15:08:38 #startmeeting zaqar 15:08:39 Meeting started Mon Jan 18 15:08:38 2016 UTC and is due to finish in 60 minutes. The chair is vkmc. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:08:40 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:08:43 The meeting name has been set to 'zaqar' 15:08:50 #topic roll-call 15:08:54 o/ 15:08:57 hi guys :) 15:09:05 hi 15:09:11 since flwang1 is not around to drive the meeting, I think I can give it a try 15:09:45 so let's check the agenda... 15:09:46 what we will discuss today? 15:10:19 since we are about to cut Mitaka-2, I'd like to focus on the changes that we consider important to get merged now 15:10:38 #link https://launchpad.net/zaqar/+milestone/mitaka-2 15:10:52 there is no many triaged features/bug fixes there, we need to fix that 15:11:05 #action vkmc triage bugs/features for Mitaka-2 15:11:07 I'd really like to get https://review.openstack.org/#/c/253731/ in for M2 15:11:13 ryansb++ 15:11:24 #topic mitaka-2 15:11:29 ryansb, Client is released independently 15:11:36 So I don't think it matters much 15:11:38 ah, well that makes it easy 15:12:13 therve: I think I'll take that "bug" we were talking about last week (that Queues need a .subscribe method in the Python lib) 15:12:13 checking the server side, I see many patches Eva-i submitted 15:12:32 those are already there for review 15:12:48 we need to check the bug list to see if we should get something else in 15:13:00 like that one... "queues need a subscribe method" 15:13:15 vkmc, Client as well 15:13:20 yup 15:13:31 My websocket binary support patch needs to be released in m-2, but now I think it needs further improvement. 15:13:48 I opened many bugs recently. 15:14:28 so... 15:14:52 #action check bugs both server-side and client-side for m-2 15:15:15 vkmc, Again, client doesn't matter for m2 :) 15:15:23 AFAIK, at least 15:15:55 https://bugs.launchpad.net/zaqar/+bug/1533668 check out this bug for example 15:15:56 Launchpad bug 1533668 in zaqar "Zaqar can't list queues, when enable_virtual_pool = True (affects devstack)" [Undecided,New] 15:15:56 right 15:19:01 If there was Fei Long, he would probably asked has anyone checked out zaqar-ui https://review.openstack.org/#/c/255213/ 15:19:41 yes 15:19:49 I'm looking at previous meetings logs 15:20:18 let's jump to that 15:20:21 #topic zaqar-ui 15:21:43 this one got recently merged 15:21:45 #link https://review.openstack.org/#/c/255213/ 15:21:52 and there are also other patch sets waiting for review 15:22:08 this is a wip but maybe flwang1 could use some help testing this 15:22:09 #link https://review.openstack.org/#/c/258474/3 15:22:36 and that's pretty much it on that side... flwang1 probably can update us better next time we chat with him :) 15:22:48 I will try new zaqar-ui patches today. But it's hard for me to understand how it works. 15:22:58 vkmc: yes 15:23:03 how to enable the plugin you say? 15:23:06 I need to test that too 15:23:18 I'll ping you if I can make some time to test it and we can do that together 15:23:20 I know how to enable the plugin, it's in readme.md 15:23:35 oh cool 15:23:40 so what's your concern? 15:24:27 I just don't know how the code is getting glued together. 15:24:40 I just need to spend some time understanding it. 15:26:28 ok ok 15:26:35 we can check that later eitherways 15:26:35 We can move to another topic now 15:26:39 I want to discuss one thing about websocket transport on today's meeting. 15:27:21 k 15:28:20 yes 15:28:24 let's move to that then 15:28:34 #topic binary support for the websocket transport 15:28:43 Eva-i, floor is yours 15:30:54 When the client is communicating with Zaqar through http requests, it always sends requests to API version-specific zaqar endpoint ("v2" for example), so Zaqar knows how to process this request depending on the API. 15:33:12 When the websocket transport is used, there's no way the client can tell Zaqar which version of websocket API it wants to use. Though zaqar has only one websocket API version now. 15:34:28 Eva-i, It only supports v2. There is no way to choose an API just yet 15:34:41 Yeah, but *one day* we'll need to version it 15:35:23 But I'd like the client able to specify the version as it can help us in future if we will release the new version of websocket API. 15:35:49 Is it too early to think about it? 15:35:59 is there a reason they can't be locked to the version with which they initiated the websocket connection? 15:37:47 ryansb, We'd need to have routing in the websocket side, which we don't 15:38:01 But that seems like the easiest option yeah 15:38:07 good point. 15:38:44 you specify the version when you instantiate the api 15:38:53 right now v2 is the only one 15:39:11 but for newer versions we can inherit from that and add the required changes, and you are done 15:39:14 or what am I missing? 15:43:25 Eva-i, ^? 15:43:59 I think I should have prepared more for this conversation 15:44:32 I'm just worrying about future compatibility 15:44:48 Eva-i: I think for right now, what you're doing is fine 15:45:01 Maybe I'll bring this topic up when API v2 will be stable 15:45:44 agreeing that websocket connections are going to stick to the version they were initiated with in the HTTP API is probably the way forward 15:48:03 +1 15:48:04 sounds good 15:48:12 let's talk more about this on the design session 15:48:24 since this something that will probably happen on the next cycle 15:48:28 agree? 15:48:50 There's http handshake during websocket connection initiation, but Zaqar can't control it AFAIK 15:49:30 yeah no, that's implemented on the websocket library 15:50:03 The client don't have a way to know websocket api version or a way to tell it to Zaqar 15:50:09 If we don't do some things right before making websocket API v2 stable, the future versioning for websocket transport might be a bit problematic. 15:50:43 But okay, let's move to another topic 15:50:57 we are almost close to the end of our meeting 15:51:08 so let's move things to open discussion 15:51:11 #topic open discussion 15:51:52 Eva-i, I understand your concerns, let's work on designing how we can handle that 15:55:28 BTW, I'm bringing back Zaqar installation guide in production environment https://review.openstack.org/#/c/268851/. I'd like your help figuring out which things might are outdated in this guide. 16:01:32 sounds good, thanks Eva-i 16:01:35 we are over time 16:01:38 let's call this a meeting 16:01:41 thanks all 16:01:43 #endmeeting