15:00:51 <sjmc7> #startmeeting openstack search 15:00:52 <openstack> Meeting started Thu Oct 6 15:00:51 2016 UTC and is due to finish in 60 minutes. The chair is sjmc7. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:53 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:57 <openstack> The meeting name has been set to 'openstack_search' 15:00:57 <RickA-HP> Morning! 15:01:06 <TravT> o/ 15:01:16 <lei-zh> o/ 15:01:36 <sjmc7> #link https://etherpad.openstack.org/p/search-team-meeting-agenda 15:02:04 <sjmc7> i hope it is sunnier wherever people are than it is here 15:02:12 <TravT> nope, raining here 15:02:32 <RickA-HP> Sunny and perfect waves here! 15:02:48 <lei-zh> rainy Beijing now : ) 15:02:49 <sjmc7> i haven’t seen either of those things for months 15:02:51 <rosmaita> cloudy here 15:03:04 <sjmc7> ok. that’s all i had to discuss, the weather 15:03:11 <sjmc7> we shall get started 15:03:12 <sjmc7> #topic Newton release status 15:03:20 <sjmc7> it’s very almost done 15:03:30 <sjmc7> think the release team are or have tagged the final builds today 15:03:40 <sjmc7> so we are almost officially 1.0! 15:04:09 <RickA-HP> Doug H sent out an email that Newton has been released. 15:04:25 <sjmc7> ah yes, with his commit message “Make it so" 15:04:45 <sjmc7> so good job everyone that helped get it done 15:04:51 <TravT> \o/ 15:04:56 <sjmc7> which moves me swiftly onto 15:05:00 <sjmc7> #topic Ocata summit 15:05:23 <sjmc7> TravT: we’ve got 1 fishbowl and a couple of working rooms, correct? 15:05:38 <TravT> yes 15:05:44 <sjmc7> #link https://etherpad.openstack.org/p/searchlight-ocata-summit 15:05:56 <sjmc7> sadly i will not be there, which is a terrible shame since barcelona is an excellent city 15:06:15 <TravT> and 1 piece of info i learned yesterday is that our fishbowl collides with a horizon working session. 15:06:23 <sjmc7> ah, wonderful 15:06:40 <sjmc7> is it at least nearby? 15:06:40 <TravT> the horizon operators feedback fishbowl is right before ours in the same room 15:06:57 <sjmc7> not all the working sessions are of interest to everyone 15:07:30 <TravT> yeah, their working session is a retrospective and angular planning session 15:07:32 <TravT> https://docs.google.com/spreadsheets/d/1TQ-RSlbiBBEclkonIbfUP7R1ExZSJylF1uiEKV2G_Cw/pubhtml?gid=1107826458&single=true 15:07:44 <sjmc7> hmm… won’t that affect many of the same people? 15:08:22 <sjmc7> maybe could do some horsetrading 15:08:37 <sjmc7> in any case, i still think your fishbowl idea for searchlight is a good one 15:08:49 <TravT> yeah, unfortunately, i'm sure will loss robcresswell, r1chardj0n3s, david-lyle though 15:09:12 <sjmc7> :( 15:10:15 <TravT> this one: Searchlight and Searchlight UI Plugins - How to enable your service in Searchlight 15:10:26 <sjmc7> yeah 15:10:47 <sjmc7> that’s a possibility that wouldn’t need so much horizon input 15:11:03 <TravT> piet did say he could also do a usability results review as well... which is more of interest to horizon 15:11:03 <sjmc7> can maybe include something on enhancing horizon with the aggregation queries 15:11:19 <sjmc7> yeah, btu not if they’re all in a working group 15:13:47 <sjmc7> well, my vote would be to go with that idea 15:14:22 <sjmc7> and maybe see if we can have rob switch something on horizon 15:14:34 <TravT> well, i could tell david, richard, rob that we will be doing that idea and will be showing enhancing horizon with agg queries as part of it. 15:14:42 <sjmc7> yeah 15:15:15 <sjmc7> for the working rooms, the two things that seemed to get interest were ‘ease of use of searching’ and ‘error handling’ 15:15:36 <sjmc7> the working rooms in a sense are much more flexible anyway 15:15:46 <sjmc7> especially if brian brings pie 15:16:32 <rosmaita> not sure what the pie situation is 15:16:49 <rosmaita> i can handle USA pie, don't want to cause an international incident 15:17:20 <sjmc7> it’ll likely be octopus or something 15:17:47 <sjmc7> ok, so any objections to those three things going on the agenda? for the working rooms we can add stuff at any time 15:18:25 <lei-zh> nope 15:19:17 <TravT> For fishbowl, maybe we could do a 5 minute overview and demo, then have piet do 10 minutes on usability results, then myself or lei-zh doing 2 - 5 slides on plugin structure and steps to implement in SL. matt doing 10 mins on sl-ui, and finally a demo of aggregation queries enhancing the images table. 15:19:42 <sjmc7> that’s quite a tour de force 15:19:45 <TravT> it is... 15:19:50 <TravT> maybe too much 15:19:53 <TravT> would need to be crisp 15:20:04 <rosmaita> i wonder what the word for 'smorgasbord' is in Catalan? 