14:01:46 <rosmaita> #startmeeting glance 14:01:48 <openstack> Meeting started Thu Dec 1 14:01:46 2016 UTC and is due to finish in 60 minutes. The chair is rosmaita. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:01:50 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:01:52 <openstack> The meeting name has been set to 'glance' 14:01:58 <rosmaita> #topic roll call 14:02:03 <rosmaita> hello everyone 14:02:21 <dharinic> Hello. o/ 14:02:45 <abhishekk> o/ 14:02:59 <alex_bash> o/ 14:04:05 <hemanth|away> o/ 14:04:46 <rosmaita> ok, guess we should get started 14:04:57 <rosmaita> #topic updates - operator surveys 14:05:10 <rosmaita> we've got 2 operator surveys happening right now 14:05:23 <rosmaita> one about the community images database migration 14:05:27 <rosmaita> (more on that a bit later) 14:05:40 <rosmaita> and one about what people use multiple image locations for 14:05:56 <rosmaita> we're getting some good feedback 14:06:31 <rosmaita> i'm going to send a reminder to the operators list about the multiple image locations survey, since the USA holiday happened right after the first email 14:06:40 <rosmaita> we plan to close that survey on dec 9 14:07:10 <rosmaita> we'll discuss another survey (still to be designed) a bit later in the meeting 14:07:39 <rosmaita> i think these are a good way to keep touch with operators and get some good vibes going in the wider community 14:07:57 <rosmaita> so, this was just an update that operators surveys are happening 14:08:11 <rosmaita> but i'll be happy to answer questions if anyone has them 14:08:26 <alex_bash> when is the CI survey overe? 14:08:50 <rosmaita> got the last response about a week ago, so it's effectively over 14:09:01 <rosmaita> i forgot to specify a closing date on that one 14:09:26 <rosmaita> i'll tell you the results at agenda item 4 14:09:54 <rosmaita> anything else? 14:10:10 <rosmaita> (i know i'm supposed to wait quietly for questions, but i can't stand the silence!) 14:10:23 <hemanth> rosmaita: do we have enough feedback take a decision based on the survey? 14:10:36 <hemanth> *to 14:10:37 <rosmaita> i think so, will give my recommendation in item 4 14:10:42 <hemanth> ok 14:11:14 <rosmaita> all right, moving along 14:11:24 <rosmaita> #topic unfinished business 14:11:41 <abhishekk> hi, about 1st patch, https://review.openstack.org/#/c/319960/ - this design was rejected during cross-project discussion with reason described in the commit message 14:12:07 <abhishekk> so I have reverted this patch after discussion with erno 14:13:53 <abhishekk> and regarding 2nd patch, https://review.openstack.org/#/c/352892/ - I am waiting for the feedback, requesting community members to give their opinions on the same 14:14:19 <rosmaita> i think the 2nd patch is a priority, so reviewers, please review it! 14:14:42 <rosmaita> as far as the first patch goes ... 14:14:57 <rosmaita> has there been a release of the glanceclient with your original patch in it? 14:15:08 <rosmaita> because your reversion patch has been sitting since May 23 14:16:04 <abhishekk> no that was not my patch, I have noticed that it was already merged so added a revert for that 14:16:59 <rosmaita> ok, sorry about that 14:17:25 <rosmaita> key thing is it looks like 2.5.0 has that undesirable code in it 14:17:32 <abhishekk> no issues 14:17:41 <abhishekk> yes 14:18:41 <abhishekk> that's all from my side 14:19:23 <rosmaita> abhishekk: what's your opinion on this now? is it important to revert right away, or instead do some kind of thread-local fix as you suggest? 14:19:52 <abhishekk> IMO we should revert this 14:20:47 <rosmaita> OK, can you put a comment on the patch saying that it's important to revert this to prevent incorrect info being returned to the user 14:20:48 <abhishekk> as it has already been rejected in cross-project by many peoples 14:21:11 <abhishekk> yes, I will add it in detail 14:21:24 <rosmaita> thanks, that would be helpful 14:21:42 <rosmaita> i'll put this patch in the list of things for people to pay attention to for the coming week 14:22:06 <rosmaita> ok, and the second patch i think is pretty much ready, just needs reviews 14:22:19 <abhishekk> yes 14:22:57 <rosmaita> ok, great ... thanks for bringing these up (again) and hopefully we can get better action on them this time 14:23:17 <rosmaita> any questions for abhishekk ? 