14:00:00 <rosmaita> #startmeeting glance 14:00:01 <openstack> Meeting started Thu Jan 5 14:00:00 2017 UTC and is due to finish in 60 minutes. The chair is rosmaita. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:00:02 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:00:06 <openstack> The meeting name has been set to 'glance' 14:00:10 <rosmaita> #topic roll call 14:00:12 <sigmavirus> o/ 14:00:20 <rosmaita> sigmavirus: you are too quick! 14:00:20 <abhishekk> o/ 14:00:21 <sigmavirus> Happy New Year! 14:00:28 <sigmavirus> rosmaita: or am I QUIC? 14:00:35 <stevelle> o/ 14:00:38 * sigmavirus hopes that joke doesn't go over anyone's TCP layer 14:00:38 <jokke_> o. 14:00:38 <abhishekk> Happy New Year.. 14:00:42 <dharinic> \o 14:00:57 * sigmavirus is in a punnnchy mood 14:01:06 <rosmaita> well, happy new year, anyway 14:01:18 <jokke_> Happy New Year everyone! 14:01:29 <rosmaita> ok, we're expecting a light turnout today, so we can get started 14:01:50 <rosmaita> #topic priority item review 14:02:08 <rosmaita> #link http://lists.openstack.org/pipermail/openstack-dev/2016-December/109375.html 14:02:23 <rosmaita> looks like not much action, but it was the holidays 14:02:38 <abhishekk> for request-id, I have added comment for sigmavirus, please have a look on the same 14:02:48 <rosmaita> abhishekk: ty 14:03:25 <sigmavirus> will do abhishekk 14:03:30 <abhishekk> also, I am working on member-id schema issue (oneOf implementation) 14:03:38 <rosmaita> great 14:03:40 <abhishekk> thank you sigmavirus 14:03:59 <abhishekk> soon propose a patch for same 14:04:05 <rosmaita> sounds good 14:04:32 <rosmaita> ok, let's move quickly to user survey question, then i want to update on the community images situation 14:04:41 <rosmaita> #topic user survey question 14:04:58 <rosmaita> #link http://lists.openstack.org/pipermail/openstack-dev/2016-December/109376.html 14:05:22 <rosmaita> jokke_ had a suggestion on the ML 14:06:08 <rosmaita> I'd like to have question "Which Images Api Version are you using? 14:06:08 <rosmaita> Pick any that applies. a) v1 b) v2; If v1, Why?" 14:06:08 <rosmaita> Could help us prioritizing the work needed to get everybody off from 14:06:11 <rosmaita> v1 and it out of support. 14:06:16 <rosmaita> (that's a quote from jokke_ ) 14:06:21 <jokke_> ++ 14:07:04 <sigmavirus> I think more specifically "Why haven't you moved to v2?" 14:07:10 <rosmaita> ++ 14:07:10 <sigmavirus> or "What prevents you from moving to v2?" 14:07:17 <rosmaita> sigmavirus: you're reading my mind 14:07:40 <rosmaita> let's ask it in a way that it's obvious that v1 has been deprecated and is being removed soon 14:07:52 <jokke_> for once I tried, I really tried to turn that question setting to positive tone ;) 14:08:06 <rosmaita> jokke_: your efforts are appreciated! 14:08:12 <sigmavirus> it's too opaque though 14:08:19 <sigmavirus> the effort was good one though 14:08:27 <sigmavirus> it wasn't condescending ;) 14:08:32 <jokke_> :P 14:08:50 <rosmaita> we can have a "short answer" style question 14:09:16 <jokke_> we could also put there "Which Images API version is still supported? Pick one: a) v2" :P 14:09:22 <rosmaita> :) 14:09:33 <rosmaita> win fabulous prizes! 14:09:59 <rosmaita> ok, i'll put up an etherpad and we can refine the question along the lines sigmavirus suggests 14:10:16 <rosmaita> i think it's a good suggestion, so thanks jokke_ 14:10:28 <jokke_> yeah ... I'm happy with that ... I'd just like to get everyone over with that specific topic :D 14:11:04 <rosmaita> #action rosmaita put up etherpad to revise survey question & send email to dev list to get comments 14:11:22 <rosmaita> ok, moving along 14:11:35 <rosmaita> next item isn't on the agenda, but what the heck 14:11:42 <rosmaita> #topic community images impasse 14:12:05 <jokke_> I was just gonna ask if we should discuss this as part of the priority or do we have own topic for it 14:12:21 <rosmaita> yeah, can be own topic 14:12:29 <rosmaita> the