16:01:16 <rakhmerov> #startmeeting Mistral
16:01:17 <openstack> Meeting started Mon Jan 16 16:01:16 2017 UTC and is due to finish in 60 minutes.  The chair is rakhmerov. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:01:18 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:01:20 <openstack> The meeting name has been set to 'mistral'
16:01:22 <rakhmerov> hi all
16:01:26 <ddeja> hello
16:01:33 <d0ugal> Hey!
16:01:42 <rakhmerov> ok, I'll wait a minute..
16:01:52 <rakhmerov> someone else may join
16:03:42 <rakhmerov> ok
16:03:45 <rakhmerov> hi again
16:03:46 <rbrady> o/
16:03:58 <rakhmerov> #topic Review Action Items
16:04:33 <rakhmerov> 1. rakhmerov: review https://review.openstack.org/#/c/411412/
16:05:06 <rakhmerov> #action review https://review.openstack.org/#/c/411412/
16:05:07 <rakhmerov> :)
16:05:21 <rakhmerov> 2. ddeja: close https://bugs.launchpad.net/mistral/+bug/1624284 and file a new more specific bug with the description of a corner case which is not yet fixed
16:05:21 <openstack> Launchpad bug 1624284 in Mistral "MessagingTimeout when executing mistral actions" [Critical,Fix committed] - Assigned to Dawid Deja (dawid-deja-0)
16:05:27 <rakhmerov> ddeja: it's done, right?
16:05:45 <ddeja> rakhmerov: yes, I'm just not sure if 'Fix committed' is OK
16:06:01 <ddeja> or if it should be 'Fix released'
16:06:31 <rakhmerov> it needs to be "Fix released"
16:06:35 <rakhmerov> ok
16:06:53 <ddeja> rakhmerov: OK
16:07:06 <rakhmerov> d0ugal, rbrady: as far as https://review.openstack.org/#/c/411412/, sorry
16:07:09 <ddeja> done ;_
16:07:11 <ddeja> ;)
16:07:30 <rakhmerov> I'm still getting used to work after long holidays
16:07:40 <rakhmerov> and buried with tons of things
16:07:41 <d0ugal> rakhmerov: haha, me too I think.
16:07:45 <rakhmerov> ok
16:07:45 <d0ugal> (and I got back a week earlier)
16:07:53 <rakhmerov> will do it this week for sure..
16:07:58 <rakhmerov> yes
16:08:04 <rbrady> ack
16:08:29 <rakhmerov> #topic Current status (progress, issues, roadblocks, further plans)
16:09:41 <rakhmerov> my status: last week mostly reviewing patches, sent a small client patch to add user domain name and project domain name params into "TARGET" group, now working on https://review.openstack.org/#/c/420650/ which describes a new feature that we need internally
16:10:19 <ddeja> My status:
16:10:21 <ddeja> * Prepared a fix for tests that depends on order of returns from DB
16:10:23 <ddeja> * Make kombu driver work in multi-thread mode
16:10:25 <ddeja> This week:
16:10:27 <ddeja> * Refactor RPC Utils (it is needed for next item)
16:10:29 <ddeja> * Make kombu driver understand multiple rabbit hosts
16:10:33 <d0ugal> I sent a tiny patch today which resolved a TripleO issue. https://review.openstack.org/#/c/420547/ - I have been thinking this afternoon how to avoid this happening again - so I am interested if anyone has ideas for better testing OpenStack actions.
16:11:11 <d0ugal> I am also looking into improving the OpenStack action loading, I a bug with it - I need to open a bug in launchpad still.
16:11:26 <ddeja> d0ugal: better testing of OpenStack action is our Holy Graill ;)
16:11:49 <d0ugal> ddeja: haha, indeed. We need all the other OpenStack projects to test themselves with Mistral :-D
16:13:03 <rakhmerov> d0ugal: as far as I remember Hardik Parekh was proposing some idea on how to test openstack actions but we put it on a shelve
16:13:16 <d0ugal> rakhmerov: it is a difficult problem :(
16:13:17 <rakhmerov> do you remember by any chance?
16:13:23 <rakhmerov> yes, indeed
16:13:30 <d0ugal> I don't remember.