14:23:44 <abhishekk> thank you 14:23:51 <rosmaita> ok, moving along 14:24:06 <rosmaita> #topic Next operators survey: swift backend in multi-tenant configuration 14:24:20 <rosmaita> this is prompted by a patch dharinic has up 14:25:02 <rosmaita> (which i suddenly can't find) 14:25:22 <dharinic> https://review.openstack.org/#/c/387719/ 14:25:29 <rosmaita> thanks! 14:25:58 <rosmaita> that patch could use some reviews, by the way 14:26:24 <dharinic> And the patch that could ideally follow: https://review.openstack.org/#/c/388944/ 14:26:44 <rosmaita> 387719 corrects a bad regression in the swift multi-tenant store 14:27:11 <rosmaita> a regression so bad, that since we haven't heard any complaints, it makes me wonder whether anyone is actually using swift in multi-tenant mode 14:27:26 <rosmaita> or, they are using it in older distributions 14:27:38 <rosmaita> so i'd like to do an operators survey to get some info 14:28:04 <rosmaita> we'd take the survey after the currently open one closes on dec 9 14:28:16 <rosmaita> so here's a etherpad to get some ideas: 14:28:27 <rosmaita> https://etherpad.openstack.org/p/glance-swift-multitenant-usre-survey 14:28:42 <rosmaita> and yes, i misspelled 'user' in the above 14:28:49 <rosmaita> but that's the correct url 14:29:47 <rosmaita> ok, that's it 14:30:07 <rosmaita> summary: please review https://review.openstack.org/#/c/387719/ so we can release glance_store without this bad regression 14:30:17 <rosmaita> and think about what we should ask operators 14:30:44 <hemanth> should we ask if operators ever switch from one store to the other? 14:30:47 <rosmaita> or, if you know people who use swift in multi-tenant config, please add that info to the etherpad 14:31:06 <rosmaita> hemanth: sure ... we can debate that on the etherpad 14:31:26 <rosmaita> i suspect people don't because migration is a big pain point 14:31:32 <rosmaita> but, who knows 14:31:54 <rosmaita> ok, any questions about that? 14:32:09 <jokke_> tumbles in ... sorry for being late 14:32:18 <rosmaita> jokke_: greetings! 14:32:50 <rosmaita> jokke_: you can read through the meeting log later, abhishekk 's discussion earlier will interest you 14:33:00 <rosmaita> ok, moving along 14:33:18 <rosmaita> #topic Community Images: discuss design decision of default visibility = 'shared' 14:33:48 <rosmaita> we talked about this at the last meeting 14:34:14 <rosmaita> i sent out a survey to get some info from operators, here's what i told them: 14:34:27 <rosmaita> #link http://lists.openstack.org/pipermail/openstack-operators/2016-November/012107.html 14:34:42 <rosmaita> you haven't seen that if you don't subscribe to the operators list 14:34:55 <rosmaita> anyway, the results 14:34:58 <rosmaita> (drum roll) 14:35:11 <rosmaita> #link https://review.openstack.org/#/c/396919/ 14:35:21 <rosmaita> take a look at the last comment on that patch 14:35:57 <rosmaita> i will give everyone a few min to read through that 14:38:03 <jokke_> wow 9 responses :( 14:38:23 <rosmaita> yeah, i don't know how to interpret that 14:38:39 <rosmaita> in one sense, great! > 5 responses 14:38:53 <rosmaita> but on the other hand, that's a small number of deployments represented 14:40:27 <jokke_> rosmaita: also regarding the poll setting. What happens to the pre diablo image records? :P 14:40:57 <rosmaita> is_public == 0 -> 'shared' 14:41:06 <rosmaita> (or is that not what you're asking?) 14:42:03 <jokke_> the option A) talks only about Doablo-to-Newton images ... what is the plan on that option for pre diablo created images? 