15:01:41 <TravT_> #startmeeting openstack search
15:01:41 <openstack> Meeting started Thu Sep  3 15:01:41 2015 UTC and is due to finish in 60 minutes.  The chair is TravT_. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:01:43 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:01:45 <openstack> The meeting name has been set to 'openstack_search'
15:02:12 <nikhil_k> o/
15:02:21 <lakshmiS> o/
15:02:26 <rosmaita> o/
15:02:31 <TravT> o/
15:02:40 <sjmc7> morning
15:02:52 <TravT> sigmavirus24:  are you around today?
15:03:06 <sigmavirus24> Yeah
15:03:08 <sigmavirus24> Sitting in a meeting
15:03:18 <sjmc7> livin the dream!
15:03:32 <TravT> as usual, eh
15:04:09 <TravT> #link https://etherpad.openstack.org/p/search-team-meeting-agenda
15:04:09 <TravT> I think we have quite a few items today
15:04:11 <TravT> #topic General Status Update
15:04:46 <TravT> so, big news is that nova patch is in
15:05:06 <TravT> thanks for the continued effort on that sjmc7
15:05:17 <sjmc7> sure
15:05:31 <TravT> #topic release stuff
15:05:49 <TravT> So, we all know the date, I think.
15:06:05 <TravT> https://wiki.openstack.org/wiki/Liberty_Release_Schedule
15:06:05 <TravT> Liberty 3 done feature freeze today
15:06:05 <TravT> RCs Start: September 21st
15:06:07 <TravT> Final RC October 9th
15:06:50 <TravT> Last wee we talked about some release planning and release tags.
15:07:02 <TravT> i want to review that again.  i have a proposal on that
15:07:15 <TravT> but first, i think we should review the current status on blueprints for liberty
15:07:26 <TravT> #link https://blueprints.launchpad.net/searchlight/liberty
15:07:53 <TravT> All but one of the essential BPs is implemented.
15:08:03 * david-lyle slinks in late
15:08:12 <TravT> hi david-lyle
15:08:18 <david-lyle> o/
15:08:28 <TravT> Our high's also look to be in good shape
15:08:36 <TravT> last night i bumped the designate to rc1
15:08:53 <sjmc7> we'll get it in!
15:08:57 <TravT> yep
15:09:25 <TravT> only thing higher priority is functional tests
15:09:46 <TravT> lakshmiS: looks like you have really good progress based on your patch sets?
15:09:52 <TravT> for functional tests?
15:10:11 <lakshmiS> yeah. 6 more tests needs to be added. will be there by tomorrow morning
15:10:24 <TravT> looks like you found a bug in metadefs in the process?
15:10:41 <lakshmiS> yes thats more critical bug
15:11:02 <TravT> ok, we'll look at bugs in just a moment...
15:11:19 <TravT> but on metadefs, we have the BP as Beta Available because we thought it needed more code review
15:11:38 <TravT> i guess you are getting to do some of that now with the functional testing?
15:12:12 <lakshmiS> which bp you are referring to?
15:12:21 <TravT> https://blueprints.launchpad.net/searchlight/+spec/glance-metadef-plugin
15:12:31 <lakshmiS> oh ok
15:12:45 <TravT> it was ported from glance, but didn't get much review
15:12:51 <TravT> so we marked it as beta
15:12:54 <sjmc7> yeah. i'll take blame for that
15:13:05 <sjmc7> i can help fix issues
15:13:09 <lakshmiS> yeah the tests should cover most of it
15:13:19 <TravT> ok.  well, i still will look as well.
15:13:43 <TravT> so then if we look at the medium and lows, on that list above
15:14:05 <TravT> query-perf-tests, i think nikhil said he was still doing work?
15:14:09 <TravT> nikhil_k?
15:15:12 <TravT> or not...
15:15:14 <TravT> :)
15:15:24 <TravT> on the horizon integration for images.
