15:02:23 <TravT> #startmeeting openstack search
15:02:24 <openstack> Meeting started Thu Apr 14 15:02:23 2016 UTC and is due to finish in 60 minutes.  The chair is TravT. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:02:25 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:02:27 <openstack> The meeting name has been set to 'openstack_search'
15:02:46 <TravT> o/
15:02:58 <RickA-HP> o/
15:03:07 <yingjun> o/
15:03:11 <lei-zh> o/
15:03:19 <lakshmiS> o/
15:03:36 <rosmaita> o/
15:03:57 <TravT> okay. i think sjmc7 is around too.
15:04:00 <sjmc7> morning
15:04:07 <TravT> https://etherpad.openstack.org/p/search-team-meeting-agenda
15:04:20 <TravT> #topic Backport 0.2.1
15:04:35 <TravT> The items we talked about last week are in
15:04:46 <TravT> https://review.openstack.org/#/q/status:merged+project:openstack/searchlight+branch:stable/mitaka
15:04:54 <TravT> everything on there since Apr 8
15:05:24 <TravT> yingjun: submited a fix this morning that i'd consider as well
15:05:29 <TravT> if we agree on handling
15:05:31 <TravT> https://review.openstack.org/#/c/305627/
15:06:26 <sjmc7> i’m not sure thiose are the only fields with disagreeing mappings
15:06:59 <TravT> i'm not either...
15:07:15 <TravT> but i'd like to backport mappings as necessary
15:07:23 <TravT> because if somebody does deploy
15:07:32 <TravT> they'll have to re-index
15:07:37 <TravT> to pick up mapping changes
15:07:51 <TravT> so, better to get those backported, i think.
15:08:10 <sjmc7> yeah, i guess i was saying if we’re going to do that we should check all the mappings rather than doing it piecemeal
15:08:35 <TravT> +1 that
15:08:42 <TravT> so...
15:08:48 <sjmc7> i’ll run the es 2 migration tool on an index today
15:09:19 <sjmc7> there’s another related bug - https://bugs.launchpad.net/searchlight/+bug/1532010
15:09:20 <openstack> Launchpad bug 1532010 in OpenStack Search (Searchlight) "Identically named fields with different definitions not allowed in elasticsearch 2.0" [Undecided,New]
15:09:44 <TravT> hmmm
15:10:01 <sjmc7> i’ll run the migration tool after this to get the list of fields now we’ve added a load of stuff
15:10:11 <TravT> ok, that'd be very helpful.
15:10:14 <sjmc7> and it really should be a test job
15:10:26 <TravT> interesting thing about that...
15:10:31 <lakshmiS> well removing doc type will be a major change
15:10:39 <sjmc7> doc type isn’t being removewd
15:10:59 <lakshmiS> isnt 2.0 doc type is not suggested
15:10:59 <TravT> is a ramification might end up being that some types have to go in different resource groups (once that patch lands to re-enable it)
15:11:42 <TravT> we can try to keep field mapping in sync for as long as possible.
15:11:50 <sjmc7> lakshmiS: doc types still exist in 2.0, but they’re enforcing rules more strictly
15:12:11 <sjmc7> TravT: yeah, if we really can’t make two fields agree they’d need renaming or a separate group
15:12:36 <sjmc7> though i guess i’d prefer to map the field twice and store it both ways
15:13:06 <TravT> like raw strings?
15:13:25 <sjmc7> yeah. anyway, i’ll see how many fields are affected
15:13:33 <TravT> Okay, cool
15:13:35 <TravT> thanks sjmc7
15:13:46 <TravT> based on that, will decide whether or not to hold up 0.2.1 tag
15:14:00 <RickA-HP> sjmc7: Storing it twice, wil that complicate the queries?
15:14:38 <sjmc7> i don’t know
15:15:28 <TravT> for things like volume size...
15:15:43 <TravT> for now changing that to long is fine with me
15:15:47 <TravT> and let the buyer beware
15:16:11 <TravT> on horizon side now, we apply filters to the output to show correct unit
15:16:31 <TravT> but possibly we could have an enhancement to provide a normalized field size too.
