16:00:49 #startmeeting Mistral 16:00:50 Meeting started Mon Oct 28 16:00:49 2013 UTC and is due to finish in 60 minutes. The chair is rakhmerov. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:00:51 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:00:53 The meeting name has been set to 'mistral' 16:01:08 hi 16:01:24 So let's review the action items from the previous week first 16:02:04 "rakhmerov, collect use cases and publish them on wiki" 16:02:28 it's done 16:02:44 #link https://wiki.openstack.org/wiki/Mistral#Use_cases 16:03:26 however, it now has a detailed description only for one use case: Cloud Cron 16:03:38 others are described really briefly 16:05:01 #info rakhmerov and stanlagun are working on detailed description of use cases: Long-running Business Process and Post Deployment Configuration (working name) 16:05:33 next item 16:05:49 "stanlagun (with rakhmerov's help) to finish Q&A document that covers use cases, relations with other frameworks, existing services, development plans etc" 16:05:52 done 16:06:06 #link https://wiki.openstack.org/wiki/Mistral#FAQ 16:07:20 this section will be populated periodically 16:07:28 next item 16:08:09 "stanlagun to describe the possible ways of executing particular actions on VMs in the tech specification" 16:08:27 this is in progress and, in fact, is a part of the other one 16:08:55 "Renat and Stan finish the modified specification and share it with the community" which is in progress too 16:10:08 "Renat, Stan, Timur to synchronize DSL and REST API with the latest conceptual changes" was assigned as an ongoing AI since the discussion keeps going on how it all should look like 16:10:29 any questions on that? 16:11:51 so basically, the team keeps working on DSL/API spec and it will take I believe around 1.5-2 weeks (due to the summit, otherwise it could be done earlier) 16:12:44 #action rakhmerov, stanlagun, tsufiev, keep working on DSL/API Mistral specification 16:13:30 On the summit we'll be able to share some insight with the community and get some feedback ) 16:13:46 exactly 16:14:05 we're hoping to clarify lots of things that may not be clear now for many people 16:14:11 rakhmerov: as for me, I currently almost completely occupied with upcoming Murano 0.4 release 16:14:25 tsufiev: sure, no problem 16:15:14 the fact that I mentioned you in AI doesn't mean you have to forget about everything else :) I would rather hope to get your input on some results 16:15:24 it should be valueable for us 16:16:26 at the summit we're also planning to talk with harlowja and his team (TaskFlow folks) and use this chance to sync our vision and most of the technical details 16:16:38 #topic Advanced use cases 16:17:18 I included this agenda item in case some people want to ask additional questions on project mission, use cases 16:17:39 if you have some, please fire away :) 16:18:38 rakhmerov: ok, I'm keeping an eye all your recent use-case proposals :) 16:18:40 I don't have any questions, but I've got an idea of a new good use-cases 16:18:46 rakhmerov: ok, I'm keeping an eye on all your recent use-case proposals :) 16:18:49 Like I said before, we're now intensively working on describing additional use cases and we're going to share this information soon in both mailing lists and wiki 16:19:00 Renat, I don't have advanced use cases out of top of my head but wanted to ask you about the areas war are going to prototype of beginning 16:19:18 Are we going to have some part of the meeting for that? Or you can describe this now? 16:19:47 we have an item "Start discussing development plans" so let's postpone it for a bit 16:20:11 ativelkov: what's your idea? Could you share pls? 16:20:54 #action rakhmerov, finish detailed description of "Long-running Business Process" use case 16:21:20 #action stanlagun, finish detailed description of "Post Deployment Configuration" use case 16:21:21 Last several days I was trying to build a murano workflow for advanced networking scenario: I had a network, a router, a subnet, several servers in it. I needed to connect all this stuff into a different network. 16:21:42 yep 16:21:53 This should be done in Heat, but it is unable to do it in one step 16:22:18 you first have to "unplug" the networks, then create a new configuration 16:22:20 what do you mean by "one step"? 16:22:41 Well, you cannot do a single stack-update command 16:22:50 gotcha 16:23:02 so, you have to first create a "temporary" heat stack 16:23:10 update an existing stack to this temporary state 16:23:19 and then update the temporary to the final 16:23:26 quite a tricky thing 16:23:36 yes, I see 16:23:39 especially if you have some optional steps and parts of the templates 16:23:49 do you think Mistral could be useful for that? 16:23:59 and this is not a bug in Heat or seomthing - that's "as designed" for Heat 16:24:12 yes, Mistral would be very appropriate for such scenarios 16:24:16 is it representable in a form of a task graph? 16:24:21 ok, cool 16:24:26 Yeah, I think so 16:25:11 could you prepare a couple of paragraphs containing what you've just told us but a little bit more detailed? 