15:20:06 <sjmc7> like a pie crust 15:20:15 <sjmc7> you’ve got three weeks to find out 15:20:24 <sjmc7> ok, i’ll put those things on the summit agenda thing 15:20:30 <rosmaita> google says it's 'smorgasbord' 15:20:41 <sjmc7> disappointing 15:20:43 <TravT> make it sound interesting... 15:20:55 <sjmc7> i will do my utmost 15:21:07 <sjmc7> anything else on this before we move on? the last couple of things’ll be pretty quick 15:21:10 <TravT> that might attract a few of the operators from the horizon session 15:21:22 <sjmc7> yeah. you can do some schmoozing too 15:21:26 <TravT> so, what are the working sessions? 15:21:44 <sjmc7> ‘ease of use for searching’, which more of in a second 15:21:57 <sjmc7> and improving robustness with indexing 15:22:37 <TravT> lei-zh, would want to do the few minutes on building a searchlight plugin? Should only be about 10 minutes? 15:23:05 <lei-zh> sure 15:23:14 <lei-zh> about writing a new plugin, right? 15:23:23 <TravT> yep 15:23:44 <lei-zh> ok, I'll take that 15:23:52 <TravT> would be kind of a powerpoint version of http://docs.openstack.org/developer/searchlight/authoring-plugins.html 15:24:07 <TravT> but simpler 15:24:15 <TravT> just an outline 15:24:18 <TravT> sounds great! 15:24:42 <TravT> we could do a hangout in a week or so to practice / put slides together 15:25:11 <lei-zh> sounds a good idea 15:25:12 <TravT> i can start a google slides project that we just collaborate on and you can add your slides into 15:25:36 <sjmc7> i will leave that in your hands, and look over shoulders/point out spelling mistakes 15:26:11 <TravT> thanks lei-zh 15:26:42 <lei-zh> I'll do my best :) 15:26:44 <sjmc7> and i mainly mean travis’ spelling mistakes 15:27:08 <sjmc7> ok. i’ll get those added into the summit scheduling system 15:27:42 <sjmc7> i had two last things, one of which goes in hand with improving usability 15:27:55 <sjmc7> #topic Transforming data 15:28:05 <sjmc7> i put up https://review.openstack.org/#/c/382572/ after some feedback from one of our developers 15:28:12 <sjmc7> that IP searches were very difficult 15:28:45 <sjmc7> it standardizes IP addresses for resources where it makes sense, in the same way we did for ‘name’ 15:29:31 <sjmc7> this obviously isn’t compatible with the openstack APIs, but i think it’s ok to enhance/add stuff with the aim of making it easier to find stuff 15:29:35 <sjmc7> thoughts/disagreements? 15:30:12 <rosmaita> i like it 15:30:15 <RickA-HP> I haven't looked at it yet, but this is a superset of the APIs, correct? In other words we are only adding fields. 15:30:16 <TravT> i agree, but had a question about boosting 15:30:18 <sjmc7> everyone’s still thinking about octopus pies 15:30:25 <sjmc7> RickA-HP: correct 15:30:34 <sjmc7> and we have done that in other place sin a limited fashion 15:30:49 <sjmc7> TravT: yeah, in some cases this’ll make IP addresses more boosted 15:30:51 <sjmc7> since they’ll match more 15:31:20 <sjmc7> though actually, there’s only one place we map IPs as “type”: “ip” currently because ipv6 is troublesome 15:31:53 <sjmc7> and also, in general you need to specify a field when you’re doing things like CIDR or IP range searches 15:32:20 <sjmc7> is that what your question about boosting was? or did i just go off on a tanget? 15:34:20 <sjmc7> i’ve stunned you into silence 15:34:38 <TravT> that was basically it 15:34:50 <TravT> if i just type 10.0.1.1/24 15:34:56 <TravT> not in a specific field 15:35:17 <TravT> is the relevancy going to still be correct? 15:35:21 <sjmc7> that’s an interesting one - if you do that, it’s searching the _all field 15:35:31 <sjmc7> which is a smorgasbord of all the fields 15:35:38 <rosmaita> :) 15:35:49 <TravT> yeah, and in watching a couple of the usability studies (i only could go to a couple) 15:35:58 <TravT> they kind of expected to just type something like that 15:36:01 <sjmc7> yeah 15:36:09 <sjmc7> and that won’t work in some cases 15:36:19 <TravT> but, i don't think that downgrades what your patch does 15:36:35 <sjmc7> no; the patch makes it possible to say “ipv4_addresses:10.0.1.1/24" 15:36:39 <sjmc7> and get all things on that subnet 15:36:53 <sjmc7> i toyed with adding start and end ranges for subnets too 15:37:31 <sjmc7> i think though that now we’re pretty good on indexing data, we can look at improving usability 15:37:40 <sjmc7> so the studies you watched might give us some other clues 15:38:38 <sjmc7> i shall wait a few moments for that to soak in before moving on to the last topic 15:40:36 <TravT> i agree. 