14:01:02 #startmeeting glance 14:01:03 Meeting started Thu Feb 1 14:01:02 2018 UTC and is due to finish in 60 minutes. The chair is rosmaita. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:01:04 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:01:06 The meeting name has been set to 'glance' 14:01:11 #topic roll call 14:01:15 hello everyone! 14:01:18 o/ 14:01:20 hello 14:01:39 o/ 14:01:43 o/ 14:02:37 well, guess we might as well get started 14:02:46 #link https://etherpad.openstack.org/p/glance-team-meeting-agenda 14:02:55 #topic updates 14:03:13 ok, after much anguish, Q-3 was released early this week 14:03:22 i don't even remember what day, maybe tuesday 14:03:31 anyway, i sent out something to the ML about that 14:04:01 the key thing to keep in mind is that the gate to merge stuff has not gotten any faster 14:04:07 o/ 14:04:20 still up near 24 hrs rosmaita? 14:04:49 i think something got through yesterday in 18 hrs 14:04:54 so maybe it is getting better 14:05:04 hm still pretty unreasonable 14:05:23 yes, i think there will be a lot of discussion at the ptg 14:06:00 I know there was one of the cloud providers that gave us a higher quota, but if we actually used them all it was overloading the network. 14:06:08 So they've dropped that down. 14:06:16 yeah, there are a lot of moving parts 14:06:27 Rather than having a higher number running at a time, but slower, we'll have a smaller, faster number. 14:06:32 and improving one thing can degrade other stuff 14:06:32 At least that's the hope I think. 14:07:00 o/ 14:07:10 yes, and i should stress that the infra team have been busting their butts 14:07:33 there were several unfortunate occurrences that messed things up 14:07:49 plus maybe the first time zuul3 was under extreme load conditions 14:08:19 anyway, i have taken the liberty to reach out to the infra team for a "consulting" session at the PTG 14:08:30 just want to go over our actual jobs with them 14:08:38 make sure they're configured correctly 14:08:42 good plan rosmaita, hopefully they can provide some guidance 14:08:50 and maybe see if we can do something better 14:09:01 i hope so, have not heard back yet 14:09:04 also, 14:09:16 i took the liberty of reaching out to the keystone team 14:09:24 want a consulting session with them too 14:09:50 good to have 14:09:51 where we explain what our problems are (particularly with policies and image locations) and figure out what is available to help us 14:10:13 want to make sure that whatever we want to use isn't planning to be obsoleted or anyghint 14:10:16 *anything 14:10:38 i did hear back from lance, they will schedule something with us early in the week, monday or tuesday 14:10:50 with the option to maybe follow up later in the week if necessary 14:11:04 i have gone off-topic 14:11:14 anyway, don't forget the ptg planning etherpad: 14:11:27 #link https://etherpad.openstack.org/p/glance-rocky-ptg-planning 14:11:41 ok, so with Q-3 out the feature freeze is in effect 14:12:02 and bugfixes at this point should only be release-critical 14:12:23 though, i will interpret that a bit liberally 14:12:57 there is a FFE request already, we will discuss later 14:13:08 the image import stuff is also automatically FFE'd 14:13:25 i don't think we will be able to remove v1 this cycle, however 14:13:32 :( 14:14:12 but, removing v1 and merging the multihash stuff should be undertaken as soon as possible after stable/queens is cut 14:14:18 well, let's say 14:14:21 multihash first 14:14:49 and actually, the glance_store part of multihash can be done as soon as we have reviewer bandwidth 14:15:04 sounds good 14:15:05 but i am going off topic again 14:15:12 probably need to review that a bit with someone as the code was finished a while ago 14:15:19 make sure we won't create merge conflicts and the like 14:15:23 McClymontS makes sense 14:15:41 good news is I have .patch files for both repos so the changes are evident 14:16:31 on that note, there are some patches that need reviews pronto 14:16:41 https://etherpad.openstack.org/p/glance-quick-reviews 14:16:47 #link https://etherpad.openstack.org/p/glance-quick-reviews 14:16:56 i would ninja them in, but most of them are mine 14:17:15 i bring it up now because one of the patches is functional tests in glance_store 14:17:31 we've always had some, but they've always been experimental 14:17:49 and they still are, but i had to rewrite them a bit to get them to work again 14:17:59 * jokke_ is bit multitasking with another meeting atm. 14:18:14 anyway, abhishekk and jokke_ please check out that etherpad at your earliest convenience 14:18:27 rosmaita, ohh I have added one patch in that 14:18:33 I will review when I have time as well 14:18:42 abhishekk good! 14:18:44 McClymontS ty 14:19:00 last update 14:19:17 PTL self-nomination period ends on Wednesday 14:19:28 so far, erno has put up a patch 14:19:42 i guess everyone has seen my note on the ML? 14:20:06 a few weeks ago i said that i had the support of my management to be PTL again, which was entierly true 14:20:10 i did have support 14:20:11 will do 14:20:25 but i decided it would be better to let someone else do it for a change 14:20:29 before i burn out 14:20:41 thank you rosmaita for your organized efforts 14:20:43 like i said in the email, i'm going to continue to work on glance 14:20:55 but concentrate on one or two things instead of 18 14:21:04 Thanks for all your work rosmaita. Better to not get burned out. 14:21:06 abhishekk yw 14:21:23 smcginnis ty 14:21:37 yes, we've got a good community and i want to continue working with y'all 14:22:01 ok, that's all for the updates 14:22:15 #topic Update the error response code in document 14:22:18 Brin this is you 14:22:33 I am here 14:22:47 great, tell us something about this 14:22:51 Update the error response code in document, consistent with cinder and nova document libraries.