14:00:15 #startmeeting Glance 14:00:20 Meeting started Thu Feb 25 14:00:15 2016 UTC and is due to finish in 60 minutes. The chair is flaper87. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:00:21 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:00:23 The meeting name has been set to 'glance' 14:00:25 o/ 14:00:26 o/ 14:00:29 o/ 14:00:36 o/ 14:00:38 o/ 14:00:51 o/ 14:01:00 o/ 14:01:02 o/ 14:01:06 o/ 14:01:17 ok, I'd call that a quorum :D 14:01:20 #topic Agenda 14:01:23 #link https://etherpad.openstack.org/p/glance-team-meeting-agenda 14:01:24 o/ 14:01:28 There, as usual 14:01:49 where is agenda? :) 14:01:58 Everything that's not in the agenda, let's keep it for Open Discussion 14:02:04 ah 14:02:11 there 14:02:16 I forgot to update the date 14:02:19 :D 14:02:20 yup, that's better 14:02:20 #topic Updates Glare 14:02:34 okay, let's begin 14:02:49 first, as it was promised the spec is here 14:03:05 #link https://review.openstack.org/#/c/283136/ 14:03:09 ++ 14:03:32 after that I have several meetings with folks from murano and app-catalog 14:03:55 they gave me several advice, but in general they like the proposal 14:04:24 today we're going to join api-wg meeting with Nikhil 14:04:26 ok, We can start reviewing those specs in a couple of weeks from now. As soon as FF is over and we have a tag 14:04:32 that sounds great 14:04:43 joining the API-WG meeting is definitely good 14:04:47 after that I'll update the spec based on all comments 14:05:21 second, Ina Vasilevskaya is back after maternity leave 14:05:40 And she's happy to join Glance/Glare team again 14:05:46 w0000h000, glad to have her back :D 14:06:21 there are several administrative issues 14:06:45 but should get better next week. 14:06:58 ok, got it! All good news 14:07:02 thanks for working on the spec 14:07:12 np, it's my job :) 14:07:21 let's know the feedback from the api-wg 14:07:24 anything else? 14:07:36 there was some ping 14:07:47 about plugins requirement from murano 14:07:56 there is a couple of minor things 14:07:58 and that hard coding was tough for them 14:08:11 but that's still in discussion I sup? 14:08:20 plugins as in API extensions ? 14:08:49 about plugins and API extensions 14:08:50 hm, not exactly. loosely coupling of the code with those of the artifacts types 14:08:56 We don't like it 14:09:24 currently we want that all plugins will be placed in glance/objects folder 14:09:32 mmh, understood. I'll have to read their request a bit better 14:09:34 like oslo.vo requires 14:09:36 I just wanted to point this out, I did not intend to start a discussion (ftr) 14:09:45 ok 14:09:46 but Murano needs some external 14:09:56 let's move on. We can discuss this further on the spec 14:10:11 I'll told you about our decision after this meeting in glance channel 14:10:27 ok 14:10:37 #topic updates nova v1->v2 14:10:42 There's not much to update here 14:10:48 This work was moved to Newton 14:10:52 yeah 14:11:04 the one thing I did want to mention is that we need to schedule a session or something with the nova team 14:11:06 code is still there, no review at all 14:11:20 flaper87: ++ 14:11:32 Also, we need to start the work on glanceclient to have that compatibility layer 14:11:45 flaper87: is it a final decision? 14:11:52 mfedosin: what part? 14:12:00 compatibility layer 14:12:26 oh, yeah. I mean, we were going to do it anyway. The thing is we wanted to postpone it for after the nova migration happened 14:12:49 We need to work on a compatibility layer that favors v2 and doesn't break v1 14:13:06 and then release it in glanceclient so we can move nova to v2 in N-1 14:13:23 The sooner we start this, the better 14:13:32 I see 14:13:42 So we need to change Mike's code, do we? 14:13:54 kairat: I hope no 14:13:55 And implement migration from scratch> 14:14:03 Ok 14:14:03 no 14:14:09 That's good 14:14:17 we can take the code from Nova and put in the client 14:14:26 The compatibility layer should (hopefully) fit into what Mike did 14:14:45 mfedosin: sorta, we might want to change it a bit 14:14:47 we just need to replace several utilities 14:14:50 anyway, that's the gist 14:14:58 We should have a spec for this, tbh 14:15:02 who wants to take this on? 14:15:13 * mfedosin hides 14:15:20 mfedosin: I won't let you >.