00:01:34 #startmeeting CongressTeamMeeting 00:01:35 Meeting started Thu Feb 11 00:01:34 2016 UTC and is due to finish in 60 minutes. The chair is thinrichs. Information about MeetBot at http://wiki.debian.org/MeetBot. 00:01:36 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 00:01:38 The meeting name has been set to 'congressteammeeting' 00:01:48 hi 00:02:50 masahito, pballand: courtesy ping 00:02:51 hi 00:03:06 Here's my agenda for today… 00:03:23 1. Austin summit 00:03:28 2. Mitaka timelines 00:03:34 3. Distributed arch updates 00:03:36 Anything else? 00:03:51 bryan_att: do you have anything specific for us? 00:04:16 I can give an update on testing and install on OPNFV, open issues etc 00:04:49 bryan_att: Sounds good. I'll call on you in a bit. 00:04:54 #topic Austin summit 00:05:19 masahito: would you like to tell us a little about the talk you proposed for Austin? 00:05:35 yes 00:05:54 I submitted a presentation about Congress. 00:06:25 Congress in NFV-based Mobile Cellular Network Fault Recovery 00:06:35 https://www.openstack.org/summit/austin-2016/vote-for-speakers/presentation/7199 00:07:26 feel free to add +3 on it :) 00:08:29 sorry, I have to leave here in a min. so please send me a mail if you have any question. 00:09:42 masahito: thanks! The talk looks interesting. I'm hoping it gets accepted! 00:10:09 No other announcements from me about Austin. Questions or comments? 00:10:40 #topic OPNFV update 00:10:47 bryan_att: want to give us an update? 00:11:11 OK, last time I gave an update was 12-17-15\ 00:11:35 Since then I have been focused on getting to a code freeze for the OPNFV B release, with a Congress installer support. 00:12:02 I didn't make the code freeze, but I will in a service release e.g. April 00:12:34 I still have the list of issues I noted in Dec to send in an email. That's my open item for this team. 00:12:50 Sounds like you're getting the installer problem under control. That's great! 00:12:57 But I do have now a fairly repeatable, scripted install (bash scripts). 00:13:15 I need to turn it into Ansible / Puppet etc. 00:13:46 No other specific update - just I am plugging away, and developing as a user. 00:14:02 Thanks for the update. Let us know if you need anything from us. 00:14:07 I will sne the issue list asap - believe me I haven't forgotten. 00:15:02 bryan_att: that's much appreciated. Thanks. 00:15:13 #topic Mitaka 00:15:37 The Mitaka3 deadline is coming up the first week of February. 00:15:48 That means feature freeze for the server, which we're basically already in. 00:16:07 It also means that the command-line client needs to be released in early Feb. 00:16:49 The CLI is released earlier than the servers so every project using another project's client can iron out problems that arise when the client changes. 00:17:32 The only project that I know of that's using Congress today is murano, so we may have some wiggle room there. 00:17:58 But let's try to get the client in shape by Feb 1. 00:18:22 In particular that means knowing whether we'll still be able to create/delete datasources. 00:18:53 As long as we don't decide to pull that feature out in Mitaka, we could probably release the client today and be fine. 00:18:57 Any thoughts? 00:19:47 thinrichs: client should wrk as is now with old architecture right 00:19:55 ramineni: right 00:20:12 It's only if we try to transition entirely to the new arch and we don't support create/delete datasources in the new arch. 00:20:20 that the client would need to change. 00:20:42 thinrichs: ya 00:21:04 In my mind, the plan is to keep create/delete datasources whether we're in the new arch or the old arch. 00:22:02 thinrichs: i think logic for add/delete datasources is in datasource manager right 00:22:02 If we do that, the client will work pretty much as is. 00:22:08 thinrichs: do you mean Mar 1 instead of Feb 1 for client release? 00:22:09 ramineni: right 00:22:23 March1, not Feb1. Right ekcs 00:22:30 thinrichs: a clarification - if you can't create datasources, is there a fixed list that are always available? 00:23:00 thinrichs: then i think we might need more changes , as currently it doesnt touch datasource-mgr in new arch 00:23:01 bryan_att: the idea was that you would control the datasources at configuration time, not at run-time. 00:23:19 OK, clear. 00:23:24 bryan_att: so if you wanted to restart the datasources, you'd need to restart the system. 00:23:32 ramineni: right they're in the datasource manager. 00:23:41 ramineni: ekcs has started moving that functionality to DSE2 00:24:10 thinrichs: oh, great 00:24:11 The idea would be that for Mitaka, we could use DSE2 but just put everything on a single node. 00:24:35 That would make creation/deletion of datasources work just like it does today. 00:24:51 thinrichs: ok 00:24:57 Then after Mitaka, we could work on having multiple DSENodes, and creating/deleting services on each o them. 00:25:27 Anyone see any problems with that plan? 00:25:43 thinrichs: no, sounds good 00:26:21 We've naturally fallen into the next topic… 00:26:25 #topic distributed arch 00:26:39 There's been a lot of activity on this since last week. 00:26:50 Which is great! 00:27:12 I took stock of where we are earlier today. 