16:13:41 <rakhmerov> and yes, it's our Holy Grail
16:13:57 <rakhmerov> if we do this we'll attract much more users, I'm sure
16:14:17 <d0ugal> Yeah
16:14:37 <d0ugal> How does openstackclient test against different projects?
16:14:41 <rakhmerov> d0ugal: ok, honestly I don't remember all the details of Hardik's proposal, let me try to find it in the mailing list
16:14:55 <d0ugal> They have the advantage that users write plugins, but they do still support some projects in the core.
16:14:57 <rakhmerov> #action rakhmerov: find Hardik's proposal on testing OpenStack actions
16:15:04 <d0ugal> Thanks!
16:15:38 <rakhmerov> d0ugal: interesting, we can probably find out what they do in order to test things
16:15:58 <d0ugal> I also have a couple of ideas, but I am only just starting to think about it :)
16:15:58 <ddeja> rakhmerov, d0ugal http://lists.openstack.org/pipermail/openstack-dev/2016-June/098301.html
16:16:25 <rakhmerov> I was also hoping that we could consider better testing as part of this Actions refactoring initiative
16:16:36 <d0ugal> rakhmerov: +1
16:16:50 <rakhmerov> ddeja: ooh, yes, that's it
16:16:51 <rakhmerov> thanks
16:16:56 <ddeja> no problem
16:17:26 <rakhmerov> d0ugal: we also need to think how to make versioning etc.
16:17:38 <rakhmerov> remember Dmitri Tantsur who came to our channel with this
16:17:38 <d0ugal> rakhmerov: Yeah, it becomes a big problem quickly :-D
16:18:02 <rakhmerov> yeah, the fact of the matter is.. It's f..ing huge
16:18:05 <d0ugal> Yeah, I know Dmitri :)
16:18:18 <rakhmerov> we'll be moving towards that
16:18:21 <d0ugal> Maybe a good place to start it identify the different issues, then we can tacklie a bit at a time.
16:18:44 <d0ugal> I'll try and spend some time looking into this.
16:18:48 <rakhmerov> yeah, the good time to discuss all that stuff is PTG I think
16:18:55 <d0ugal> That would be good
16:19:06 <rakhmerov> with you participating remotely :)
16:19:07 <rakhmerov> may be
16:19:23 <rakhmerov> d0ugal: yes, please do spend time if you have it
16:19:27 <rakhmerov> it'd be great
16:19:28 <ddeja> d0ugal: BTW, did you try the support program (assuming money is the problem)
16:19:42 <rakhmerov> I'll participate as much as possible
16:20:02 <d0ugal> ddeja: I didn't - that isn't the issue tho :)
16:20:04 <rakhmerov> I think, speaking practically, I'll have much more time in 2-3 weeks
16:20:27 <rakhmerov> now have to turn my look inside at some internal stuff
16:20:41 <ddeja> rakhmerov: ACK
16:20:45 <rakhmerov> okay
16:21:08 <rakhmerov> d0ugal: ooh, money isn't the issue?
16:21:33 <rakhmerov> I know many folks using this program
16:21:36 <rakhmerov> really
16:21:46 <rakhmerov> ok, I see
16:21:51 <d0ugal> nah, I think Red Hat would have sent me, but I am unavailable that week.
16:22:12 <rakhmerov> aah... too bad
16:22:21 <rakhmerov> ok, it is what it is
16:22:27 <d0ugal> Indeed. Not to worry.
16:22:33 <rakhmerov> yes
16:22:35 <d0ugal> I'm happy to not have jet lag :)
16:22:39 <rakhmerov> :)
16:22:55 <d0ugal> In other news, I started writing a Mistral Workflow linter
16:23:06 <rakhmerov> ok, I included the topic "PTG preparations" so let me formally set it
16:23:15 <rakhmerov> ooh, what's that?
16:23:17 <d0ugal> My initial goal is to catch common issues I am seeing in TripleO - but it could be useful more generally too.
16:23:26 <rakhmerov> hah, cool
16:23:32 <d0ugal> rakhmerov: like a flake8 for workflows I guess :)
16:23:38 <rakhmerov> yes, I see
16:24:06 <rakhmerov> btw, there's one guy who made a prototype of sublime plugin to edit Mistral workflows
16:24:10 <rakhmerov> just FYI
16:24:29 <rakhmerov> it's still just a prototype though working with an older version of DSL
16:24:56 <rakhmerov> I was also thinking about something like this but it seemed to be a very far future
16:25:05 <rakhmerov> #topic PTG preparations
16:25:47 <rakhmerov> so, I was mostly going to ask you about maybe some ideas as far as what else we can do to make our PTG more productive/interesting
16:26:17 <d0ugal> How much time do you have?