quick summary is 14:12:40 <rosmaita> stevelle explained this at the last meeting 14:12:56 <rosmaita> the qa team has a problem with a patch he proposed 14:13:11 <rosmaita> they don't like the backward-incompatibility in the workflow for image sharing 14:13:40 <rosmaita> the incompatibility is that now you get a 409 if you try to add a member to an image that is not in visibility == shard 14:14:00 <rosmaita> previously, the operation would succeed on an image with visibility==private 14:14:19 <rosmaita> we've discussed the heck out of this change and why it makes sense 14:14:23 <jokke_> I did not reply to the ML chain but I added my comment on the etherpad 14:14:49 <rosmaita> #link https://etherpad.openstack.org/p/glance-ocata-community-images-tempest-continued 14:14:57 <rosmaita> jokke_: thanks for the reminder 14:15:24 <jokke_> I'd also like to point out that the timeline was misleading and I took it wrong ... that test change has been inflight for months, it was not something super new that popped up conveniently just before we merged the change 14:15:37 <rosmaita> good to know 14:15:47 <sigmavirus> Yeah, the QA team is not out to get us 14:15:54 <rosmaita> that's a relief! 14:15:55 <sigmavirus> I can sympathize with both us and them 14:16:03 <jokke_> same here 14:16:37 <jokke_> and like I commented to the etherpad IMHO now we need to decide which one we disappoint. The QA team or users and the community 14:16:45 <stevelle> confessing I didn't read that etherpad yet 14:16:46 <sigmavirus> we dont' have time for microversioning so keeping private->shared auto-transition functionality seems like the best path forward 14:16:47 <rosmaita> my problem ATM is that if we take the api stability guidelines literally, this entire change cannot happen 14:17:02 <jokke_> rosmaita: ++ 14:17:07 <rosmaita> i.e., we cannot even have 'community' visibility 14:17:10 <sigmavirus> I'm not sure I agree rosmaita 14:17:15 <stevelle> it could with microversioning 14:17:33 <jokke_> well microversioning is just excuse to break the rules 14:17:40 <sigmavirus> expansions of the API are okay, and as I understand this change, it's an expansion if we implement private -> shared transition for users 14:17:55 <jokke_> and I'm really not liking how it's treated ... not fond of us going there 14:18:04 <dharinic> agree with sigmavirus 14:18:07 <sigmavirus> I don't agree jokke_ 14:18:30 <rosmaita> let's take a look at the docs 14:18:33 <rosmaita> #link http://specs.openstack.org/openstack/api-wg/guidelines/evaluating_api_changes.html 14:18:55 <rosmaita> under "The following types of changes are generally not considered acceptable:" 14:19:05 <rosmaita> "Changing or removing a property in a resource representation." 14:19:43 <sigmavirus> rosmaita: so the property in question is visibility which has possible values of {"private", "public"} yes? 14:19:44 <rosmaita> the guidelines mention "API extensions" which have been eliminated in favor of microversions (in nova, at least) 14:19:50 <rosmaita> sigmavirus: yes 14:19:53 <sigmavirus> extension != expansion 14:19:58 <rosmaita> right 14:20:05 <rosmaita> extensions are ok 14:20:18 <sigmavirus> rosmaita: and we're talking about adding states to "visibility" to make the possible values {"private", "shared", "public", "community"} yes? 14:20:29 <rosmaita> sigmavirus: yes 14:21:02 <jokke_> sigmavirus: and that changes the representation of the property, no? 14:21:19 <sigmavirus> so while the underlying semantics have changes, the representation contract hasn't strictly been broken, it's been expanded 14:21:48 <stevelle> this may fall under "Changing the semantics of a property in a resource representation which may be supplied by clients." 