15:40:45 <TravT> make the searches more useful and accurate 15:40:57 <TravT> can't see what there is to complain about there 15:41:11 <sjmc7> okey doke. on a somewhat similar vein then: 15:41:13 <sjmc7> #topic Openstack API recommendations versus Elasticsearch 15:41:33 <sjmc7> in https://review.openstack.org/#/c/381956/ i added ‘size’ and ‘from’ as synonyms for ‘limit’ and ‘offset’ 15:42:03 <sjmc7> it was pointed out that ‘limit’ and ‘offset’ are openstack (or rather, SQL) standards which is why we had them 15:42:15 <sjmc7> but size and from are elasticsearch DSL standards 15:42:41 <sjmc7> since our API largely maps directly to elasticsearch, i’m fine with having both 15:43:21 <sjmc7> and i think it would be defensible to, say, the api working group, but there’s a risk we could end up conflicting at some point 15:44:13 <rosmaita> it's probably worth running by the api wg 15:44:22 <RickA-HP> I think it may be confusing to have both. It seems that we should stay consistent with the OpenStack API. 15:44:53 <rosmaita> yeah, but we should at least reserve 'size' and 'from' so they don't get used by os api at some point 15:44:55 <RickA-HP> I think the only ones who may want "size:" and "from" are the SL developers who are used to dealing with Elastricsearch :) 15:45:00 <sjmc7> rosmaita: i will, though i can forsee what they’ll say 15:45:05 <sjmc7> RickA-HP: that would be me :) 15:45:13 <RickA-HP> And me! 15:45:17 <sjmc7> it does trip me up about every time i use it 15:45:41 <RickA-HP> Besides, these are parameters and not really a part of the DSL query. 15:46:06 <sjmc7> the entire payload is compatible with the DSL for the most part 15:46:40 <sjmc7> ok. i’ll run it by the WG 15:47:27 <sjmc7> i can always have my shell substitute ‘limit’ for ‘size’ :) 15:47:37 <TravT> i think both would be good 15:47:46 <TravT> because we refer to ES docs quite often 15:47:55 <TravT> and ES docs have info on pagination 15:48:03 <TravT> having both can be confusing... 15:48:04 <TravT> but 15:48:33 <sjmc7> yeah. the problem comes when there’s a clash, i guess. if ‘size’ ever got used to mean something else we could always drop it i suppose as an API change 15:48:46 <sjmc7> i’ll ask the WG and add a comment on the review 15:50:12 <sjmc7> your hanging ‘but’ has thrown me 15:50:34 <TravT> i don't know what to say 15:50:38 <sjmc7> :) ok 15:50:40 <sjmc7> that was all i had 15:50:41 <TravT> i can see both arguments 15:51:01 <TravT> and at the same time the API wg couldn't actually ratify their pagination recommendation anyway, right? 15:51:03 <sjmc7> yeah. i’ll see what the WG say, people can vote in comments. it’s not gonna make me cry if we leave it 15:51:15 <sjmc7> no.. which is why the marker thing is still in use in most places but not all 15:51:20 <rosmaita> well, i think the api wg could have a position on openstack not using query parameters that are used by other well-known apis for different purposes 15:51:38 <sjmc7> yeah, that’s true 15:51:48 <sjmc7> ok, will ask them 15:51:54 <sjmc7> we’re nearly out of time, bizarrely 15:51:58 <sjmc7> #topic open discussion 15:52:03 <sjmc7> anything else anyone wanted to bring up? 15:52:17 <rosmaita> but, the images api already allows you to filter by size 15:52:39 <sjmc7> so do we, for cinder/glance 15:53:42 <sjmc7> but yeah, “size” could be ambiguous in lots of circumstances 15:53:43 <lei-zh> our team has an openstack book for second edition, think I will write something to introduce Searchlight 15:53:56 <sjmc7> oo, that’d be good lei-zh 15:54:16 <sjmc7> TravT: you said mirantis had included it in their containerized deployment too? 15:54:28 <TravT> cool! 15:54:37 <lei-zh> but it's in Chinese : ) 15:54:38 <TravT> i saw that they mirantis had a repo set up 15:54:51 <TravT> also, core os has something called stackenetes 15:55:12 <TravT> they include searchlight 15:55:13 <TravT> https://github.com/stackanetes/stackanetes 15:55:16 <sjmc7> lei-zh: maybe it’ll sell enough to get an english translation 15:55:33 <TravT> Basically, stackenetes is the core projects + searchlight 15:55:40 <sjmc7> it’s certainly good we’re getting more exposure, especially if it brings more users/contributors in 15:55:46 <TravT> cinder, glance, horizon, keystone, neutron, nova, searchlight 15:56:07 <admin0> \o/ 15:56:56 <TravT> o/ 15:57:24 <TravT> i have a feeling the current meeting should end. 15:57:38 <TravT> lei-zh: i just put together a slide outline on google slides. 15:57:46 <TravT> will send link to you 15:58:04 <sjmc7> yep, think we’re done 15:58:11 <sjmc7> thanks everyone, enjoy the rain 15:58:15 <sjmc7> unless you’re RickA-HP 15:58:25 <RickA-HP> :) 15:58:30 <lei-zh> lol 15:58:36 <sjmc7> #endmeeting