16:26:41 <rakhmerov> so, my question is: do you see any other important topics not listed in https://etherpad.openstack.org/p/mistral-ptg-pike ?
16:26:47 <rakhmerov> d0ugal: what do you mean?
16:27:00 <d0ugal> rakhmerov: at the PTG, is it two full days? or?
16:27:16 <rakhmerov> 3
16:27:50 <mgershen> it says Feb 20-24, 2017 on the ptg page
16:28:06 <rakhmerov> mgershen: hi, I didn't realize you were here :)
16:28:23 <mgershen> I didn't say hi. Hi all :)
16:28:24 <mgershen> https://www.openstack.org/ptg/
16:28:25 <d0ugal> mgershen: I don't think every project has rooms for every day
16:28:29 <rakhmerov> yes, first 2 days are reserved for various horizontal discussions etc
16:29:31 <rakhmerov> so, we can quickly go over these items and make sure that all important things are there
16:29:56 <rakhmerov> after that I'll fill them with more details and make them usable for the PTG
16:30:16 <ddeja> rakhmerov: do we also like to talk about upgradability?
16:30:36 <rakhmerov> ddeja: can you explain what you mean by that?
16:31:11 <ddeja> rakhmerov: I mean the rolling upgrades effort, that many projects took in the past 2 years
16:31:19 <ddeja> so that they can be live-upgraded
16:31:53 <rakhmerov> what "they"?
16:32:02 <rakhmerov> you mean Mistral components?
16:32:07 <rakhmerov> like engine, api etc.
16:32:46 <ddeja> well I meant the other project, like nova, cinder etc.
16:33:49 <rakhmerov> yes, I mean.. I know what rolling upgrade is but I heard so many times in different contexts so I'm confused about what specifically you mean
16:34:00 <rakhmerov> you're talking about upgrading what?
16:34:09 <ddeja> mistral components
16:34:18 <rakhmerov> ok
16:34:32 <ddeja> (btw, it's not a priority for myself, just it came to my mind when you asked what may be missing ;)
16:34:37 <rakhmerov> so that we have 1 version of Mistral installed and we need to upgrade it to a newer version
16:34:39 <rakhmerov> I see
16:35:01 <ddeja> yes, and end user should not notice that something is happening ;)
16:35:12 <rakhmerov> I thought maybe you're talking about a WF example that does it for something else
16:35:20 <ddeja> no no
16:35:20 <rakhmerov> ok, yes
16:35:34 <rakhmerov> well, ok, let's include this
16:36:04 <rakhmerov> but I'm not sure it's a near future for us unless someone new decides to do it
16:36:24 <ddeja> I think that rather we should discuss if it is needed
16:36:33 <ddeja> it may be too early
16:37:35 <rakhmerov> that's ok
16:37:50 <rakhmerov> on this one I don't expect us to have a long discussion at the PTG
16:38:00 <ddeja> OK
16:38:34 <rakhmerov> anything else you'd like to add?
16:38:36 <ddeja> oh, can we also submit bugs as topics?
16:38:42 <rakhmerov> ddeja: btw, new RPC
16:38:47 <rakhmerov> ddeja: sure
16:38:50 <rakhmerov> go ahead
16:38:57 <rakhmerov> whatever is important
16:39:05 <ddeja> https://bugs.launchpad.net/mistral/+bug/1656862 this one
16:39:05 <openstack> Launchpad bug 1656862 in Mistral "MessagingTimeout when run-action is used to run another action" [Undecided,Confirmed] - Assigned to Dawid Deja (dawid-deja-0)
16:39:12 <ddeja> important and hard to fix :/
16:39:48 <ddeja> rakhmerov: sure, new RPC
16:39:51 <d0ugal> Could we just detect this and block it?
16:39:57 <rakhmerov> ddeja: just included the item
16:39:58 <d0ugal> i.e. solve it by not allowing the behaviour?
16:40:11 <ddeja> d0ugal: well, that's the idea
16:40:18 <ddeja> but still it's not easy
16:40:23 <d0ugal> right
16:40:34 <rakhmerov> can you elaborate?