> 14:15:27 you have too many things on your plate 14:15:29 :P 14:15:34 no one? 14:15:39 ping? :D 14:15:41 I want to do it 14:15:45 Cyril's here? 14:15:47 nikhil: w000h000 14:15:49 but don't think people will like my dir in first place 14:15:58 mfedosin: he's not here, I could ping him 14:16:02 mfedosin: yeah 14:16:08 not the right nick though 14:16:09 oh, he is 14:16:11 :D 14:16:13 flaper87: it was fast 14:16:27 nikhil: why's that? 14:16:37 Steap: interested in taking the compatibility layer on? 14:16:44 flaper87: I would like to know the details on putting the decision to add compat layer to client 14:16:53 ++ to nikhil 14:16:56 Wasn't there a feature freeze that blocked us on this ? 14:16:56 Steap: writing the spec, actually 14:17:06 Steap: it's newton work 14:17:16 we need to write the spec 14:17:16 is this going in Nova or Glance? 14:17:23 seems like we need at both places 14:17:29 or glanceclient ? 14:17:39 nikhil: why would we have the compatibility layer in both places ? 14:17:40 it's common for client 14:17:50 nikhil: not really. On reason I'm asking for a spec is to also formalize this discussion 14:17:54 Steap: I meant a spec in both places 14:18:02 There have been several discussions about this 14:18:04 nikhil: oh right 14:18:05 flaper87: gotcha 14:18:07 especially on IRC 14:18:13 but nothing is in a place we can reference 14:18:15 oh 14:18:17 nikhil: a "define the layer" spec in glanceclient and a "use the layer" in Nova ? :p 14:18:21 so, this would be a good way to do that 14:18:26 dates would work! :) 14:18:40 flaper87: ah kk 14:18:41 I'm on the side that thinks this should go into glanceclient 14:18:50 makes sense to me 14:18:57 Steap: I think nova is worried about long term eval of this 14:19:08 and for client it's mostly about how to structure the code 14:19:10 and nova shouldn't have any knowledge of this (a.k.a nova.images.glance shouldn't exist) 14:19:13 nikhil: what do you mean ? 14:19:17 it will be impact on developers 14:19:18 flaper87: yeah 14:19:24 flaper87: it should be abstracted in glanceclient 14:19:53 Steap: yup, that's what we need to work on in glanceclient 14:19:57 ok, let's move on 14:20:08 I am skeptical to put this in client tbh (but will have to read the logs) 14:20:09 WE can discuss this further in -glance 14:20:28 nikhil: I wish I remembered when this was discussed, really. :( 14:20:33 I remember some in-person discussions 14:20:39 np 14:20:46 And also over the reviews of mike's work 14:20:51 and IRC with other nova folks 14:20:53 anyway 14:20:59 We need to formalize this discussion 14:21:03 hence the request for a spec 14:21:13 #topic Cross prj updates 14:21:16 o/ 14:21:28 Just a small update 14:21:47 The Cross prj initiative on quotas has started getting momentum 14:22:01 watch out! 14:22:04 initial spec 14:22:10 #link https://review.openstack.org/#/c/284454/ 14:22:18 and a weekly meeting 14:22:28 #link https://review.openstack.org/#/c/284478/ 14:22:29 that's good news 14:22:38 I will announce to ML in a few 14:23:15 Nothing to discuss on the spec itself yet 14:23:29 nice, glad to see this moving forward 14:23:35 if there is interest, I can make it a point to buzz you prior to mtg :) 14:23:38 it's been a long standing open task for several projects 14:23:38 (done) 14:23:51 agreed 14:24:07 I'm interested but I don't think I've the bandwidth to follow that mtg too :( 14:24:15 plus, you'll update us anyway 14:24:17 * flaper87 ducks 14:24:28 * flaper87 