15:01:33 <rakhmerov> #startmeeting Mistral
15:01:34 <openstack> Meeting started Mon Jul  3 15:01:33 2017 UTC and is due to finish in 60 minutes.  The chair is rakhmerov. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:01:35 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:01:37 <openstack> The meeting name has been set to 'mistral'
15:01:42 <d0ugal> o/
15:02:01 <rakhmerov> hey
15:02:07 <rakhmerov> waiting..
15:02:08 <toure> hello
15:02:12 <rakhmerov> hi toure
15:02:20 <toure> rakhmerov hello
15:02:20 <rbrady> o/
15:02:28 <rakhmerov> hi rbrady
15:02:57 <rbrady> hello rakhmerov.  How are you doing?
15:03:07 <rakhmerov> doing ok )
15:03:10 <rakhmerov> still alive
15:03:30 <rakhmerov> you?
15:03:54 * d0ugal is alive too
15:04:00 <rakhmerov> :)
15:04:06 <rakhmerov> it's funny
15:04:19 * toure is mostly alive
15:04:23 <d0ugal> but I hope you are feeling better rakhmerov!
15:04:29 <rakhmerov> i was reading a book yesterday where the author has a joke about the word "job"
15:04:30 <toure> +1
15:04:37 <rakhmerov> about what it means
15:04:43 <rakhmerov> "just over broke"
15:04:44 <rbrady> rakhmerov: I'm doing okay
15:05:13 <rakhmerov> just a little better than a bankrupt ) kind of
15:05:35 <d0ugal> hah
15:05:41 <rakhmerov> but I believe we're doing better than this ))
15:05:44 <rakhmerov> definition
15:05:48 <apetrich> :)
15:05:49 <rakhmerov> ok, let's start
15:06:00 <rakhmerov> #topic Current status
15:07:13 <rakhmerov> my status: spent significant time on reviewing patches, including namespaces spec (please do too), also almost finished Mistral java client in OpenStack4j (will be polishing and testing it) and keep working on HA tasks
15:07:49 <rakhmerov> specifically, learning OpenStack CI and started working on a new gate
15:08:03 <d0ugal> With help from apetrich I was able to remove Mistral from the global requirements, which solved a number of problems
15:08:24 <rakhmerov> excellent!
15:08:39 <rbrady> my status: slowing working on the port of keystone utils to mistral-lib.  I'm not a keystone expert and recent patches/changes have left me reading more code than writing it in the last week
15:08:55 <rakhmerov> ok
15:09:00 <toure> status: workflow error analysis has taken a turn as sub-workflow tracking proves to be difficult, d0ugal pointed out this flaw in my proposal, He and I have been working on a solution. The problem stems from the parent execution not having the ability to track it's children tasks.
15:09:32 <rakhmerov> yeah, it needs thinking
15:09:41 <apetrich> me: as d0ugal said mistral-lib is going strong. I'm fixing some documentation to reflect that
15:09:44 <rakhmerov> maybe it's even worth changing the DB model for that
15:09:59 <rakhmerov> toure: I also left some questions in the patch, take a look
15:10:02 <toure> rakhmerov that would help
15:10:14 <toure> k, will take a look
15:10:16 <apetrich> also thanks rakhmerov to make the change to mistral-lib serializer so we can remove that from mistral soon
15:10:21 <rakhmerov> apetrich: documentation where? TripleO?
15:10:28 <apetrich> mistral/docs
15:10:37 <d0ugal> https://docs.openstack.org/developer/mistral/developer/creating_custom_action.html
15:10:37 <apetrich> rakhmerov, ^
15:10:39 <rakhmerov> ooh, ok
15:11:01 <rakhmerov> woops
15:11:03 <rakhmerov> 404
15:11:14 <rakhmerov> I guess it's a temporary problem
15:11:23 <rakhmerov> ok, good anyway )
15:11:27 <d0ugal> It works for me :)
15:11:31 <rakhmerov> really?
15:11:33 <rakhmerov> haha
15:11:48 <apetrich> https://github.com/openstack/mistral/blob/master/doc/source/developer/creating_custom_action.rst
15:12:32 <rakhmerov> yes, ok
15:13:02 <rakhmerov> we need to strive to finish all main changes in mistral lib within 2 weeks
15:13:05 <rakhmerov> it's very important
15:13:15 <rakhmerov> I'll try to participate too as much as possible
15:13:33 <d0ugal> What other changes do we need in mistral-lib?