16:40:35 <ddeja> (or I'm stupid, but I'd link to think that the problem is complicated ;)
16:40:39 <rakhmerov> why is it not easy?
16:41:48 <ddeja> I tried to solve it once that way, but I ended up in a place where I needed to write some kind of static analyzer in an API
16:41:59 <d0ugal> lol
16:42:05 <ddeja> that would read each requests to see if it's not doing something illegall
16:42:15 <d0ugal> scary :)
16:42:16 <rakhmerov> ah, yes
16:42:22 <rakhmerov> sounds scary, yeah
16:42:37 <rakhmerov> ok, then I think it's a good candidate to discuss too
16:42:57 <rakhmerov> maybe we can even extend it to a more generic topic
16:42:58 <ddeja> I was thinking that we can simply block of usage of mistralclient in executor, but that block too much
16:43:17 <rakhmerov> because I believe we now can do some other crazy things too
16:43:45 <ddeja> what kind of crazy things?
16:43:45 <rakhmerov> like what if we have a workflow A that calls workflow B that calls workflow A
16:44:08 <ddeja> hm
16:44:23 <rakhmerov> under certain conditions we can have a never ending recursion which will keep generating data in DB endlessly
16:44:32 <d0ugal> I created an infinite loop in Mistral - that was fun.
16:44:37 <ddeja> assuming that your server doesn't have inifinite memory, it should kill it
16:44:41 <rakhmerov> it is fun, yes
16:45:08 <d0ugal> ddeja: it kept going for me for several days :)
16:45:09 <rakhmerov> I heard an opinion about that: if users are smart they won't shoot their own legs
16:45:13 <rakhmerov> but I disagree
16:45:35 <rakhmerov> things like that can impact the whole Mistral instance and hence other users
16:45:48 <d0ugal> yeah, it could be a big problem
16:45:52 <rakhmerov> d0ugal: haha :) How much data did you accumulate?
16:46:10 <rakhmerov> ddeja: can you please add this item?
16:46:14 <d0ugal> rakhmerov: I forget, I just deleted the VM in the end.
16:46:17 <rakhmerov> into the etherpad
16:46:20 <rakhmerov> or you already did?
16:46:26 <rakhmerov> d0ugal: I see
16:46:29 <ddeja> rakhmerov: I've added one about deadlocks in run-action
16:46:41 <ddeja> let me re-write it
16:47:09 <sharatss_> rakhmerov ddeja d0ugal: hi
16:47:10 <rakhmerov> yes, you can re-write it to a more generic topic and specify these two as examples
16:47:18 <rakhmerov> sharatss_: hi, how's it going?
16:47:47 <d0ugal> het sharatss_
16:47:50 <d0ugal> hey*
16:48:04 <ddeja> rakhmerov: done
16:48:07 <ddeja> hi sharatss_
16:48:34 <sharatss_> rakhmerov: everything's fine. Stuck in traffic :) joined in phone :)
16:48:35 <rakhmerov> ddeja: ok
16:49:30 <sharatss_> i had an idea
16:49:35 <rakhmerov> one specifically important thing (at least I think so) that I'd like to discuss in details is API/CLI v3
16:50:27 <rakhmerov> maybe I a little bit exaggerate the importance of it but seems high time for this work
16:50:43 <rakhmerov> especially, I'm fed up with our CLI
16:50:45 <ddeja> BTW, I'm rather pessimistic about me going to Atlanta. But I don't have any official info
16:50:49 <rakhmerov> I find it very inconvenient
16:51:02 <d0ugal> rakhmerov: what is the goal/motivation for the new CLI/API?
16:51:11 <d0ugal> rakhmerov: Is it simply to improve usability?
16:51:18 <rakhmerov> ddeja: pessimistic? :) you don't want to go or most likely won't go?
16:51:30 <rakhmerov> d0ugal: pretty much usability, yes
16:51:46 <d0ugal> rakhmerov: cool, I find it weird to use and explain too :)
16:52:01 <rakhmerov> very stupip CLI commands, hard to remember, human unfriendly
16:52:13 <rakhmerov> like "mistral execution-create"
16:52:22 <rbrady> +1
16:52:24 <ddeja> rakhmerov: I'd like to go, but I don't know if I'll be send there
16:52:34 <d0ugal> rakhmerov: but do we need a new API to improve the CLI?