14:21:53 <sigmavirus> yes it's a change, but I think your interpretations (rosmaita & jokke_) are too literal 14:22:11 <rosmaita> stevelle: yes, which is also not allowed 14:22:16 <stevelle> but I don't think it violates the above meaningfully if we implement private -> shared 14:22:48 <rosmaita> sigmavirus: yes, my point is that the guidelines are probably not meant to be taken literally 14:23:00 <rosmaita> so the question is, who decides what is OK and what not? 14:23:01 <sigmavirus> stevelle++ 14:23:22 <rosmaita> stevelle: you mean the automatic transition? 14:24:03 <stevelle> yes 14:24:26 <jokke_> the auto transition breaks actually two of the guidelines ... it breaks the representation and the semantics 14:24:32 <stevelle> rosmaita: I believe that question is legitimate, but not ultimately going to solve anything 14:25:11 <rosmaita> stevelle: you are probably correct 14:25:17 <rosmaita> but this is why i'm raising it 14:25:43 <rosmaita> although i am having trouble articulating it 14:25:44 <hemanthm> stevelle: I thought we didn't want to do automatic transitions. Also, if we are doing private->shared, what is the point of making shared as the default state? 14:25:53 <jokke_> rosmaita: wasn't API WG ok with the change already? 14:26:08 <rosmaita> jokke_: that was my understanding 14:26:10 <jokke_> rosmaita: and they are behind these guidelines? 14:26:21 <rosmaita> jokke_: well, that's an interesting question 14:26:29 <rosmaita> i think the guidelines predate the API-WG 14:26:46 <jokke_> well the guidelines lives under api-wg if you look the uri 14:26:48 <rosmaita> although, i am very possibly wrong about that 14:26:55 <rosmaita> was just noticing that! 14:27:50 <stevelle> API WG drafted that guideline 14:28:23 <rosmaita> does anyone know if there's an api-wg meeting this week? 14:28:28 <stevelle> hemanthm: we didn't originally, but if we did that state transition the Tempest cores don't currently have any way of stopping us from implementing the feature 14:28:32 <sigmavirus> rosmaita: it was last night 14:28:41 <sigmavirus> (if my calendar is still accurate) 14:28:47 <stevelle> I don't think that is correct 14:29:05 <sigmavirus> oh it's changed 14:29:09 <sigmavirus> it's in 2 hours 14:29:11 <stevelle> yes 14:29:15 <sigmavirus> http://eavesdrop.openstack.org/#API_Working_Group 14:29:24 <rosmaita> just not sure if it's cancelled for today 14:29:33 <rosmaita> i'm thinking it is, but can't find the email 14:29:34 <stevelle> I dont believe it is cancelled 14:29:35 <sigmavirus> hold, please 14:30:16 <sigmavirus> your call is important to us. A representative will be with you shortly ;) 14:30:32 <sigmavirus> There is 14:30:44 <sigmavirus> (from cdent in #openstack-sdks) 14:30:49 <rosmaita> great 14:31:13 <rosmaita> please scroll to the bottom of https://etherpad.openstack.org/p/glance-ocata-community-images-tempest-continued 14:31:23 <rosmaita> i put 3 questions in there 14:32:12 <rosmaita> just want to make sure these are the correct things to ask the API-WG 14:32:16 <stevelle> the way those are written, it doesn't prequalify that the minor api version is not a microversion 14:32:27 <stevelle> otherwise ++ 14:33:04 <jokke_> rosmaita: A(4) very direct and blunt way :D 14:33:07 <jokke_> waait .... 14:33:52 <rosmaita> stevelle: would you suggest i preface this with, "given that glance currently doesn't do microversions ..." 14:34:30 <jokke_> rosmaita: you can even drop that 'currently' 14:34:34 <stevelle> maybe "doesn't adhere to the WG guideline for microversions" 14:34:53 <jokke_> rosmaita: as that is not even inflight condition 14:35:25 <rosmaita> stevelle: not so much "doesn't adhere" as "has not adopted microversions" 14:35:33 <stevelle> that works 14:35:41 <rosmaita> it isn't a requirement yet, though someone did propose it as a community goal 14:35:51 <rosmaita> (to implement micorversioning in all apis) 14:36:10 <jokke_> that is horrible idea ... do you have the review for that? 14:36:35 <stevelle> don't think it made it past an etherpad stage jokke_ 14:36:43 <jokke_> pfeeew 14:36:45 <rosmaita> jokke_: i think it's on the etherpad 14:36:49 <rosmaita> what stevelle said 14:37:13 <stevelle> it is not in consideration for Pike 14:38:03 <rosmaita> ok, i'll put us on the API-WG agenda for today 14:38:18 <rosmaita> anyone who can make the meeting (1600 utc today), please do 14:38:43 <rosmaita> #action rosmaita put glance community images impasse on API-WG agenda 14:38:54 <rosmaita> any other comments? 