys share any ideas 16:49:07 <rakhmerov> I have a bunch of them in my head ) 16:49:21 <d0ugal> haha, I'm sure :) 16:49:28 <rakhmerov> just don't know how to find resources to help with all that 16:49:52 <rakhmerov> but some of that is already in progress 16:50:17 <rakhmerov> it's kind of a mystery for you probably but hopefully we'll make an announcement relatively soon 16:50:19 <d0ugal> Should we move onto open discussion? 16:50:22 <d0ugal> I guess we already are :) 16:50:26 <rakhmerov> by "we" I mean Nokia 16:50:40 <rakhmerov> ok, yes 16:50:43 <d0ugal> oh, that sounds interesting 16:51:12 <rakhmerov> #action rakhmerov: read Hardik's email about testing OS actions and share with the team 16:51:24 <rakhmerov> #topic Open Discussion 16:52:19 <rakhmerov> d0ugal: yeah :) I saw a little myself but once finished it will be a bomb! 16:52:28 <rakhmerov> no kidding 16:53:00 <rakhmerov> I'm not allowed to tell more though atm 16:53:11 <d0ugal> hah, so you just tease us :) 16:53:15 <ddeja> I have one topic: Since alternative RPC layer is done (for better or worse) I would like to start working on tagging each message for processing type (at-least-once vs at-most-once). Are you guys OK with this? 16:53:17 <rakhmerov> yes!! 16:53:48 <rakhmerov> ddeja: yes, I'm ok 16:54:10 <rakhmerov> just be ready to fix something in your patches 16:54:14 <rakhmerov> if needed 16:54:21 <ddeja> rakhmerov: yeah, sure 16:55:50 <rakhmerov> I feel like I'm done for today, nothing else from my side 16:55:58 <ddeja> I'm also done 16:56:03 * d0ugal is done 16:56:12 <rakhmerov> mgershen, jpeeler? 16:56:19 <jpeeler> all good here 16:56:38 <rakhmerov> jpeeler: I hope you will still have your lunch :)) 16:56:51 <rakhmerov> ok, guys 16:56:54 <jpeeler> ha, i'm going to go do that now! 16:56:59 <rakhmerov> thanks a lot for joining today 16:57:15 <rakhmerov> you have a great time during the week ) 16:57:24 <rakhmerov> bye 16:57:27 <d0ugal> Thanks! 16:57:28 <ddeja> bye 16:57:32 <d0ugal> bte 16:57:35 <d0ugal> bte! 16:57:38 <d0ugal> bye! 16:57:42 * d0ugal is also done, he can't type 16:57:42 <rakhmerov> #endmeeting