15:15:29 <david-lyle> he's in rel-mgr-office now
15:16:05 <TravT> i think we'll be looking at horizon integration for presentation at the Tokyo summit
15:16:14 <TravT> last time we have very POC'ish
15:16:29 <TravT> i'm hoping this time it will be actual patches up for review.
15:17:19 <TravT> there are a few other medium / low
15:17:32 <TravT> but we don't have to go over now
15:17:42 <TravT> we should talk about the release tags.
15:17:59 <TravT> My proposal is that we target these two:
15:18:09 <TravT> #link https://github.com/openstack/governance/blob/master/reference/tags/release_cycle-with-intermediary.rst
15:18:40 <TravT> ^ for now, meaning we do a liberty release.
15:18:52 <TravT> but we don't necessarily have to do milestone releases.
15:19:07 <nikhil_k> TravT: sorry trying to be in 3 places, isn't working
15:19:12 <sjmc7> l3 would mean doing today, right?
15:19:18 <TravT> l3 is milestone
15:19:28 <TravT> that would mean we were following this one:
15:19:33 <TravT> https://github.com/openstack/governance/blob/master/reference/tags/release_cycle-with-milestones.rst
15:19:35 <david-lyle> sjmc7: yes, generally
15:19:50 <sjmc7> so let's not do that :)
15:20:04 <david-lyle> well it's a tag
15:20:17 <lakshmiS> its only feature freeze anyway
15:20:18 <david-lyle> then you have most of Sept to pull together an RC-1
15:20:29 <david-lyle> L-3 is not intended to ship
15:20:41 <david-lyle> it should be working, but is not final
15:21:06 <TravT> There is the "independent" one.
15:21:13 <TravT> meaning release whenever.
15:21:20 <TravT> https://github.com/openstack/governance/blob/master/reference/tags/release_independent.rst
15:22:02 <TravT> At least for now, i think there is value in targetting a release to happen that lines up with the larger release cycle
15:22:16 <sjmc7> yes
15:22:44 * david-lyle taking his turn in relmgr-office back soon
15:23:28 <TravT> So, that means our goal for the next few weeks is fleshing out the couple of BPs and big bugs
15:23:50 <lakshmiS> that would be nice
15:24:01 <TravT> we can decide our actual version in a week or two based on how things look.
15:25:04 <TravT> david-lyle had the idea of doing a release version of 0.8.0, to signify this is an initial release
15:25:37 <TravT> the .8 correlating to the bigger OS release number of 8.0
15:25:43 <TravT> something to chew on
15:26:20 <TravT> On to some specific reviews
15:26:30 <TravT> #topic reviews
15:26:51 <TravT> i'd like to call attention to this one:
15:26:52 <TravT> Index CLI improvements: https://review.openstack.org/#/c/210759/
15:27:05 <TravT> this changes our command line interface
15:27:09 <TravT> for the better
15:27:28 <TravT> but we might get stuck with the options it proposes for some time
15:27:34 <TravT> so i think input would be good.
15:28:05 <sjmc7> input away!
15:28:05 <TravT> sigmavirus24: nikhil_k: rosmaita: david-lyle: lakshmiS: ^ https://review.openstack.org/#/c/210759/
15:28:49 * rosmaita looking
15:30:05 <david-lyle> what is the value of allow stale?
15:30:19 <david-lyle> s/value/reasoning for/
15:30:21 <TravT> A little problem statement: right now if you do an index sync command, it will go through ALL the plugins and sync ALL the data.  In addition, it currently does not clear out non-current data.
15:31:04 <TravT> so, if notifications get messed up, and say some instances were deleted, calling index sync now would not clear them out.
15:31:32 <TravT> sjmc7 put in an option to clear all data, and my feedback was that clearing all data should be the default.
15:31:59 <TravT> possibly leaving stale data being the option...
15:32:01 <david-lyle> I agree with the second part :)
15:32:14 <david-lyle> what's the benefit of stale data?
15:32:26 <TravT> that's a good question...
15:32:27 <david-lyle> if I just went and made the API call to index
15:32:30 <sjmc7> it is!