(e.g. 400 to badRequest(400)) (Brin Zhang 14:23:07 I think here need to be consistent with cinder, nova api documentation 14:23:20 because this way, the user can see a clear http error, rather than an error code, e.g.400,500,503 14:23:35 do you think so? 14:23:50 so you mean just in the documentation? 14:24:01 yes 14:24:30 so can you show an example? here's our docs: https://developer.openstack.org/api-ref/image/v2/index.html#images 14:25:10 maybe this is better: https://developer.openstack.org/api-ref/image/v2/index.html#create-an-image 14:25:19 Here is an example: https://review.openstack.org/#/c/527345/3/api-ref/source/v2/qos-specs-v2-qos-specs.inc 14:25:34 You can fllow this link :https://developer.openstack.org/api-ref/image/v2/index.html#create-image-member 14:26:04 My personal opinion for Cinder was the codes were mostly fine, but if someone wants to maintain the string representations, that may make it a little easier for some. 14:26:16 I don't know what is the Error response codes: 400, 401, 403, 404, 409, 413 14:26:20 So kind of a "meh, why not" in the end. :) 14:26:32 i remember working on a table for these back when anne gentle was docs ptl 14:26:44 i eventually abandoned it 14:27:34 anyway, my only concern is, is there a table somewhere of the "official" strings? 14:28:03 and why is "Request Entity Too Large(413)" not all one word like the others? 14:28:21 in other words, i don't object to making the docs more clear, i just don't want a lot of churn 14:28:28 i.e., fix it and then re-fix it 14:28:42 on the other hand, doc churn is not like code churn 14:28:51 so what the heck 14:28:52 I think this is better for users who read api 14:29:12 Instead of just seeing a single figure 14:30:22 this is the table thing i was talking about: https://review.openstack.org/#/c/349551/ 14:31:29 ok .. what did I miss 14:31:38 Hah, line 66. :) 14:31:41 rosmaita: yeah, I think it good. 14:31:59 i forget why we never followed up with that 14:32:14 other than it being a bit of work 14:32:38 anyway, i personally have no objection if someone wants to update the docs as Brin suggests 14:33:24 Brin are you interested in doing this, or is your aim to nudge projects to do this themselves? 14:34:17 For future development, I think this is something to do 14:34:51 I am willing to take the time to try this thing 14:35:11 OK, sounds good 14:35:18 so just finished with backlog 14:35:19 This should be done on the premise that everyone thinks it needs to be done 14:35:41 OK 14:35:45 ok 14:35:46 I'm not sure if that's worth of the time 14:36:02 well, not our time 14:36:15 but if someone new wants to take it on, what the heck 14:36:23 the API ref is for developers who are interacting directly with the API 14:36:26 the numbers have well-defined text to go with them 14:36:26 This is not a problem, given the time allowed :) 14:36:41 so it's not like we'll have to review really carefully 14:36:42 I think the list of codes that doc has atm. is really quick to scan through what to expect 14:37:03 and when you're developing REST client, you gotta either have cheatsheet or you know these 14:37:07 ok, tell you what 14:37:15 this would be a good topic for a spec-lite 14:37:18 Since you say that, it is not necessary 14:37:20 our clients does spell them out 14:37:27 for endusers 14:37:31 Can proceed to the next topic 14:37:34 instead of throwing just a code 14:38:01 ok, so opinion is mixed 14:38:09 so just wondering what the value is and if we spell them out, please make sure they do not become unreadable 14:38:11 yeah 14:38:24 like one per line or something that is quick to look through 14:38:34 that's the advantage of the rest_status_code table structure 14:38:52 Brin why don't you look into that and follow up with us? 14:39:21 ++ 14:39:28 #topic FFE request for --check feature 14:39:35 here's the info: 14:39:49 #link http://specs.openstack.org/openstack/glance-specs/specs/untargeted/glance/lite-spec-db-sync-check.html 14:39:54 #link https://review.openstack.org/#/c/455837/8/ 14:40:07 Perhaps I ignored this point, after the meeting I will conduct in-depth research 14:40:16 thank you. 14:40:17 thanks, Brin 14:40:38 ok, key point here is (1) it's a spec lite, and (2) it's already been coded 14:40:49 and it will be very useful for operators 14:40:55 so i am inclined to grant an FFE 14:41:03 yeah 14:41:07 +1 14:41:08 +1 14:41:12 Looks low risk with some benefit. 