15:02:08 #startmeeting openstack search 15:02:09 Meeting started Thu Feb 11 15:02:08 2016 UTC and is due to finish in 60 minutes. The chair is TravT. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:02:10 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:02:12 The meeting name has been set to 'openstack_search' 15:02:25 sjmc7: looks like everyone ganged up on you last night and -1'd your spec patch 15:02:30 haha 15:02:36 i had to jump on the band wagon 15:02:42 hey 15:02:49 o/ 15:02:54 o/ 15:02:55 rosmaita: it was heartbreaking 15:03:12 o/ 15:03:37 okey-dokey. 15:03:49 i didn't put together a big agenda today. 15:04:11 i did have something which i now can’t remember. perhaps my mind will be jogged 15:04:20 because i thought we should look over the current release and talk some priorities 15:04:52 as usual, it is here though: https://etherpad.openstack.org/p/search-team-meeting-agenda 15:05:08 so you can add or bring it up during the meeting 15:05:44 General updates is that we are closing in on the end of mitaka 3 which you all already knew 15:06:24 I did just remember one item for discussion and added to agenda 15:06:38 #topic searchlight panel 15:07:18 Interestingly enough, I've been contacted twice in the last two weeks from people wanting to install searchlight and the horizon panel. 15:07:34 last one appeared to be somebody wanting it for liberty 15:08:03 which left us in a not great spot, because it obviously wasn't merged to master 15:08:08 and it wasn't its own repo 15:08:28 oops 15:08:39 it exists as a patch on horizon. 15:09:22 so, i'm still very much trying to get it in shape for horizon mitaka, but the reality is that its a longshot to land on horizon master for mitaka. 15:09:55 i still think it may be best to be on horizon master, but if needed we should do a searchlight-dashboard / searchlight-ui repo 15:10:09 so we have something we can tag with mitaka and be deployable. 15:10:58 my question is this: should i proactively put up a patch for a searchlight-dashboard repo now? 15:11:17 TravT: given horizon’s been moving stuff out rather than in, i think that’d be a good idea for now 15:11:25 until we have a good case for making it a core piece 15:11:46 ultimately i think we have a good case for making it core and many horizon cores agree. 15:11:52 yeah, i agree 15:12:16 but i’ve heard from many they want to see it have broader use before that happens 15:12:29 i think that if we miss horizon master in mitaka, that it'll be tough to get a repo created in time for mitaka tagging 15:13:11 but i still have to keep it as a patch on horizon master until its parent patch lands on master. 15:13:37 having separate repos for horizon is a well-established pattern, so i don’t see a problem 15:13:42 ultimately it’d be much nicer to be in core 15:13:53 okay, so i'm going to request a new repo so we at least have it 15:14:19 and we might swap it over from being a patch on master to the repo right at the end of mitaka. 15:14:50 okay? rosmaita nikhil RickA-HP? 15:15:06 TravT: I agree with creating the new repo now. 15:15:27 good with me. and im sure with david :) 15:15:40 It will put us on the path to success if we don't make Horizon core. 15:15:44 yeah, david and i had a long conversation about it. 15:16:08 he likes it for horizon core, but thought a separate repo as backup / interim plan makes sense 15:16:31 i don't have an opinion on this, don't know much about horizon politics 15:16:40 but the discussion so far makes sense to me 15:17:12 horizon team has generally been supportive, but the priority backlog for horizon is pretty big with several things higher on the list 15:18:20 with a dashboard repo, it has become standard to make the service core team and the horizon core team cores on the repo. 15:18:41 in some cases, additional people are added as cores. 15:18:49 for example, magnum did that 15:19:17 so initially, i'd make searchlight core and horizon core team have rights. 15:19:36 == rosmaita 15:20:08 okay, i'll work on getting that going over the next several days 15:20:28 #agreed Create a searchlight-dashboard repo as a backup plan to searchlight landing on mitaka horizon master 15:20:38 i think that's how it works... i don't know 15:20:59 #topic Newton summit session 15:21:19 sjmc7 and I submitted a project update session to the summit. 15:21:43 * TravT needs to find a link 15:22:18 sjmc7 do you have it? 15:22:24 looking 15:22:37 talki amongst yourselves 15:22:54 https://www.openstack.org/summit/austin-2016/vote-for-speakers/presentation/7561 15:23:17 ok, anyway, if you are so inclined, voting for it might help it 15:23:32 and hopefully we won’t have to bellow over the marketplace this time 15:24:34 #topic blueprints - end of mitaka priorities 15:24:52 https://blueprints.launchpad.net/searchlight 15:25:34 sjmc7, do we have anything left on index-level-role-separation 15:25:59 there’s a bug against the designate plugins around it with a patch up, but the core code is done 15:26:23 ok, then we'll leave it not marked as complete 15:26:27 aww :( 15:26:37 so close! 