15:33:00 <david-lyle> unless you're trying to build a history up, but I think it would just cloud the results with non-existent items
15:33:01 <sjmc7> if we're ok breaking back compatability, i'm more than happy to remove it
15:33:16 <TravT> clear also recreates the mappings
15:33:29 <david-lyle> backward compatibility with kilo glance?
15:33:30 <TravT> so there are some other side effects...
15:33:42 <david-lyle> or Liberty Searchlight
15:33:56 <sjmc7> :)  well, in the sense it was already that way. if there was no reason for it, i'll take it out
15:34:12 <sjmc7> speak now, or forever hold your piece!
15:34:47 <david-lyle> if just Liberty the API isn't really made official yet by a release
15:35:05 <sjmc7> this isn't even the API, it's the management command
15:35:08 <david-lyle> TravT: such as?
15:35:14 <TravT> sjmc7, any other negative rammifications to just always clearing?
15:35:23 <david-lyle> sjmc7: even better
15:35:42 <sjmc7> nope. do you want me to drop the confirm: y/n prompt?
15:35:44 <TravT> david-lyle: i was just meaning that clearing clears data and the data mappings for the data.
15:35:48 <sjmc7> go all wild west?
15:36:16 <david-lyle> so the only downside is an API call in the middle of a sync may get partial data?
15:36:19 <sjmc7> adding options later (like to not clear mappings) is possible if we find a use case
15:36:27 <sjmc7> david-lyle - yep
15:36:50 <david-lyle> I think that's better than non-existent items
15:36:50 <rosmaita> maybe this is stupid, but maybe don't have a default?
15:36:51 <sjmc7> if someone's doing this often, there are ways to do it to avoid that (index into a different index and switch them over)
15:36:57 <rosmaita> force user to specify which one?
15:37:16 <rosmaita> then there's no change of behavior for scripts calling the manage command
15:37:45 <TravT> sjmc7: i still think we should talk about the alias idea, to better faccilitate that
15:37:50 <TravT> but another topic
15:37:53 <TravT> for another time
15:38:08 <sjmc7> the question is whether there's any reason to ever leave old data
15:38:23 <sjmc7> in effect, whether that is essentially a bug in the implementation
15:38:24 <TravT> i can't think of any.
15:38:51 <TravT> and i don't see how anybody would have reason to know they should leave something old in
15:39:18 <TravT> that would be quite the arduous task to figure out, i'd think
15:39:27 <rosmaita> i can't think of any reason to do it, either
15:39:34 <sjmc7> ok. decision made!
15:39:43 <rosmaita> and as sjmc7 says, you could keep a copy if yo ureally want to
15:39:46 <sjmc7> i'll leave the prompt in, since it is a destructive operation
15:39:48 <TravT> now, re: the prompt
15:40:10 <david-lyle> \o/
15:40:25 <sjmc7> :)
15:40:26 <TravT> i like the prompt, especially because if you don't specify a specific type, you'll re-index everything which could be quite the expensive operation
15:40:34 <sjmc7> and i should note, i took horizon as my inspiration for that
15:41:05 <rosmaita> i wonder whether the override should be --force instead of --no-input
15:41:06 <TravT> however, I was thinking the prompt could include a list of what will be re-indexed.
15:41:08 <rosmaita> sounds scarier
15:41:19 <sjmc7> :D
15:41:39 <sjmc7> --force is more unix-y
15:41:51 <sjmc7> TravT - yeah, that's easy to do
15:42:25 <TravT> so, it would be searchlight-manage index sync --type OS::Nova::Server
15:42:41 <TravT> Are you sure, you don't look like you know what you are doing (y/n)?
15:42:55 <lakshmiS> :)
15:43:01 <TravT> or searchlight-manage index sync --type OS::Nova::Server --force
15:43:06 <TravT> that sounds good to me.
15:43:21 <sjmc7> yep. and i can add a list of resources that'll be re-indexed before the prompt
15:43:27 <TravT> yeah
15:43:36 <rosmaita> that would be really useful
15:43:37 <sjmc7> "You're about to delete all your data, you fool"
15:43:51 <TravT> i'd be okay with a small bit of humor in the prompt...
