15:00:46 #startmeeting openstack search 15:00:46 Meeting started Thu Jun 23 15:00:46 2016 UTC and is due to finish in 60 minutes. The chair is TravT. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:47 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:50 The meeting name has been set to 'openstack_search' 15:00:55 o/ 15:00:56 o. 15:01:01 o/ 15:01:03 lol 15:01:06 o/ 15:01:09 0/ 15:01:10 o/ 15:01:13 o/ 15:01:39 give it another second or two for any more stragglers 15:01:40 o/ 15:02:01 morning 15:02:08 morning^2 15:02:27 morning / evening. 15:02:45 evening :D 15:02:46 so, yingjun messages me last night to let me know he is on vacation until the 27th 15:03:11 looks like we are missing lei-zh today as well 15:03:14 but let's get going. 15:03:20 https://etherpad.openstack.org/p/search-team-meeting-agenda 15:03:38 #topic ElasticSearch 2.x support updates 15:04:04 the patches that fix the functional tests for es 2.x have landed 15:04:30 sjmc7: i think we need to also backport them 15:04:38 ok, will do 15:05:06 RickA-HP: you've been looking at enabling ES 2 zuul tests 15:05:12 I have verified that the Jenkins tests (experimental) pass with the fixes. 15:05:43 As we were discussing in chat, I'm about to push the changes to infra to move them out of experimental. 15:06:25 so you are thinking py27 with es 1.x and es 2.x and then py34 with es 2.x? 15:06:26 The idea is that we don't want to test all combinations. Steve proposed enough tests to easily spot if an issue was due to Python 2.7 vs 3.4 or ES 1.7 vs 2.3 15:06:58 TravT: Yes, that is the proposal. It should cover everything and emphasis the versions that are main for Newton. 15:07:26 Any comments or objections? 15:07:29 it seems we can't yet eliminate 1.x from the mix. 15:07:43 last week, we decided we needed to get all of our stuff through before doing that 15:07:58 and we needed to figure out devstack support 15:08:12 sdague hasn't had any additional comments on https://review.openstack.org/#/c/320152/ 15:08:52 As long as we are testing Py27/Py34/ES1.7/ES2.3 the devstack support can be orthogonal. 15:09:39 yes, but we also were talking of eliminating 1.x from being officially supported. 15:09:49 i don't think we're ready to say that, yet. 15:10:07 no, probably not 15:10:19 We can say it, but we aren't ready to act on it! 15:10:36 so, from a functional testing perspective, i think the combinations make sense 15:11:47 Anyone else have a comment? 15:12:44 guess not. so, for now, we'll try to get the functional tests running on both and those additional es.2x backported to stable/mitaka. 15:12:52 next up 15:12:57 Ok, I'll make it so. 15:13:03 #topic searchlight ui plans for newton 15:13:41 I just wanted to start providing more updates / involvement on what's going for searchlight ui 15:14:11 matt-borland and tyr have been doing a lot of work in this area 15:14:29 they both have been working on upstream Horizon for nearly two years 15:15:03 and have been doing quite a few things in horizon that will enable improvement in searchlight ui 15:15:14 tyr doesn't look to be online yet 15:15:15 cool 15:15:32 so matt-borland, do you want to provide some overview? 15:15:43 sure thing! 15:15:47 Hello everybody, 15:16:07 so we've been approaching integration of searchlight into Horizon first by adding a plugin panel. 15:16:24 This panel is being developed in the repo: searchlight-ui (gerrit) 15:16:49 this panel is basically a one-stop shop for searching through your resource types, 15:17:05 and performing any actions/seeing any views that are relevant to the user. 15:17:23 The work so far has gone really well, and is accelerating. 15:17:43 We could frankly use some more eyes on the development that's being done, particularly in the searchligh-ui repo. :) 15:18:24 but I encourage you to install the panel in Horizon and try it out...it makes all the work you're doing really evident, in comparison with Horizon right now. :) 15:18:31 so thanks for all the work on the back-end!!! 15:18:49 there's a lot of information here, any basic questions? 15:18:54 thank matt-borland and tyr, really nice work 15:19:02 thanks david-lyle 15:19:21 so, as FYI. 15:19:31 +1 matt-borland 15:19:36 in horizon, there has been a lot of work on something called the resource type registry 15:19:46 Great work Matt! 15:19:46 OS::Nova::Server 15:19:53 ^ resource type 15:20:07 in that registry, we can register the actions you can take on it 15:20:18 and the views (e.g. details) 15:20:46 with SL, we are obviously using the notion of resource types, like Heat does 15:21:13 so, the support that will be building out in SL UI will be to be able to look up the resource types in the registry 15:21:22 find the actions you can do on each item 15:21:32 and specific views. 