15:01:32 <sjmc7> #startmeeting openstack search
15:01:33 <openstack> Meeting started Thu Jun  9 15:01:32 2016 UTC and is due to finish in 60 minutes.  The chair is sjmc7. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:01:35 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:01:37 <openstack> The meeting name has been set to 'openstack_search'
15:01:51 <sjmc7> morning all
15:02:00 <rosmaita> o/
15:02:07 <lei-zh> o/
15:02:24 <lei-zh> we are on holiday:)
15:02:35 <sjmc7> ah, good for you! we’ll keep this short :)
15:02:55 <sjmc7> travis is still on his eternal vacation
15:03:11 <sjmc7> so the very short agenda is https://etherpad.openstack.org/p/search-team-meeting-agenda
15:03:39 <sjmc7> give it one more minute in case there are stragglers
15:04:16 <sjmc7> ok…
15:04:18 <sjmc7> #topic Elasticsearch 1.x/2.x support
15:04:38 <sjmc7> we merged a change in CI to switch to es-2.x last week, which promply broke all tests
15:05:14 <sjmc7> we rolled back to 1.7 and i put up three reviews to fix the issues it uncovered, the last of which is https://review.openstack.org/#/c/325698/
15:05:32 <sjmc7> in doing those patches though, it became apparent we need to run CI against both versions if we support both
15:05:38 <sjmc7> as voting jobs
15:06:05 <sjmc7> so the question is whether we need to support both, or whether we can draw a line in the sand given that E-S 2.x will be more than a year old by Newton
15:06:33 <robcresswell> Got any numbers on adoption? Sorry, just hijacking your meeting sjmc7 :)
15:07:03 <sjmc7> for es-2?
15:07:11 <robcresswell> As in, E-S usage. That might help make an easier decision. I imagine people are using it for other things outside searchlight, and will stick with what they have/know, right?
15:07:26 <sjmc7> no, but we suspect it’s higher than for other similar (database like) systems because of the way it’s packaged and released
15:07:26 <robcresswell> yeah
15:07:36 <robcresswell> Ah, okay
15:07:46 <sjmc7> people don’t rely on debian-packaged releases
15:08:04 <sjmc7> elastic.co manage packages themselves and strongly encourage people to upgrade
15:08:12 <sjmc7> logstash etc support 2.x
15:08:43 <sjmc7> but yeah, i’ll make a note to ask folks that are interested in searchlight whether that would be a problem
15:08:46 <robcresswell> Given that you're pre-1.0, perhaps it makes sense to just do 1.0 on latest and greatest, in that case? If you release searchlight formally with 1.x support, that may haunt you.
15:09:08 <sjmc7> yeah. the changes are not that great, but some of the ones i found were pretty subtle
15:09:26 <sjmc7> ok. maybe a mailing list message first and then can make the call
15:10:17 <robcresswell> Sounds like a good shout
15:10:26 <sjmc7> but yeah, maybe since we are pre 1.0 this would be the opportunity to drop it
15:10:31 <rosmaita> robcresswell: +1
15:10:36 <rosmaita> sjmc7: +1
15:10:44 <sjmc7> +1s all round!
15:11:10 <sjmc7> ok, moving on, and out of order since rob’s here
15:11:14 <sjmc7> #topic UI reviews
15:11:42 <sjmc7> we’ve got a bit of an issue since none of the SL cores aside from travis know much angular :)
15:11:58 <sjmc7> i guess this must be affecting other services too though
15:12:15 <sjmc7> so typically are there horizon cores that moonlight on eg sahara or designate or whatever?
15:12:50 <robcresswell> There's a mix, often people drift away and work on other projects and their UIs.
15:13:13 <robcresswell> Right now we dont have enough cores to help the UIs unless they ask specifically for a review/help
15:13:32 <sjmc7> no, and i wouldn’t expcet that - that was the explicit reason for having projects move out of core
15:14:05 <sjmc7> the problem for me at least is that for the old django stuff it was no problem to review, but being a dinosaur i’ve struggled to keep up with the javascript rebellion
15:14:15 <robcresswell> Indeed. But if there's any difficulty or complexity, we're around to help. I'm interested in searchlight anyway, and my angular is adequate.
15:14:36 <sjmc7> you are too modest!