16:52:40 <rakhmerov> how a mentally normal person can ever suggest that it starts a workflow
16:52:51 <rakhmerov> ddeja: yes, I see
16:52:59 <rakhmerov> d0ugal: no
16:53:28 <sharatss_> rakhmerov: are we having plans to migrate our CLI to openstackclient any time soon?
16:53:37 <d0ugal> sharatss_: it is already there :)
16:53:46 <ddeja> what?
16:53:51 <ddeja> really?
16:54:00 <rakhmerov> so the idea is to start from CLI, come up with useful commands and then work on API too that would allow to implement all those commands
16:54:26 <d0ugal> sharatss_, ddeja: http://paste.openstack.org/show/595076/
16:54:28 <rakhmerov> I see API v3 as a revision of all we did in v2 sporadically
16:54:35 <rakhmerov> new endpoints, filters, sorting
16:54:35 <sharatss_> d0ugal: i have never used the osc then :(
16:54:55 <rakhmerov> I just want to make it more systematic, logical and consistent
16:54:58 <ddeja> well, openstack workflow execution create seems OK to me
16:55:17 <ddeja> and since we should migrate from each project clients to one openstack client...
16:55:42 <sharatss_> ddeja: yes. At least it is self explanatory to some extent
16:55:42 <d0ugal> ddeja: that wont work really
16:55:46 <rakhmerov> like there's "v2/executions" (what kind of executions? we have 3 kinds of them), "v2/tasks" (in fact, executions too but it's not mentioned) and "v2/action_executions"
16:56:02 <d0ugal> ddeja: openstack client doesn't expose all the features of all the openstack clients
16:56:31 <ddeja> right
16:56:37 <rakhmerov> d0ugal: ++++
16:56:47 <rakhmerov> ddeja: yes, d0ugal is 100% right
16:56:53 <d0ugal> and we also want to keep the Mistral CLI without openstackclient
16:57:14 <ddeja> oh yes
16:57:19 <ddeja> non-openstack deployments
16:57:24 <d0ugal> yup
16:57:26 <ddeja> forgotten about it
16:57:32 <rakhmerov> I see openstackclient as one point to access OpenStack capabilities but, it's really limited because it enforces to follow a certain structure
16:57:39 <d0ugal> So we have a good situation now, we just need to make it easier to use
16:57:58 <ddeja> ok, now I can totaly agree
16:58:45 <rakhmerov> I see no problem having our own top-notch CLI with all the cool capabilities and human friendliness and have an osc plugin compliant with its requirements
16:58:57 <ddeja> sure
16:59:03 <rakhmerov> if you're only getting familiar with Mistral then osc is probably the way to go
16:59:09 <sharatss_> This sounds good
16:59:21 <rakhmerov> if you're using Mistral everyday, go and use the more powerful tool
16:59:38 <rakhmerov> ok, I didn't notice that our time is up
16:59:44 <ddeja> yuo
16:59:47 <ddeja> yup*
16:59:59 <rakhmerov> ok, thanks for coming
17:00:08 <ddeja> thanks you very much
17:00:08 <d0ugal> Thanks!
17:00:09 <sharatss_> Thank you
17:00:10 <rakhmerov> have a good week
17:00:28 <rakhmerov> let's try to make the ball rolling faster :)
17:00:32 <rakhmerov> I mean development etc.
17:00:37 <ddeja> sure
17:00:39 <sharatss_> I have few things to discuss. Maybe tomorrow in our channel
17:00:53 <rakhmerov> ooh sharatss_, sorry
17:01:03 <rakhmerov> can you quickly mention what they are?
17:01:11 <d0ugal> if nobody is waiting?
17:01:13 <rakhmerov> just so we don't forget
17:01:20 <rakhmerov> d0ugal: dunno
17:01:24 <sharatss_> rakhmerov:. It takes time. We'll have it tomorrow
17:01:26 <rakhmerov> they will tell I guess..
17:01:38 <rakhmerov> sharatss_: ok, I'm open to discuss it tomorrow
17:01:40 <rakhmerov> bye
17:01:43 <ddeja> after an hour everyone can end a meeting ;)
17:01:51 <rakhmerov> #endmeeting