15:14:00 <d0ugal> I see the mistral-extra work (moving actions etc.) as a different task - I am not sure this will happen in time for P
15:14:03 <apetrich> Good question
15:14:20 <rakhmerov> d0ugal: I think we need to port OpenStack actions, it will prove that everything is ok with mistral-lib
15:14:26 <apetrich> I see the serialization that we can move from mistral into mistral-lib
15:14:37 <rakhmerov> I think it's not 100% a different task
15:14:47 <d0ugal> rakhmerov: Mistral itself uses mistral-lib, that should be proof enough :)
15:14:57 <rakhmerov> ok, I may be wrong
15:14:57 <apetrich> context will stay in mistral but we could use a contract from mistral-lib
15:14:58 <d0ugal> and tripleo does too, which is some external proof.
15:15:15 <apetrich> AFAIK
15:15:17 <rakhmerov> I haven't really looked into it for a while and I just want to make sure that it's in a good shape
15:15:28 <rakhmerov> and also documented well
15:15:33 <d0ugal> but yes, it would have been nice if mistral-extra used it as that would cover more use cases.
15:15:42 <d0ugal> +1
15:16:15 <rakhmerov> I'm talking about mistral-extra work because it may potentially point to some issues
15:16:18 <rakhmerov> hopefully not
15:16:36 <d0ugal> Right
15:16:37 <rakhmerov> btw, as with config options used by actions
15:16:49 <rbrady> if we're going to port openstack actions to mistral-extra, does that mean we'll need to port the action generator too
15:16:50 <rbrady> ?
15:17:12 <d0ugal> rbrady: yes, I think so.
15:17:23 <rakhmerov> yes
15:17:40 <rakhmerov> it's part of the framework related to actions
15:17:48 <rakhmerov> it shouldn't be in mistral
15:17:48 <d0ugal> I had thought about this a bit actually
15:17:55 <d0ugal> We have entry_points for invididual actions
15:18:07 <d0ugal> but maybe we should have a setup.py entry_point for action generators
15:18:11 <rbrady> there is more to investigate wrt to the action generator imports
15:18:18 <d0ugal> then a generator for aws (for example) could be written and registered
15:18:25 <rbrady> I like d0ugal's idea
15:18:38 <rakhmerov> yeah
15:18:39 <rakhmerov> sure
15:18:43 <d0ugal> but I have not looked into this idea at all :)
15:18:51 <d0ugal> There is lots that needs to happen first.
15:19:07 <rakhmerov> keep in mind that what we've done so far in actions is pretty primitive and not very useable
15:19:35 <d0ugal> What do you mean?
15:19:38 <rakhmerov> there should be better ways of configuring them
15:19:43 <d0ugal> aha
15:19:47 <rakhmerov> I mean your idea is good! )
15:19:58 <d0ugal> Yes, I think mistral-extra will also need a configuration file
15:20:08 <rakhmerov> don't take what we have now as a good example
15:20:16 <d0ugal> Right
15:20:26 <rakhmerov> d0ugal: sure, yes, but that will be its own config
15:20:28 <rbrady> rakhmerov: configure them?  as in configure the client constructors?
15:20:31 <rakhmerov> of a separate project
15:20:37 <rakhmerov> in my view
15:21:52 <rakhmerov> rbrady: no, I mean user experience in general. We have to update that huge mapping if something is out of date, we have to register individual entries in setup.cfg
15:21:53 <rakhmerov> etc.
15:22:34 <rakhmerov> I mean when you're working on this transition just try to think how to also improve it
15:22:46 <rakhmerov> I know there's not much time but anyway
15:22:46 <d0ugal> Right
15:22:51 <rakhmerov> this work will continue anyway
15:22:56 <d0ugal> because it would also be better if you didn't need to restart Mistral
15:23:02 <d0ugal> people always find that strange :)
15:23:03 <rakhmerov> yes, ideally
15:23:33 <rbrady> autodiscovery for actions sounds cool
15:23:38 <rakhmerov> yes
15:23:47 <d0ugal> I think we could solve that quite easily by executing actions in their own process - which would also open the door for non-python actions.
15:24:05 <rakhmerov> true
15:24:23 <rakhmerov> some time ago we also thought about the idea of an action provider
15:24:37 <rakhmerov> some abstraction that knows how to deliver actions to the system
15:24:51 <rakhmerov> its contract is basically: getActions()
15:25:01 <rakhmerov> and it would return all actions
15:25:12 <apetrich> mmm
15:25:42 <rakhmerov> once it's integrated we could have implementations of action providers that could fetch actions from whatever comes to our mind
15:26:09 <rakhmerov> they could generate them dynamically as wrappers around external processes, generate based on DB records, etc.
15:26:51 <rakhmerov> it's just a general idea but it may be worth thinking of (there's even a BP for that)
15:27:00 <d0ugal> yeah, that might be a nice feature for Q
15:27:06 <rakhmerov> right
15:27:21 <rbrady> do I dare ask about actions as containers?