ducks two times 14:24:34 ok moving on 14:24:37 heh 14:24:44 #topic Design Summit sessions allocation 14:24:48 #link http://lists.openstack.org/pipermail/openstack-dev/2016-February/087433.html 14:24:56 Sent that email... yday I htink 14:24:58 think, even 14:25:06 pls, if you have an opinion, do chime in 14:25:17 nikhil: proposed we increase the number of workrooms to 6 14:25:20 we had 5 in tokyo 14:25:33 I did 14:25:43 What are the tentative FB sessions for us? 14:26:30 OTOH, I guess it'd be something like Import Refactor, Quotas and something else (perhaps the compat layer so we can invite osclient folks ) 14:26:40 but again, nothing has been discussed yet on this 14:26:53 Maybe Glare will require 2 :) 14:26:58 hmm 14:27:02 I'd love to talk more with ppl that are going to run for the PTL position 14:27:06 it's better to plan that way 14:27:17 I think we will need 1/2 Workroom session for tasks (tbh) 14:27:22 mfedosin: I think Glare can use a workroom session 14:27:39 In tokyo, we made a very good use of the meetup time 14:27:43 flaper87: he meant 2 W for Glare (I think) 14:27:46 we split the time and the team 14:28:03 mfedosin: oh, ok. weeeeeeell, I guess we could use 1 and some meetup time 14:28:26 It's quite blury still. I'll start putting more serious thoughts on this and come up with a plan for next week 14:28:32 Since next week is FF anyway 14:28:35 I wanted to have a session about Glare architecture and another one with murano, app-catalog and heat 14:28:53 That said, I want us to be very very careful with requesting more time as there are more projects this time around 14:28:56 oh then , prolly 1 FB and 1 W 14:29:01 so, just trying to be a good citizen 14:29:09 yes, makes sense 14:29:21 I was trying to be cautious that we didn't miss on imp topics 14:29:25 mfedosin: in tokyo we shared the meetup time with app-catalog too 14:29:33 we are making some changes to glance and they are big ones! 14:29:37 it was on Friday 14:29:42 for example a lot of folks ask me about tasks 14:29:53 and image sharing 14:30:02 Also, considering the size of the glance team, there's so much we can plan for Newton, which means limiting the number of session might end up being good 14:30:04 hahaha 14:30:17 flaper87: nikhil: is friday set aside for working meetings like in tokyo? 14:30:36 I think the contri meetup is on friday 14:30:39 That said, we should also use the time at the summit to discuss things that *must* be discussed face to face 14:30:41 rosmaita: yes 14:30:50 rosmaita: we have a full day 14:30:54 just like in tokyo 14:30:58 ok, everyone please make travel plans so that you can spend a full day 14:31:07 ah yeah, that too 14:31:13 do not live on friday at 8am 14:31:17 we got a lot done in tokyo (though you wouldn't know it from the import refactor) 14:31:22 leave* 14:31:28 but surely live :D 14:31:31 lol, leave* 14:31:33 hahaha 14:31:36 nikhil: thanks :D 14:31:42 ;) 14:31:52 my brain is full of Freudian slips 14:31:56 rosmaita: I agree 14:32:05 I found the meetup in Tokyo to be super super helpful 14:32:12 ok 14:32:15 put some thoughts 14:32:26 I'll start an etherpad and send an email so we can start collecting topics 14:32:34 I think the next week in Austin is some sort of festival 14:32:42 so that might be worth considering as well 14:32:49 The week after the summit? 14:32:53 I think so 14:33:04 nnnnnice, good thing I'm booked for like a gazillion of days there 14:33:06 hahahaha 14:33:09 * nikhil double checking (remembers vaguely) 14:33:14 anyway, moving on 14:33:28 #topic Non client libraries freeze 14:33:49 Unfortunately, I think I forgot to mention this last week but this week is the non client libraries freeze 14:34:13 This means that this week libraries like glance_store can be released for the last time during the Mitaka timeframe 14:34:22 #link https://review.openstack.org/#/c/251851/ 14:34:28 Some patches landed yday 14:34:34 so kairat work with Swift driver moved to Newton 14:34:45 it's sad 14:34:48 but there are 2 big ones for the "trusts" work that still can make it 14:34:55 mfedosin: check the link >.