16:25:34 it's definitely not urgent but it may be included on our wiki some time later 16:26:08 I know you guys are really busy with other things but humbly hoping to get this info from you :) 16:26:31 Yes, I need to finish this feature in current Murano conductor - and this has to be done before the summit 16:26:40 ok 16:26:48 But I'll try to formalize it on the paper somewhere 16:27:41 #action ativelkov, collect the info about use case related to setting up a network (not a high priority) 16:28:09 so, let's move on 16:28:28 #topic What is left on documentation (use cases, DSL, API) 16:29:05 I think we've done some good job since the last meeting on writing docs 16:29:14 however, we still have a lot of things to do 16:29:42 complete use cases and complete the initial version of DSL/API specification 16:29:53 this is in our action items already 16:30:15 if you have other things to mention please let us know 16:30:46 I would say the crutial thing that is left is the technical spec: DSL/API 16:31:07 we'll get it done in a couple of weeks I believe, after the summit 16:31:17 *crucial 16:31:57 as far as the design and the implementation, it will be populated simultaneously with the corresponding work 16:32:12 which I hope will start pretty soon 16:32:45 anything else on that? 16:32:51 Do you update wiki or ether pads? 16:33:17 katyafervent: not sure what you mean :) 16:33:38 do ypu publish docs on openstack wiki? 16:33:44 *you 16:33:47 ooh, yes, sure 16:33:54 great) 16:34:06 #link https://wiki.openstack.org/wiki/Mistral 16:34:15 it already has lots of good information 16:35:15 I also added a back reference to Mistral from Convection proposal (harlowja was ok with that) so that people know it's started 16:36:10 #link https://wiki.openstack.org/wiki/Convection#PROPOSAL_ONLY:_TaskSystem-as-a-Service_.28Convection.29 16:36:34 may be we should emphasize it even more clearly 16:36:58 so, the next topic 16:37:07 #topic Start discussing development plans 16:38:17 so, my plan was to start sketching out a Roadmap this week 16:38:29 and I actually did start doing that today 16:38:51 but don't have too much to share yet, only some fragments 16:39:12 however, we have lots of ideas in mind that we need to formalize somewhere 16:40:03 first of all, I would suggest we finish DSL/API documentation and polish it out with other interested folks 16:40:33 it'll be the most important step towards the beginning of development 16:41:40 at the same time we have resources to start prototyping some of our ideas in parallel 16:42:11 this is NikolayM 16:42:31 Great, now we come back to the topic I was asking:) 16:42:38 yes, right :) 16:42:38 So what's the plan, what do we prototype first? 16:44:21 well, I would suggest we start with the basic things we'll need in ether case. Like very simple DSL grammar and the parser for it 16:44:49 Hm 16:44:55 I would suggest to start from the engine 16:44:58 in that regard we had an idea of automatic REST APIs clients generation and inclusion into DSL 16:46:02 ativelkov: yes, I'm telling now about the things that may become an interesting part of our spec but we're not sure about them yet since it requires some hands-on testing 16:46:42 the order of the things may be different 16:46:42 I mean, we may try testing some things without having DSL 16:46:59 by just defining the task flows manually 16:46:59 sure 16:47:21 what do you mean by "manually"? 16:47:40 Well, by defining tasks by python code in tests 16:47:47 or something like this 16:47:49 yes, sure 16:48:02 I mean DSL parser is easy to implement - you just need a complete spec 16:48:11 and you do not need to prototype it ) 16:48:34 hm.. I guess you didn't get what I told on 100% :) 16:48:44 while prototyping the engine (context passing and data flow, dependency analysis etc) may require some PoC 16:48:50 may be ) 16:49:08 I didn't mean to implement DSL fully 16:49:38 only some interesting ideas we have regarding it 16:50:19 just to be able to complete the spec 16:50:51 overall, I agree with you: we'll start with the engine and the model (tasks, dependencies, actions) as the first serious dev item 16:51:47 ok :) Thanks for making this clear 16:51:55 ok 16:52:22 but thanks for this note, it does make sense for sure 16:53:28 just wanna repeat: I have just a fragmentary vision on how we're going to approach our development, I just see some risky things I'd try before start the serious work 16:53:33 it's all up to discussion 16:53:57 and I would like to clarify that by the next meeting actually :) 16:54:06 It's fine, I think I've got the idea. Are there topics uncovered yet for today? 16:54:29 we have a vision sync up with Y! in 5 minutes, don't we& 16:54:31 ?* 16:54:35 #action rakhmerov, prepare the first version of Roadmap and the development plan 16:54:39 yes, we do 16:54:48 #topic Open discussion 16:54:53 anything else guys? 16:55:05 not from my side 16:55:59 ok 16:56:00 I'm also done 16:56:09 thanks to everyone! 16:56:20 see you all soon 16:56:25 #endmeeting