15:14:55 <robcresswell> haha
15:15:11 <sjmc7> ok, will maybe ask for a some quick input on a couple of patches at least until travis is back
15:15:30 <robcresswell> Man, that guy is just always on holiday
15:15:41 <sjmc7> yeah, it’s like he’s french or something
15:15:49 <robcresswell> hahaha
15:16:01 <sjmc7> ok, moving on
15:16:10 <sjmc7> #topic Remaining big features/changes
15:16:29 <sjmc7> last release we got in trouble with ourselves because we were making big changes up to milestone-3
15:16:35 <sjmc7> so we don’t want to do that again
15:16:53 <sjmc7> i only know of one big feature we’re looking at, and that’s the publishing stuff lei-zh has been pushing
15:17:24 <sjmc7> lei-zh: have you been able to spend any time prototyping that?
15:18:06 <lei-zh> I just start on doing it, perhaps still need some time
15:18:21 <sjmc7> sure. there’s no immediate rush
15:18:36 <lei-zh> I hope to put out a prototype soon
15:18:38 <sjmc7> n-2 isn’t for a while. if there’s any help we can give though, please ask
15:19:04 <lei-zh> maybe just clone a repo and do something on that
15:19:43 <sjmc7> ok. that was all i had for topics
15:19:51 <sjmc7> @topic Open discussion
15:19:53 <sjmc7> oops
15:19:56 <sjmc7> #topic Open discussion
15:20:20 <sjmc7> if anyone’s got anything they’re itching to bring up, you’ve got 40 minutes. otherwise will let folks get back to work / vacation
15:20:35 <lei-zh> so a question about the pipeline
15:20:42 <sjmc7> sure
15:21:02 <lei-zh> do we need to make transformers configurable
15:21:17 <sjmc7> transformers being the serialization?
15:22:04 <lei-zh> yes, give people freedom to configure serialization, and maybe other transformers
15:22:15 <lei-zh> I'm not sure what's the meaning of that
15:22:34 <sjmc7> what kind of configuration were you thinking?
15:22:58 <sjmc7> we haven’t done that already because it then creates problems for consumers; you have to know how the system’s configured in order to use it
15:23:32 <lei-zh> yes, that makes things complicated
15:23:54 <lei-zh> I just wonder if this senario useful
15:23:58 <sjmc7> were you hoping to add configuration?
15:24:48 <lei-zh> I'm not sure if we need this use case, like by configuring transformers, can produce different dataset from one kind of notification
15:25:28 <sjmc7> in a sense that’s why we’ve got our own listener rather than adding an output publisher to ceilometer
15:26:40 <sjmc7> i can’t see a use case right now, but that doesn’t mean there won’t be one. i would like it ultimately if the various service notificaitons instead became more standardized
15:27:38 <lei-zh> yes, we can gradually do that, first start from configuring publishers
15:28:32 <sjmc7> yeah, i think that might be the way to do it. if there’s a publisher for zaqar it can potentially change things, and the e-s publisher will add fields for role-based access
15:29:07 <lei-zh> that's a good idea
15:29:09 <sjmc7> IMO the data going into publishers needs to be a documented and static structure otherwise it makes it hard to know what you will get
15:30:15 <lei-zh> agree, that's what I'm towards : )
15:30:35 <lei-zh> ok, my question is done
15:30:48 <sjmc7> ok. anything else?
15:30:56 <sjmc7> seems like everyone’s on vacation today :)
15:31:11 <lei-zh> yes, it's chinese danwu festival
15:31:17 * david-lyle was late
15:31:19 <lei-zh> duanwu
15:31:30 <rosmaita> let's all vote ourselves raises and call it a day
15:31:35 <sjmc7> see me after class, Lyle
15:31:41 <lei-zh> haha
15:31:43 <sjmc7> haha, good plan!
15:31:44 <david-lyle> yes sire
15:31:47 <david-lyle> *sir
15:31:52 <sjmc7> hahaha
15:31:56 <sjmc7> sire is fine.
15:31:57 <david-lyle> little too much title there
15:32:11 <sjmc7> did you have anything to add, other than unearned titles?
15:32:26 <david-lyle> no, I'm useless
15:32:33 <sjmc7> and on that note...
15:32:38 <sjmc7> thanks all!
15:32:47 <RickA-HP> bye
15:32:53 <rosmaita> bye!
15:32:54 <lei-zh> bye
15:32:55 <sjmc7> #endmeeting