14:41:15 thnaks brin and rosmaita 14:41:31 ok, cool, sounds like FFE should be granted 14:41:37 that was easy! 14:41:46 I think that is crucial with our EMC approach that the operators have tools to verify on which stage they are ... defo +2 for that 14:41:48 thanks. 14:41:56 #topic policy-in-code community goal 14:42:04 http://lists.openstack.org/pipermail/openstack-dev/2018-January/126827.html 14:42:09 thank you :) 14:42:19 ok, so lance sent out an email wondering what's up with our status 14:42:28 i haven't responded yet 14:42:45 because as far as i;m concerned, it's very low 14:42:50 priority, that is 14:42:55 two things: 14:43:10 you need a policy file to effectively use property protections 14:43:17 these changes don't touch that at all 14:43:29 and they really cant 14:43:50 basically, i am going to reply that we're not doing it this cycle and i can take the heat for that as Queens PTL 14:44:12 plus, we have the policy changes we want to make 14:44:29 so i don't see the point in doing this right now 14:44:50 agree 14:45:03 #link https://review.openstack.org/#/c/528021/ 14:45:14 yeah, we definitely should catch up with Lance in PTG and walk through where this is going to impact us 14:45:16 ^^ that's what i mean by "policy changes" 14:45:33 and why this hasn't been done 14:45:38 jokke_ i will make sure i say that in my response to the ML 14:45:54 I think it's easier to walk through in person than trying to put into an email 14:46:18 yes, the email will just say (1) not doing it in queens, and (2) let's discuss in person in Dublin 14:46:23 Glance policies are unfortunately complex af atm 14:46:37 rosmaita: sounds good 14:46:38 don't see how it affect property protections, maybe i missed something, it just change defaults from policy files to code 14:46:57 kairat it makes them asymmetrical 14:47:10 we also better get rid of policy layer 14:47:11 kairat: it will not achieve it's goals as property protections are relying on that file anyways 14:47:18 if possible=) 14:47:31 kairat exactly, i think the layer riddance should be done first 14:47:38 ++ 14:48:06 ok, that's everything on the agenda 14:48:07 ++ 14:48:10 #topic open discussion 14:48:38 Did we ever figure out the system package upgrade that was causing the services to fail to start with the functional tests? 14:48:53 smcginnis yes, erno put up a patch, it has merged 14:49:00 Awesome. 14:49:00 it was an eventlet problem 14:49:04 ok if anyone missed it, I threw my name into the hat for PTL after lengthy discussions with various people including Brian 14:49:13 We may need that backported. I saw some stable-maint checks failing. 14:49:15 fixed in eventlet 0.22, but we are u-c at 0.20 14:49:16 http://lists.openstack.org/pipermail/openstack-stable-maint/2018-February/007232.html 14:49:41 Not sure if it's the same thing or not, but that will have to be sorted before any stable releases can be done most likely. 14:50:00 jokke_: +1! 14:50:01 there is still time until tomorrow to nominate yourselves, otherwise you'll get me as default :P 14:50:19 actually, the deadline is wednesday! 14:50:20 jokke_, +2 14:50:33 and jokke_ thank you for stepping up 14:50:38 rosmaita: oh, cool 14:51:01 they monkeyed with the schedule this cycle, not sure why 14:51:27 so this is as much for me as for the community like I lined out in my nomination. I've been past couple of cycles all over the place and done half good job everywhere burning myself out as well 14:51:49 smcginnis i guess they run those stable-maint jobs with the updated images? 14:51:55 so I need to focus more what I do and this is clear indication where I need to put my focus in 14:51:57 rosmaita: I believe so. 14:52:09 yeah, then they are all broken 14:53:28 i'll put up a patch cherry-picking the change into stable/pike and stable/ocata 14:53:38 ++ 14:53:55 so one thing I can promise ye all. If you'll take me as PTL, I won't be writing the next big feature we want to implement in Glance during my term, but I'll do everything I can to enable you guys do it ;) 14:54:04 #action rosmaita patch stable/pike and stable/ocata 14:54:10 jokke_ ++ 14:54:38 :D 14:54:46 rosmaita: and thanks once more for your great work, support and mentoring 14:55:03 glad to be of service :) 14:55:13 ok, anything else? 14:55:25 copy from 14:55:37 yes, so that has a FFE 14:55:41 as part of image import 14:55:43 are you guys ok still pushing that in for the release? 14:55:56 i definitely am 14:56:01 cool ... just wanted to bring it up in public if anyone has heavy objections 14:56:02 +1 14:56:05 at this point 14:56:10 because it will mean we can remove v1 at beginning of rocky 14:56:15 yes 14:56:20 otherwise, we'll be in same position we are now 14:56:31 and were last cycle 14:56:39 so I am gung-ho on image import copy-from 14:56:44 and the cycle before that 14:56:48 :) 14:57:37 jokke_, kindly review https://review.openstack.org/523366 when you get time 14:58:37 abhishekk: will do 14:58:54 jokke_, thanks 14:59:15 ok, i guess that's all for this week ... thanks, everyone! 14:59:22 abhishekk: sorry, have slipped from me 14:59:24 Thanks all 14:59:26 Thanks all! 14:59:37 jokke_, no issues 14:59:41 thank you all !!! 15:00:00 #endmeeting