15:16:36 <TravT> (new field)
15:16:45 <TravT> but that would need to be separate patch, i think
15:17:44 <TravT> i mean using https://www.elastic.co/guide/en/elasticsearch/reference/current/_multi_fields.html
15:17:47 <TravT> do something like
15:18:15 <TravT> have long for the base type def, then have a .mb .gb field
15:18:58 <TravT> anyway... let's take that discussion to the patch
15:19:33 <TravT> #topic Summit Finalization
15:20:05 <TravT> I've filled out our fishbowl and 1st working session
15:20:11 <TravT> #link https://www.openstack.org/summit/austin-2016/summit-schedule/#day=2016-04-28&summit_types=2&tags=1728
15:20:34 <TravT> fishbowl is the whole notifications talk
15:20:46 <TravT> and i've tagged several other projects to see if we get participation
15:20:57 <TravT> 1st working session is for cross region
15:21:38 <TravT> of note is that horizon has a session where searchlight has been tagged
15:21:57 <sjmc7> yeah, that’s good
15:22:03 <TravT> for the 2nd searchlight session
15:22:18 <TravT> we have a few options
15:22:19 <TravT> https://etherpad.openstack.org/p/searchlight-newton-summit
15:22:49 <TravT> I'm thinking of splitting it and just seeing if we get to more than the first topic
15:23:26 <TravT> First topic would be the pipeline architecture if we get representation for lei-zh's team in attendance
15:23:27 <sjmc7> yeah, i think last time we went over a bunch of stuff
15:23:36 <TravT> second topic would be priorities review
15:23:37 <sjmc7> in our sweaty little room
15:23:44 <TravT> yes, that was quite packed last time
15:23:45 <rosmaita> TravT: the first work session is up against a nova/glance session on moving to images v2
15:23:58 <rosmaita> i had to sit on the floor
15:24:06 <sjmc7> you can sit on a chair this time
15:24:10 <TravT> rosmaita: oh...
15:24:16 <lei-zh> ok, I'll double check with yuntong
15:24:29 <TravT> so it seems like you'd be very important for cross region
15:24:35 <TravT> so i could move that to the second session
15:24:38 * rosmaita blushes
15:24:42 <sjmc7> :D
15:24:42 <rosmaita> that would be cool
15:24:49 <TravT> okay, i'll do that
15:25:08 <sjmc7> be good to grab brad and david for that one, too
15:25:10 <sjmc7> and maybe eric
15:25:17 <TravT> yeah, definitely
15:25:20 <rosmaita> just checked, that 2nd time doesn't conflict with any glance stuff
15:25:32 <sjmc7> ok. does it conflict with horizon?
15:25:45 <TravT> it shouldn't
15:25:48 <TravT> but let me check
15:25:49 <sjmc7> i guess we can figure it out
15:26:11 <sjmc7> thanks for pointing that out, rosmaita. timing clashes are the theme of summits :)
15:26:14 <david-lyle> IIRC horizon and searchlight had no conflicts
15:26:30 <sjmc7> splendid
15:27:01 <TravT> here's horizon schedule https://www.openstack.org/summit/austin-2016/summit-schedule/global-search?t=horizon%3A
15:27:13 <TravT> no conflicts
15:27:44 <TravT> okay, so we'll make that change
15:28:03 <rosmaita> thanks!
15:28:08 <nikhil> o/
15:28:09 <TravT> and i'll make the first working session be the pot of stew
15:28:10 <nikhil> quick ques
15:28:16 <TravT> we'll throw whatever we can in
15:28:19 <TravT> nikhil: shoot
15:28:31 <rosmaita> i was supposed to bring pie, can't remember why now
15:28:33 <nikhil> * should I tag SL against Glare session? (trying to see if there was interest)
15:28:54 <TravT> that was the artifact and app catalog right?