> 14:35:02 flaper87: ah 14:35:02 We have until tomorrow to release it 14:35:09 oh sorry, I was too tired last night to review that one 14:35:21 I never do releases on friday but I'm happy to stay "on call" during the weekend in case things go south 14:35:22 flaper87: is that still on? 14:35:45 (I'll be online during the weekend too) 14:35:50 not sure if sabari is around but there was also the vmware on python-requests one 14:35:51 i got some comments from mclaren, but nothing more then. 14:36:07 so it would be perfect to have some comments tomorrow 14:36:08 kairat: it's a big one :) 14:36:33 I am personally will try to review all patches for glance_store tomorrow 14:36:33 actually big two 14:36:37 I'll review them with my limited knowledge of how the swift store works 14:36:41 mfedosin: right 14:36:47 it's always a problem with big patches 14:36:53 here's the second part https://review.openstack.org/#/c/251850/11 14:36:57 no promises made, it's possible they won't make it 14:37:02 can we have a list of the ones we are targeting ? 14:37:11 mfedosin: I posted the top most one 14:37:21 nikhil: I think just that one, TBH 14:37:34 I don't think we've bandwidth for anything else today/tomorrow 14:37:35 great 14:38:15 ok, moving on 14:38:31 #topic Preparing for Feature Freeze 14:38:43 Now, next week is FF 14:38:50 yup, it's that time of the cycle again 14:38:55 crazy, isn't it? 14:38:59 Anyway 14:39:09 we don't have that many pending features in glance 14:39:24 filters 14:39:30 https://review.openstack.org/#/c/248359/ 14:39:31 mfedosin: I said "that many" 14:39:42 there are 6 filters :) 14:39:44 * flaper87 takes mfedosin's keyboard 14:39:51 hahahha 14:39:53 jokes apart 14:40:05 yeha, filters is perhaps the biggest one we have now 14:40:21 so, I wanted to propose having a review sprint on Monday (as early in the week as possible) 14:40:45 ++ to review day 14:41:03 filters review day :))) 14:41:05 In addition to that day, I was thinking we should have reviews days every week until the end of the cycle 14:41:12 but I'll let that for open discussion 14:41:14 :D 14:41:21 I would love it personally 14:41:48 All that said, I'm hoping to cut M-3 on Wednesday so, let's target that as a deadline 14:42:03 It'd be awesome to be able to do it on Tuesday but that might be too tight 14:42:21 Questions ? 14:42:28 otherwise we can move on 14:43:21 ok, moving on 14:43:25 #topic Spec lite proposal 14:43:30 #link https://review.openstack.org/#/c/282517/ 14:43:33 jokke_: wrote this 14:43:35 :D 14:44:16 The proposal is quite small 14:44:20 and easy to review and adopt 14:44:28 LEt's drop comments there 14:44:44 if there are no complaints, I'll proceed with approving it 14:45:14 ok, unless there are comments, I think we can go into Open Discussion 14:45:27 or as I prefer to call it: "Go nuts section" 14:45:37 do we say why we're changing the spec lite process? 14:46:01 mclaren: I think it's in the commit message 14:46:13 if it's not, we can then put it there 14:46:50 unless I misunderstood your question and you're actually asking why we're changing it 14:47:31 mclaren: ^ 14:47:33 I'm intrested why 14:47:52 I don't see anything in the commit message. It adds overhead, so explaining the advantages would be useful 14:48:06 Perhaps it is worth mention in spec itself 14:48:18 to explain future supporters why it was chosen 14:48:32 and not to spent their time 14:48:43 The TL;DR is: Using bugs doesn't seemed to work. Triaging them was sometihng we did in the drivers meeting which we don't have anymore. Using specs encourages a more distributed review process across core reviewers 