15:43:58 <sjmc7> kidding :)
15:44:23 <sjmc7> i'll get another patch back up after this. any other issues with it briefly? or submit reviews
15:44:35 <TravT> ok, cool, thanks for letting me publicly harass you this time sjmc7
15:44:44 <sjmc7> :)
15:44:47 <TravT> i just saw this as an important change
15:44:50 <TravT> worthy of discussion
15:45:32 <TravT> ok, so let's quickly look at other bugs
15:45:45 <TravT> #link https://bugs.launchpad.net/searchlight
15:46:18 <TravT> The sort parameter is critical and could use another set of eyes on it
15:46:29 <TravT> https://review.openstack.org/#/c/206268/
15:46:31 <sjmc7> yeah. it's also reasonably simple
15:46:40 <TravT> yep
15:46:59 <rosmaita> good catch, that's a key feature request
15:47:48 <TravT> There is another big one we found
15:48:02 <TravT> https://bugs.launchpad.net/searchlight/+bug/1491085
15:48:03 <openstack> Launchpad bug 1491085 in OpenStack Search (Searchlight) "Searchlight not updating on member change" [High,New]
15:48:26 <TravT> so, we don't get notified when glance members change
15:48:53 <TravT> lakshmiS: did you find any old code ready to submit on that to glance?
15:49:22 <lakshmiS> yes, i will put a initial patch. will need some input from glance team though(may be nikhil?)
15:49:26 <rosmaita> i agree with sjmc7 that glance should consider this
15:49:33 <sjmc7> i did take a look at glance yesterday. it doesn't seem like a really simple fix; adding another layer to CRUD pipeline isn't as simple as i'd hoped
15:50:43 <TravT> lakshmiS: when you get something up, can you notify rosmaita: and nikhil_k:
15:50:56 <TravT> so we can get some attention to it
15:50:56 <lakshmiS> hope glance treats this as a bug for liberty
15:50:59 <lakshmiS> sure
15:51:16 <sjmc7> it was a bug for kilo :)
15:51:23 <rosmaita> kind of funny, we were discussing feature requests disguised as bugs in the glance meeting earlier
15:52:11 <TravT> yesterday we debated on bug vs bp a bit.
15:52:15 <TravT> decided to start with bug
15:52:26 <sjmc7> talking of glance, https://bugs.launchpad.net/searchlight/+bug/1490697 is related to the glanceclient 1.0 release
15:52:27 <openstack> Launchpad bug 1490697 in OpenStack Search (Searchlight) "python glance-client for image-members has upgraded" [High,In progress] - Assigned to Lakshmi N Sampath (lakshmi-sampath)
15:52:27 <TravT> but if we need ffe on it, then we need to know asap
15:52:37 <david-lyle> it's a bug that searchlight doesn't index neutron, go
15:52:43 <sjmc7> :D
15:54:22 <TravT> Do we need another dedicated review hour?
15:54:26 <TravT> like we did last week?
15:55:09 <TravT> guess not?
15:55:27 <sjmc7> i'm happy to do it
15:55:42 <sjmc7> since i have a few changes still need reviewing
15:55:44 <lakshmiS> guess everyone busy with releases...
15:55:50 <TravT> Yeah, today is rough.
15:56:09 <TravT> let's plan on having one next Tuesday again.
15:56:16 <TravT> just an hour to be online
15:56:20 <TravT> and get through open questions
15:56:26 <sjmc7> ok
15:56:27 <lakshmiS> sounds good
15:56:49 <TravT> ok, i'll send out a courtesy meeting request
15:57:05 <TravT> That's it for today, unless there are other items
15:57:16 <TravT> thanks for baring with me.
15:57:20 <TravT> :)
15:57:59 <lakshmiS> 3 minutes early. signing out!
15:58:01 <TravT> #endmeeting
15:58:31 <TravT_> #endmeeting