15:29:02 <nikhil> or people can choose to sync with app-catalog folks off-the-band too
15:29:06 <nikhil> TravT: yeah
15:29:31 <TravT> seems like there will be some discussion of interest there, so i'm okay if you tag it
15:29:38 <nikhil> this one https://www.openstack.org/summit/austin-2016/summit-schedule/events/9162?goback=1
15:29:53 <nikhil> ok, sounds good. will do that then.
15:30:22 <nikhil> done
15:30:28 <nikhil> and done (tagging)
15:30:30 <nikhil> thanks
15:30:36 <TravT> oh boy, that one does conflict with a horizon topic of particular interest as well
15:30:39 <TravT> horizon scaling
15:31:14 <sjmc7> :)
15:31:19 <nikhil> oopsie
15:31:31 <TravT> well we'll try to figure it out.
15:31:40 <nikhil> well I hope rosmaita can be a liaison then ? :)
15:31:56 <rosmaita> for which one?
15:31:56 <TravT> yes. he can also bring the pie
15:32:15 <nikhil> rosmaita: for Glare and app catalog interactions and the need for SL
15:32:25 <rosmaita> sure
15:32:27 <nikhil> thx
15:32:29 <TravT> thx rosmaita
15:32:52 <rosmaita> actually, i better check
15:33:52 <rosmaita> ok, that's already on my calendar ... duh! so just to get this straight, i will represent searchlight at the glare/app catalog glance working session?
15:34:09 <TravT> that's our hope!
15:34:23 <rosmaita> ok, got it
15:34:27 <rosmaita> but no pie for that session
15:34:32 <rosmaita> i only bring pie to searchlight
15:34:39 <sjmc7> woo!
15:34:43 <TravT> sweet
15:34:46 <TravT> maybe yingjun can also make that one.
15:34:50 * rosmaita wonders how to get pie through the TSA
15:34:58 <TravT> hide a knife in it
15:35:06 <nikhil> heh
15:35:17 <rosmaita> i can use the knife as a distraction
15:35:19 <nikhil> I think you can carry a swiss knife these days (but not entirely sure)
15:35:51 <rosmaita> yes, but can you carry a pie?
15:36:06 <nikhil> yeah as long as it's not too soggy
15:36:19 <rosmaita> it can only be 6 ounces
15:36:19 <TravT> well, i'm thinking that if you put a knife in it, it will become invisible on their radiation screens
15:36:22 <nikhil> the liquid part needs to be <<< crust
15:36:36 <TravT> so you'll slip straight through
15:36:40 <nikhil> :)
15:36:51 <yingjun> TravT which one?
15:36:57 <rosmaita> i don;'t know, didn't they catch a guy the other day with a pot grinder shaped like a hand grenade?
15:36:58 <TravT> pumpkin
15:37:20 <TravT> yingjun: https://www.openstack.org/summit/austin-2016/summit-schedule/events/9162?goback=1
15:37:25 <rosmaita> i guess we are getting off topic here
15:37:50 <TravT> basically app catalog is considering using a glance API called glare and they are also talking about indexing into elastic search
15:38:13 <TravT> so there might come up some discussion about using searchlight for their ES needs
15:38:28 <rosmaita> yingjun: it would be good for you to make it, as someone who would actually know something about searchlight
15:39:18 <rosmaita> i will go too, but i don't have a lot of depth
15:39:28 <yingjun> TravT, thanks, seems interesting, i’ll try to make that
15:39:33 <TravT> okay, great!
15:40:12 <TravT> okay, so only other thing is if we want to try to get together for lunch or dinner some day?
15:40:16 <TravT> summit is always hectic
15:40:43 <rosmaita> i want to go to this on tuesday evening: http://www.antonesnightclub.com/event/1125203-greg-izor-box-kickers-austin/
15:41:10 <rosmaita> don't think there's anything "official" yet on tuesday evening
15:41:20 <TravT> that looks interesting.
15:41:39 <rosmaita> doors for the above open at 9 and tickets are only $5, so i think you have to show up early
15:42:04 <rosmaita> will talk to a guy i know in austin who's familiar with the music scene to find out
15:42:13 <TravT> okay, that'd be great
15:42:20 <TravT> maybe you can report back next week?