15:21:51 as additional actions are added to upstream horizon for core services (e.g. nova) 15:22:04 they should be available via lookup to the registry 15:22:19 so they'll become available to the search screen "auto-magically" 15:22:53 TravT: Where does the registry live? 15:23:05 in addition, we'd hope new plugins to horizon for other services will also write their actions and views in a way that is compatible with the registry 15:23:15 so as you add a plugin to horizon 15:23:28 you register its resource types, actions, views 15:23:48 and they'll become available across horizon and be available to searchlight 15:23:58 RickA-HP: the registry lives in horizon master 15:24:10 SL UI just uses it 15:24:17 It's part of the Angular application context 15:24:36 so any Angular features can access any of it (register or retrieve registrations) 15:25:02 matt actually did a "registry in 60 seconds video" 15:25:06 and put on youtube 15:25:09 do you have the matt? 15:25:12 * matt-borland looking.... 15:26:19 in horizon, you can see how images are registered 15:26:21 https://github.com/openstack/horizon/blob/master/openstack_dashboard/static/app/core/images/images.module.js#L50 15:26:29 ^ that is where the basic image info is registered 15:26:49 https://www.youtube.com/watch?v=nBpZebKSWiE is that 60-second overview 15:26:57 and this module registers the image actions: https://github.com/openstack/horizon/blob/master/openstack_dashboard/static/app/core/images/actions/actions.module.js#L41 15:28:06 matt-borland and tyr will be putting up more patches for SL to better take advantage of this work in upstream horizon. 15:28:06 My main goal today is to just advertise that we're working on this Horizon plugin, and to get you looking at it if you haven't already. 15:28:50 Our goals for Newton are to have at least full actions/views for Instances, Images, Volumes...however there is additional support for many more, and the ability to just use existing actions in any resource type. 15:28:56 so we may go faster than our expectation :) 15:29:20 thanks, matt. 15:29:25 any questions from anybody? 15:30:06 no questions, looks like nice work 15:30:09 nope. although i will try to fulfill my promise to look at some patches 15:30:11 thanks! 15:30:26 Thanks for the explanation TravT and nice work matt-borland 15:30:55 FYI, tyr who is here has also been heavily involved. 15:31:13 correction tyr is not here today. 15:31:26 okay. 15:31:30 thanks, next topic 15:31:45 #topic Nova API hit reduction 15:32:09 sjmc7 has done some nice work in trying to reduce API call backs for instances 15:32:35 I have sat down with a debugger multiple times watching the notifications flow from nova for various actions 15:32:53 and the outcomes all seem to be correct to me 15:33:05 https://review.openstack.org/#/c/317100/ 15:33:53 there are a couple of actions on instances that this patch doesn't cover, yet 15:34:18 it started to exceed its remit a little 15:34:23 but, i'd be okay with pushing it on through. 15:34:48 my only suggestion was that we put in a few more comments to specifically say what events are not covered by the code 15:34:53 like image shelving 15:35:15 sure 15:35:31 we are still hoping that once basic versioned notifications from nova land for instances 15:35:53 that they can then release a new version that includes more data so we don't have to do any callbacks 15:36:06 it looks like that probably won’t be in newton at this stage 15:36:10 yep 15:36:36 just the basic versioned notifications, which don't seem to provide any incremental value to us for newton. 15:36:54 sjmc7: this is nice work 15:37:21 thanks. the notifications especially on instance boot are a little excessive 15:38:18 well, with a couple comments, i'll add my +2 15:38:28 and if we can get a few more reviews, that'd be great! 15:38:56 That's all i have for formal topics 15:39:03 #topic open discussion 15:39:05 +1 sjmc7 15:40:12 oh 15:40:26 looks like release management is fixing our release model tag 15:40:27 https://review.openstack.org/#/c/333420/ 15:40:40 Well, I have finished off the Flavor Plugin; I was thinking of advancing myself more; since I have a better understanding of the code base now. 15:41:16 GB21, thanks for your continued dilligence on this! 15:41:49 Pleasure is all mine, TravT , I have more confidence now :D 15:42:16 do you have ideas on what you'd be interested in working on next? 15:42:49 as FYI, our bug and bp list is quite deep 15:42:52 Well, there are many things open in launchpad which are due for the Newton cycle. 15:42:54 yeah 15:43:11 so, maybe take a look through and see if there are things that interest you more than others. 15:43:18 I have had a chance to look at it; I was thinking of picking one up 15:44:00 Also, can I work for the Searchlight UI 15:44:07 as well 15:44:17 yeah, if you have an interest in SL UI, that'd be good 15:44:48 Kevin_Zheng: you had a BP up on server groups. 15:44:58 awsome; I know a little angular js and Django, but I can work my way out 15:45:02 have you had any chance to start that one? 15:45:09 little knowlege* 15:46:04 maybe Kevin_Zheng fell asleep. ;) 15:46:27 okay, any other topics for today? 15:46:34 nope 15:47:00 all right 15:47:02 thanks! 15:47:10 #endmeeting