14:39:59 <dharinic> here is the patch for the auto transition (the only way as of now that gets us past tempest) https://review.openstack.org/#/c/415306/ 14:40:42 <rosmaita> dharinic: thanks for being on top of this 14:41:02 <dharinic> Sure rosmaita 14:41:27 <stevelle> comment: I'm not happy about the concern jokke_ raised, even if I don't feel it is a fight we should take right now 14:41:36 <dharinic> I know it kind of voids the purpose of "shared", but it does prevent the backward incompatibility as the qa team say. 14:41:47 <jokke_> stevelle: which one? 14:41:58 <stevelle> using the WG guideline as a cudgel 14:42:17 <rosmaita> stevelle: how do you mean? 14:43:14 <stevelle> rosmaita: it's related to the question you posed before: who decides what is OK and what not? 14:43:28 <rosmaita> here's my real worry about this. we can make the change in dharinic 's patch, to pass the tempest tests. but that doesn't necessarily mean that we are living the spirit of the api guidelines 14:43:51 <rosmaita> and i'm worried that if we do follow the guidelines, the entire CI patch is not allowable 14:44:19 <jokke_> rosmaita: we would not, we would be just working around what QA tests 14:44:32 <stevelle> the WG was chartered with a heavy emphasis on consensus and non-binding guidelines. I feel as if this issue with Tempest tests is against the spirit of that 14:44:50 <rosmaita> stevelle: that's well-put 14:45:07 <jokke_> and thus I'm against that approach. I was once convinced why the auto transform is not good idea. I don't think we should do it just because it gets us around what QA tests 14:45:12 <stevelle> but I know the WG internally struggled with that non-binding nature since the charter, so it's not an open-closed thing 14:46:02 <rosmaita> let's move to open discussion 14:46:13 <rosmaita> (we can continue this if there's nothing else) 14:46:17 <rosmaita> #topic open discussion 14:46:41 <rosmaita> i'll put a quick summary of the issue and our questions for the api-wg on an etherpad 14:47:21 <rosmaita> #link https://etherpad.openstack.org/p/glance-ocata-community-images-api-stability 14:47:24 <dharinic> rosmaita: How much time do we have with respect to CI getting in this cycle? 14:47:33 <rosmaita> dharinic: not much 14:47:44 <rosmaita> we need to decide this quickly 14:48:21 <rosmaita> like, within a few days 14:48:33 <dharinic> Yes. Re reading the 2 email responses we got from tempest on the ML, the 2 things they suggest are 1) microversioning 2)auto transition private->shared 14:48:43 <dharinic> okay 14:49:11 <dharinic> Hope the api-wg meeting helps us decide faster 14:50:45 <rosmaita> dharinic: ++ 14:51:56 <hemanthm> We should probably involve tempest folks also when talking to API WG. 14:52:24 <stevelle> ++ 14:52:33 <sigmavirus> hemanthm: good idea 14:53:40 <rosmaita> how should we notify them? 14:53:48 <sigmavirus> the QA folks? 14:53:54 <sigmavirus> Probably poke them in #openstack-qa 14:54:43 <rosmaita> #action rosmaita to poke tempest folks in #openstack-qa 14:56:57 <rosmaita> ok, guess there's nothing new to discuss? 14:57:42 <jokke_> Thanks all, good discussion again! 14:57:48 <jokke_> Starting the year strong 14:57:55 <jokke_> :) 14:58:50 <hemanthm> rosmaita: We aren't making much progress on the rolling upgrades front. 14:59:13 <rosmaita> hemanthm: what's the blocker? community images? 14:59:23 <stevelle> times up 14:59:24 <hemanthm> rosmaita: and reviews 14:59:42 <rosmaita> ok, will note in priority email 14:59:48 <stevelle> we should continue this topic in #openstack-glance 14:59:48 <rosmaita> ok, thanks everyone! 14:59:51 <rosmaita> and happy new year 14:59:59 <rosmaita> #endmeeting