15:27:27 the deletion-journal stuff i think will be much simpler after digging up the barely documented feature elasticsearch has (and again, i think rosmaita tried to bring this up and i didn’t pay enough attention, so apologies) 15:28:07 +1 to rosmaita 15:28:31 sjmc7: i think you can mark es-deletion-journal as something besides unknown delivery now, right? 15:28:36 yes 15:29:08 i can’t edit it :( 15:29:13 ah, there it is 15:29:31 RickA-HP: i just marked zero downtime indexing as approved (since spec went through) 15:29:42 can you change delivery to appropriate status? 15:30:04 TravT: Sure. As an update, the code is written. 15:30:05 per-resource-type-control i think will not be tremendously difficult now the role separation is in 15:30:12 i’m saving it for a rainy afternoon 15:30:22 I'm in the middle of testing and I still need to write more unit tests. 15:30:28 yes, i think we will need it in mitaka 15:30:48 when i discussed searchlight in the horizon irc meeting yesterday, "permissions" was one of the questions raised 15:31:09 ok. i’ll maybe talk to you about specifics 15:31:18 richard asked if we'd addressed the permissions issues we talked about in tokyo. 15:31:19 i don’t want to get too deep into a complicated hole 15:31:33 sure, my brain isn't yet firing on all cylinders this morning 15:31:49 so later in the day would be better 15:31:52 ;) 15:32:25 steve, i think this bp is actually done. it appears to overlap with role base documents spec https://blueprints.launchpad.net/searchlight/+spec/nova-server-config-admin-properties 15:32:33 woo-hoo! 15:32:41 i can mark it superseded if you agree 15:32:46 yes 15:33:01 freebie blueprint! 15:33:10 woot! 15:34:16 ok, so other than pre resource type policy control, i contend that we really need cinder and neutron plugins for mitaka. 15:34:29 we need to fill out the core openstack services. 15:34:39 i looked at cinder yesterday 15:34:54 and i think the existing notifications are sufficient for something without hitting the API all the time 15:35:05 neutron i’m waiting for the parent/child patch to land 15:35:21 okay, i saw lakshmiS added comments on that one last night 15:35:36 yep. the data model is more complicated for neutron 15:36:08 yeah, i've been trying to psych myself up to learn it enough to provide some review on it 15:36:11 but i think i can put up a first stab for discussion 15:36:59 if we get a cinder plugin patch up 15:37:07 maybe duncan could still get the notification side up 15:37:11 on cinder 15:37:31 yep. i’m intending to work on both today or tomorrow 15:38:03 the versioning patch is probably the most crucial one now 15:38:19 it has the potential to introduce unforseen bugs 15:38:29 so it’d be good to get as much testing time as possible 15:39:04 well, looks like just waiting for another patch set from lei 15:39:15 https://review.openstack.org/#/c/255751/ 15:39:48 yeah, it’s gone back and forth a bit. timezones make it more difficult 15:40:30 nikhil, have you had any time to look at zuul functional tests? 15:42:33 maybe he stepped away 15:43:49 i still would like to see swift get closer to reality, but that's probably something to look to get worked out for the summit... maybe we should have put up a session on searching swift... 15:44:10 * TravT :( just realized we probably did miss the boat on that one... 15:44:27 maybe the IBM guys will be doing it again :) 15:44:36 we really should have done that! 15:44:54 we'll have to propose it as a design session with swift perhaps 15:45:07 walk in with working code. 15:45:32 anyway, i don't have any other big topics for the week 15:45:39 #topic open discussion 15:45:41 i remembered something i had 15:46:30 i am still worried about hitting the nova API. i wondered about not listening to state change notificaitons, or at least making it optional 15:46:47 power events spawn at least 4 notifications 15:47:07 the transitionary ones don’t seem that crucial for us to pick up 15:47:13 sjmc7: Are we able to filter out notifications? 15:47:20 oh, actually, we talked about putting in some smarter logic there as well... 15:47:28 RickA-HP: yeah, we can choose which ones 15:47:39 i’m thinking a config option 15:47:44 maybe we should only react to final state 15:47:55 i have a strong concern that on a big install, hammering the API on state change is gonna be bad news 15:48:02 yes - config option would be good. 15:48:12 if we can get more info into the notifications next cycle we can do better 15:48:25 ok. i’ll put up a BP. i can’t remember the other thing 15:49:11 i haven't gone back to see if nova actually implemented any of that notification discussion that i attended at the summit 15:49:33 i haven’t looked for a while. i’ll take another look 15:49:38 if they did, then they were going to put full data into notification, but it would just be the raw data 15:49:45 not looking like the API output 15:49:53 but we could transform it 15:50:10 yep 15:50:21 i’l ltake a look. something for the summit, definitely 15:50:28 I had asked about them giving a transform library and they were basically like "oh man, no way, not how tangled up the code is" 15:50:44 :D 15:51:14 if they did that, it would be pretty awesome... 15:51:31 okay, anything else for today? 15:51:54 i think we'll have yingjun back next week. 15:51:56 nope. reviews on the versioning patch 15:51:58 he was out this week on holiday 15:52:26 all right, thanks! 15:52:51 #endmeeting