14:42:12 <jokke_> it was not in the poll :P 14:42:21 <alex_bash> I think jokke_ is asking why you called out diablo-to-newton specifically 14:42:21 <rosmaita> oh 14:42:23 <jokke_> so they must stay intact, right :P 14:42:59 <rosmaita> i'm assuming there are no pre-diablo images 14:43:08 <rosmaita> anyway, sharing was introduced in Diablo 14:43:18 <rosmaita> but i get your point now 14:43:41 <rosmaita> i don't think anyone ever deployed anything before essex 14:45:06 <rosmaita> anyway ... 14:45:20 <rosmaita> any other comments about my recommendation? 14:45:43 <jokke_> I'm still very much against that option A 14:45:58 <jokke_> purely user perspective 14:46:02 <rosmaita> yeah, but the operators are split, and jay pipes and monty are for it 14:46:16 <rosmaita> and that assumes that users actually look at visibility 14:46:42 <jokke_> rosmaita: I haven't seen either contributing to Glance for long enough that I'm willing to ignore the opinions of them :P 14:46:42 <rosmaita> i think the (B) migration causes more problems, for example, breaking v1 14:46:56 <hemanth> rosmaita: I still have to catch up on the comments on that review, will comment on it in a bit 14:47:02 <rosmaita> jokke_: i have to admit, if they disagreed with me, i'd say the same thing 14:47:05 <rosmaita> :) 14:47:37 <rosmaita> this is one of those things where no matter what we do, someone is going to be disappointed 14:47:47 <rosmaita> so we just need to go with what makes the most sense 14:47:56 <rosmaita> so my request to everyone is: 14:48:01 <jokke_> Talk again in Pike? :P 14:48:03 <rosmaita> please read through and comment 14:48:06 <rosmaita> NO!!!! 14:48:16 <jokke_> that's easy to achieve 14:49:18 <rosmaita> ok, let's move along before jokke_ makes any more suggestions 14:49:26 <rosmaita> :P 14:49:44 <rosmaita> actually, that brings us to open discussion 14:49:49 <rosmaita> #topic open discussion 14:50:46 <rosmaita> so, anything to discuss? 14:50:54 <bhagyashris> I would like some insights and opinions for bug https://bugs.launchpad.net/glance/+bug/1585917. :) 14:50:54 <openstack> Launchpad bug 1585917 in Glance "member-create will raise 500 error if member-id is greater than 255 characters" [Undecided,Confirmed] - Assigned to Abhishek Kekane (abhishek-kekane) 14:51:41 <jokke_> bhagyashris: is that still a thing? :o 14:52:21 <rosmaita> bhagyashris: i looked into it a bit, even the 'simple' fix is not going to be easy 14:52:54 <bhagyashris> jokke_: yes 14:53:32 <rosmaita> so bhagyashris, it seems like the metadefs fix should be straightforward 14:53:43 <rosmaita> the problematic one is the member string 14:54:27 <rosmaita> so maybe split the work into fixing metadefs first 14:54:29 <bhagyashris> rosmaita: yeah 14:54:50 <rosmaita> but you are right to bring this up, because it's sitting from inertia 14:54:57 <jokke_> rosmaita: ref abhishekk's topic up there, my recommendation was valid when the change wasn't released yet 14:55:38 <bhagyashris> ok. 14:55:49 <rosmaita> bhagyashris: we may have to accept 500s on bad member ids until we work out the json schema support in glance 14:56:03 <rosmaita> but at least the metadefs can be fixed 14:56:35 <rosmaita> jokke_: how do you feel now that the change has been released? 14:56:47 <rosmaita> abhishekk thinks we should still revert it 14:56:59 <rosmaita> he's going to update the patch with a comment to that effect 14:57:18 <jokke_> rosmaita: imo we still need to fix it. We just made ourselves clowns by releasing it, claiming it fixes the issue even we knew it doesn't 14:57:53 <rosmaita> well, i guess it's nice that we can amuse the openstack community 14:57:59 <bhagyashris> rosmaita: metadefs fix are already merged 14:58:17 <rosmaita> bhagyashris: that was quick work! 14:58:54 <jokke_> rosmaita: but I do not have strong feelings of revert vs. just patch that fixes the behavior 14:58:56 <abhishekk> yes, IMO even we fix using thread local storage its not guaranteed to get right result 14:59:39 <rosmaita> ok, let's continue that discussion on the patch 14:59:40 <jokke_> abhishekk: that's not great either ... do you have case in mind when it wouldn't? 14:59:52 <rosmaita> for both abhishekk and bhagyashris patches 15:00:05 <rosmaita> gotta go 15:00:12 <rosmaita> #endmeeting