00:27:31 I went through the bug list and blueprint list and added a few bugs. 00:27:46 What I'm doing now is marking all the DSE2 issues as High importance 00:28:06 If it's something that we need to get done before mitaka3, I added mitaka3 as the milestone. 00:28:20 If it's something that we can wait on until after Mitaka, I didn't add the mitaka3 milestone. 00:28:58 Generally, we want to focus on functionality and correctness, and leave optimization and code cleanliness to later. 00:29:17 Questions on any of that? 00:30:10 Let's hear some status updates on what people have done, what they're stuck on if anything, and what they're hoping to do next. 00:30:25 ramineni: want to start? 00:30:31 thinrichs: sure 00:31:26 thinrichs: iḿ done with migrating api models , and exceptions patch is in progress , once that done i have one more bug assigned of migrating test_congress to dse2 00:32:05 thinrichs: for exceptions patch, i have got all services registered on peers from control bus 00:32:38 ramineni: cool! 00:33:16 Are you going to add a method to DataService that encapsulates that? Or is it not generally useful? 00:33:19 thinrichs: ill raise a dependent patch to enable all the tests already merged 00:33:36 ramineni: sounds good 00:33:53 ramineni: nice job on the api models. Very clean. Example… 00:34:04 https://github.com/openstack/congress/blob/master/congress/api/status_model.py 00:34:36 thinrichs: :) thanks 00:35:10 thinrichs: currently we have a function in dse node to get the dse status 00:35:18 I'm moving the couple of API models I worked on over to your new basecalss. 00:35:24 thinrichs: but that doesn include services running on current node 00:36:30 https://github.com/openstack/congress/blob/master/congress/dse2/control_bus.py#L127 00:36:54 ramineni: I think there were several places we wanted to know if a dataservice existed, so having that functionality as part of the DSE2 makes sense, I'd say. 00:37:21 thinrichs: yes, but should we status of current node too, to get all the data at one place? 00:37:36 should we add** 00:38:04 I think the operation I saw in the code was: "does service X exist"? 00:38:35 But yes I would think we should have an abstraction that combines local and remote services. 00:39:11 ramineni: anything else? 00:39:36 thinrichs: no, migrting test_congress is pending from my side 00:39:51 ramineni: sounds good. 00:39:56 ekcs: want to go next? 00:40:13 thinrichs: sure. 00:41:39 I made a minimal change to managers/datasource to get it working on deepsix2. I also looked at whether we can rely on migrating datasource_model instead, but i’m not I totally understand the vision. 00:43:56 I think what we can do is move the creation/deletion of datasources out of the datasource manager and into the DseNode class. 00:44:26 I'd think the rest of the functionality in the datasource manager is obsolete with the new API-models. 00:45:10 ekcs: ya, migrating to dsenode would be better 00:45:38 #link https://github.com/openstack/congress/blob/master/congress/managers/datasource.py 00:45:59 I'm perusing the code now… 00:46:55 The other way to look at it is that if we look at the API datasource_model, we basically need to migrate that functionality into the DseNode 00:48:51 After perusing, it looks like most of the functions inside the datasource manager are used for creation/deletion. 00:49:18 Some definitely aren't, but most seem to be. 00:50:26 ekcs: maybe your existing datasource manager migration makes sense for now. 00:50:36 We'll want to move that code out of the manager and into the DSeNode eventually, but that's more cosmeetic than functional. 00:50:47 ekcs: what do you think? 00:51:31 thinrichs: I think either way is fine. Maybe I don’t understand the implications enough to have a strong opinion either way. 00:51:43 Anyone else have opinions? 00:52:17 thinrichs: but how do we invoke the functionalities? 00:52:33 thinrichs: rpc is not possible 00:53:42 ramineni: for now the API could send an RPC call to the DseNode it is running in (or just have a python reference to the DseNode) 00:54:01 Remember that the API will be deployed as a service inside the 1 DseNode. 00:54:23 But you're right that we'll need to work that out to figure out the best way. 00:55:38 5 minutes left. 00:55:51 I’m not following. Will need to catch up in my understanding. 00:56:51 ekcs: okay. Let's move ahead with the migration you already did for the datasource manager. I'll dig into it a little to see if there's anything left to do. 00:56:58 #topic open discussion 00:57:00 Before we run out of time, we have a newcomer. 00:57:15 tsandall: could you introduce yourself? 00:57:25 thinrichs: thanks! 00:58:25 hey all, name is Torin Sandall. I started looking into congress last summer after the Vancouver summit. 00:58:40 I'm excited about making more contributions to congress in the future! 00:59:07 ...and as of recently, I'm working at Styra, Inc. 00:59:22 that's all for now :) 00:59:26 tsandall: awesomeness. 00:59:45 tsandall: great, welcome :) 00:59:48 tsandall: Welcome (officially)! 01:00:18 And we're out of time. Let's keep pushing to hit that mitaka3 deadline! 01:00:22 #endmeeting