15:42:25 <rosmaita> anyway, my point was, if we do dinner, need to be done by 8:30 or so if we do it tuesday
15:42:37 <sjmc7> i do eat pretty fast
15:42:40 <rosmaita> ok, will do
15:42:48 <TravT> okay.
15:42:55 <TravT> one other topic
15:42:58 <TravT> #newton specs
15:43:01 <rosmaita> also, i have some info to put on the cross region spec
15:43:04 <TravT> #topic newton specs
15:43:25 <TravT> okay
15:43:30 <TravT> Cross region searching discussion (sjmc7): https://review.openstack.org/301227
15:43:36 <TravT> you can add there
15:43:39 <TravT> or discuss now
15:43:43 <TravT> we also have the session
15:44:06 <sjmc7> yeah, all information is good. won’t be committing to implementing anything til after the summit
15:44:18 <rosmaita> i will put it on spec
15:44:28 <rosmaita> can also summarize here if we have time
15:45:04 <TravT> well, other topic is an update on swift
15:45:08 <TravT> nothing else i know of
15:45:26 <rosmaita> ok, it's a good news/bad news kind of thing
15:45:33 <TravT> ok
15:45:46 <rosmaita> good news is that most customers seem to have all resources in a single region
15:46:03 <rosmaita> around 90%, will put exact number on spec
15:46:16 <rosmaita> so good in the sense maybe don't need to do it
15:46:20 <rosmaita> bad news is,
15:46:40 <rosmaita> of course we want to do it!  HA means multiple regions for most people
15:46:54 <rosmaita> and those customers tend to be the biggest
15:47:36 <rosmaita> anyway, that's the summary
15:47:42 <sjmc7> yeah, we’ve got a similar issue
15:47:53 <sjmc7> most people don’t need it, but the ones that do are the big ones
15:48:11 <TravT> and the big ones probably would appreciate search more...
15:48:28 <rosmaita> but everyone here is +1 on the "stats" idea sjmc7 put on the spec
15:49:15 <TravT> i think so
15:49:18 <TravT> trying to find it again
15:49:45 <sjmc7> https://review.openstack.org/#/c/301227/
15:50:12 <TravT> i don't think we'll be able to discuss deeply further today
15:50:29 <TravT> so might give a quick update on swift if that's okay
15:50:52 <TravT> #topic     Swift: (LakshmiS): https://review.openstack.org/#/c/305404/
15:51:28 <TravT> lakshmiS: do you want to summarize?
15:51:39 <lakshmiS> sure
15:52:05 <lakshmiS> we met with swift team yesterday to discuss the client library
15:52:24 <lakshmiS> since they were hesitant to go with oslo/rabbit mq route
15:52:56 <lakshmiS> they were agreeing to use elasticsearch python client library to communicate directly with elasticsearch behind searchlight
15:53:24 <lakshmiS> but still they haven't quite figured where in their code we can make those calls from
15:53:41 <lakshmiS> they have given some pointers on calling it during container sync process
15:54:39 <lakshmiS> #link http://docs.openstack.org/developer/swift/overview_container_sync.html#what-s-going-on-behind-the-scenes-in-the-cluster
15:55:11 <TravT> i think conceptually we may be looking at a big paradigm shift for that plugin.
15:55:43 <TravT> where searchlight really only needs to have an API plugin for RBAC
15:55:51 <TravT> but not the indexing.
15:56:13 <TravT> will be interesting to explore this area further
15:56:50 <lakshmiS> right. i was thinking to see if we can provide a interface with common methods if we have more than one plugin doing it that way
15:58:01 <TravT> well thanks for the quick work on it
15:58:20 <TravT> i know we gave them something pretty primitive relative to our real plugins
15:58:28 <TravT> but it was enough to continue the conversation
15:58:41 <lakshmiS> yeah it was just for the meeting
15:58:57 <TravT> okay,
15:59:02 <TravT> well i think we're out of time
15:59:18 <TravT> thanks everybody\
15:59:29 <TravT> #endmeeting