15:27:22 <apetrich> rakhmerov, do you think have a rabbitmq topic that actions register themselves or a two way communication for getActions to work?
15:27:24 <rakhmerov> by default, let's say we could have a provider that provides all built-in standard actions
15:27:43 <rakhmerov> apetrich: yeah, sounds nice
15:27:53 <apetrich> s/actions/actionproviders/ actually
15:27:59 <rakhmerov> got it, yes
15:28:19 <apetrich> nice
15:28:19 <rakhmerov> anyway, all these ideas can possibly co-exist
15:28:24 <d0ugal> yeah
15:28:46 <rakhmerov> how to configure them, how to develop the them more efficiently, how to deliver them to the system
15:28:51 <rakhmerov> lots of ideas )
15:29:23 <rakhmerov> let me formally open an open discussion
15:29:27 <rakhmerov> #topic Open Discussion
15:29:40 <rakhmerov> one thing to share
15:30:05 <rakhmerov> you have probably seen the news about migrating docs
15:30:13 <rakhmerov> kind of a sad news
15:30:15 <rakhmerov> but
15:30:30 <rakhmerov> for us it's kind of a good news because we don't need to move docs )
15:30:36 <rakhmerov> they are already in our repo
15:31:01 <rakhmerov> we'll just need to adjust docs according to the spec, basically change folder structure
15:31:22 <rakhmerov> and it's good that we got volunteers (my colleagues) to do this work
15:31:42 <rakhmerov> and they also volunteered to improve docs in general, including installation/configuration guides
15:31:49 <rakhmerov> RPMs, etc.
15:32:41 <rakhmerov> I think that's pretty much all from me
15:33:12 <d0ugal> Great, looking forward to that docs work.
15:33:17 <rakhmerov> any other topics?
15:33:19 <rakhmerov> yeah
15:33:37 <d0ugal> Nothing from me.
15:33:40 <rakhmerov> ok
15:33:50 <d0ugal> oh, actually
15:33:55 <toure> nothing from me either...
15:34:04 <d0ugal> [openstack-dev] [tc][all] Move away from meeting channels
15:34:07 <rakhmerov> ok
15:34:12 <rakhmerov> d0ugal: yeah?
15:34:29 <d0ugal> I seen this discussion on openstack-dev, it seems there might be a move away from these channels
15:34:49 <d0ugal> and it also sounds like some projects are moving away from meetings in general
15:35:07 <d0ugal> we don't need to change anything, but it is worth reading and we can then discuss what we want to change if anything.
15:35:14 <rakhmerov> making them ad-hoc?
15:35:30 <rakhmerov> yes, I read it briefly
15:35:32 <d0ugal> I think so
15:35:35 <rakhmerov> not all the content though
15:35:39 <rakhmerov> ok, no problem
15:35:50 <rakhmerov> I'm actually ok with the meetings we have
15:36:08 <rakhmerov> I still think that they are useful
15:36:26 <rakhmerov> but we can make them more flexible, for example
15:36:33 <rakhmerov> not the same day, not the same time
15:36:34 <d0ugal> yup
15:36:47 <rakhmerov> or on demand
15:37:04 <rakhmerov> if someone wants to discuss something important we can have them
15:37:10 <rakhmerov> so ok, let's think about it
15:37:26 <rakhmerov> I don't have any immediate proposals
15:37:31 <rakhmerov> regarding that
15:37:56 <rakhmerov> ok
15:38:04 <rakhmerov> just one more thing from me: as far as my personal contribution, I planned to participate actively in all things like docs and actions but I'll have to be focused on HA because of our internal very strict deadlines
15:38:35 <rakhmerov> essentially I have to make it work in HA in the next 2 months :)
15:38:41 <rakhmerov> no choice
15:39:00 <rakhmerov> but I'll be reviewing and helping anyway
15:39:14 <rakhmerov> ok, so let's end the meeting earlier then?
15:40:01 <rakhmerov> d0ugal, rbrady, toure, apetrich?
15:40:06 <toure> ack
15:40:09 <apetrich> aye
15:40:11 <rbrady> ack
15:40:14 <rakhmerov> ok
15:40:19 <d0ugal> sounds good
15:40:26 <rakhmerov> as usually, thanks for joining )
15:40:43 <rakhmerov> and happy Independence Day, us folks! :)
15:40:48 <toure> Thansk
15:40:50 <toure> thanks
15:40:51 <rakhmerov> .. US folks ..
15:40:54 <toure> :)
15:40:58 <rakhmerov> ok, bye
15:41:11 <rakhmerov> #endmeeting