14:48:56 but yeah, I think we should have this in the spec 14:48:56 let's just slap that into the commit message 14:49:03 erm, commit message 14:49:16 I don't think we need to put this in the contributions guidelines 14:49:25 +1 to keep out of guidelines 14:50:20 I'll ping jokke_ and ask him to write this down 14:50:30 also ... moving it to the glance-specs repo, we get all our specs listed in the same website 14:50:35 maybe not using his words because we know what that commit message will look like 14:50:38 jokke_: oh, HAI! 14:50:41 ;) 14:51:08 jokke_: ah listing, that was the other one. That one is actually super useful 14:51:28 ok, 10mins left 14:51:28 so one does not need to go hunting closed bugs and merged specs to figure out what new functionality has gone in 14:51:47 #topic Open Discussion 14:51:52 what about deprecating use_user_token? 14:51:54 https://review.openstack.org/#/c/237742/ 14:52:04 So, I mentioned that I'd like to have 1 review day per week until the end of the cycle 14:52:14 I obviously don't expect everyone to sign up for this 14:52:24 but it'd be cool to establish a day for this. 14:52:35 The upcoming weeks will require bug fixing and reviews like crazy 14:52:44 i guess monday is as good as any 14:52:57 rosmaita: 14:53:00 ops 14:53:01 rosmaita: ++ 14:53:08 mfedosin: that's a breaking change afaict 14:53:27 on another note, a CVE fix broke something on how locations work: https://review.openstack.org/#/c/280789/ 14:53:39 we need to help reviewing that fix 14:53:41 jokke_: ^ 14:53:56 flaper87: I'm aware of that :( 14:54:07 jokke_: :( 14:54:10 do we have trusts now to avoid using that option? 14:54:24 mfedosin, ^ 14:54:29 yes 14:54:35 Ok, good 14:54:44 this option was introduced to work with registry 14:54:47 kairat: I think in some deployments may not be able to use trusts 14:54:52 when token expires 14:54:57 correct 14:55:08 but it doesn't work 14:55:17 so there's a dependency here 14:55:24 if you disable it, you cloud will die 14:55:36 Also v2 is deprecated already 14:55:40 because every person will behave as an admin 14:55:44 it works for when the operation is concise (like trusts) performed using nova 14:55:48 we need to consider that during Newton 14:56:15 please read 14:56:17 #link https://wiki.openstack.org/wiki/OSSN/OSSN-0060 14:56:26 for long running operations if the token is invalid and then this will allow the image to go to active (given you have the right setup) 14:57:41 but there's no getting rid of it yet so those who are using it will keep using it 14:57:51 is the intention to remote it in Mitaka like the OSSN says? 14:58:15 and that's what I am aiming to talk about 14:58:31 no 14:58:36 sorry to interrupt, but looks like we are running out of time ... please open your browsers to: 14:58:36 if we remove it then we need to send proper communication that people need to adopt trusts 14:58:39 we will remove it in Newton 14:58:39 #link http://lists.openstack.org/pipermail/openstack-dev/2016-February/087519.html 14:58:42 it will take you 3 minutes to read, there's a link to an etherpad with more stuff if you want to dig, but it would be good to have a consensus on what the recommended behavior should be, since it's different in v1 and v2 14:58:49 in Mitaka we deprecate it 14:58:49 I need to take a better look at mfedosin's patch 14:58:57 rosmaita: ++ 14:59:02 I remember that bug (btw) 14:59:21 and I think hemanthm agreed with the deprecation 14:59:30 rosmaita: yeahm thanks. we need to remember about it in compat layer 14:59:53 ok, that's it 14:59:56 thank you all! 14:59:57 I am just not sure the wider impact on this so removing in newton might be okay, let's start the broadcast on the deprecation this release 14:59:58 tty next week 15:00:03 and REVIEW ALL THE PATCHES 15:00:03 ok, thanks 15:00:06 PRETTY PLEASE!!!!!!!!!!! 15:00:11 #endmeeting