14:01:06 #startmeeting glance 14:01:07 Meeting started Thu Aug 23 14:01:06 2018 UTC and is due to finish in 60 minutes. The chair is jokke_. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:01:09 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:01:11 The meeting name has been set to 'glance' 14:01:15 #topic roll-call 14:01:19 o/ 14:01:21 o/ 14:01:23 o/ 14:01:38 w/ 14:01:40 o/ 14:01:44 * rosmaita bbiab, coffee in final draw down stage 14:02:31 lets wait rosmaita back 14:02:42 sorry about that 14:03:01 no worries 14:03:10 #topic release updates 14:03:20 ok, so the key question is: 14:03:27 will RC-2 be our final RC? 14:03:34 computer says yes ? 14:03:41 I guess it is 14:03:53 That's better. o/ 14:04:00 i haven't seen any bugs 14:04:07 though i did not check this morning 14:04:14 I haven't seen anything release blocking critical popping up, so I'd say RC2 is our final Rocky 14:04:26 no new bugs 14:04:30 I checked last today 14:04:35 ok, cool 14:04:48 smcginnis what do we need to do to make RC-2 the actual release? 14:05:03 \o/ a toast 14:05:17 IIRC we don't need to do anything 14:05:23 \o/ congratulations jokke_ 14:05:41 ? :o 14:05:44 jokke_ yes, i think it's all automatic from here? 14:06:05 i will close the RC-3 milestone in launchpad, though 14:06:08 rosmaita: so IIRC the release tooling will pick up the last tagged rc for the final 14:06:30 ok, great 14:06:49 so since the tag is there, we can go ahead and merge test backports to stable/rocky 14:07:53 ok, next item is periodic tests 14:08:01 looking mostly good 14:08:05 details on the agenda 14:08:14 which i seem to have lost on my desktop 14:08:21 the test stuff doesn't really affect anyways ... I'd wait green light from smcginnis before merging tests stuff ... traditionally the week or two leading to release has been very very limited wat's going in to that branch apart from release blocking bug fixes 14:08:33 ok 14:09:13 one thing is that we are still running the periodic jobs on all 3 components on stable/rocky 14:09:26 rosmaita: Sorry, got pulled away. Releasing RC2 automatically makes it the final release. 14:09:32 the fix for that is to change .zuul.yaml in master in each component 14:09:44 and then backport to stable/rocky 14:09:53 i have a patch up for glance_store master: 14:10:00 https://review.openstack.org/#/c/586334/ 14:10:09 smcginnis has kindly +2d it 14:10:23 And abhishekk has kindly approved it. :) 14:10:29 oh, thanks abhishekk ! 14:10:37 :D 14:10:43 ok, soon as it merges i will propose a backport to stable/rocky 14:10:50 and we can see if it actually works 14:10:59 sounds good 14:11:06 I'd assume _store and glanceclient are fine, just dunno how strict smcginnis wants to be with the release branch for now 14:11:22 running some extra periodicals is not end of the world anyways 14:11:25 jokke_: As far as backporting those before the RC? 14:11:35 well, just the glance_store one 14:11:52 smcginnis: yeah, doing the testing changes into glance stable/rocky 14:11:52 jokke_: Since those are non-functional as far as the delivered product goes, they can go in before or after RC. It really doesn't matter at that point. 14:12:16 ok, cool 14:12:26 smcginnis: ok, so you don't mind the brnach living a bit as long as they are not functional changes 14:12:29 ? 14:13:11 jokke_: Yeah. Don't need to hold up an RC2 waiting on those, but also OK if they do end up in there for that. 14:13:44 ok, so the periodic test results: it's kind of funny, the thing that was motivating me to make sure we run them only in master is that glance_store stable/rocky were all failing, but now they are all green! 14:13:45 smcginnis: we have rc2 and were not planning to tag rc3 for test changes 14:14:22 Oh, right. Then yeah, they can be merged to stable/rocky and nothing more needs to be done. 14:15:18 to continue, glanceclient tests are all green 14:15:31 cool, I think that's our release clear then :D 14:15:44 glance is showing some weirdness in the functional-py35 for oslo and keystone 14:15:51 but, they never fail together in master and stable/rocky 14:16:00 which are pretty much the same code atm 14:16:29 anyway, one thing i do want to mention is the large amount of MoxStubout deprecation messages being spit out 14:16:59 yeah they have been spat out all over the get for past months 14:17:04 we can get rid of them during S? 14:17:07 just throwing that out there, i personally am not interested in dealing with them right now 14:17:44 there was somebody putting up some patches, chuck short maybe? 14:17:58 but he drew the line at the fixtures we are using, they are a bit complicated 14:18:07 we can get rid of them when ever someone finds time and motivation to figure out the changes needed from the current stubout model to what ever needs to be used 14:18:33 ok, I will see if I can find someone to work on that 14:18:49 that's fine, i just wanted to mention them as part of the periodic test report 14:18:57 abhishekk that would be great! 14:18:59 it's all coming somewhere from oslo 14:19:57 so there was this community goal to get rid of all mox stuff, and apparently oslo instead of fixing what the fixtures used underneath decided to throw deprecations out of them 14:20:25 yeah, i think abhishekk got rid of all 'import mox' in our code early in the cycle 14:20:32 yup 14:20:38 yup 14:20:52 we haven't been directly using any since like before milestone 1 14:21:15 are we done with updates? 14:21:24 yes, that's all from em 14:21:27 *me 14:21:36 #topic Open Discussion 14:21:57 edmondsw : you are up 14:22:00 jokke_: Do we have a time slot for the PTG group picture? Everyone going to be there? 14:22:06 rasmaita thanks 14:22:07 Sooo ... I will try to find some time to look into that PTG planning etherpad by monday 14:22:35 so I'm trying to write up some help for an internal team where I work that involves setting up devstack 14:22:36 smcginnis: no we don't and no we won't be having everyone there :( 14:22:47 :( 14:22:48 and as I've been trying things, I stumbled over this uwsgi mess 14:22:48 i have been waiting for to hear a scream clear across the atlantic ocean for one of the topics i put on there 14:23:02 I saw the mail from Kendal now before meeting so I'll drop time for us into the list 14:23:07 sorry, edmondsw, you have the floor 14:23:18 there are patches to make uwsgi work in glance, but they haven't merged 14:23:29 there are patches to make it possible to not use uwsgi in devstack, but they haven't merged 14:23:36 so... we're kinda stuck here 14:23:41 one of those needs to merge 14:23:44 yeah, this has been sitting since the ptg: https://review.openstack.org/#/c/545483/ 14:23:48 i use it personally to run devstack, seems ok 14:24:01 but neutron-grenade hates it 14:24:14 yeah 14:24:27 personally I'd prefer that glance just work in uwsgi 14:24:35 edmondsw: yes, so the patches that claims to make glance working under uwsgi, won't and I forgot to chase to make sure we get the devstack deploying glance without being under Apache 14:24:39 but one way or the other, devstack shouldn't have to be hacked up to work 14:24:40 I think everyone would prefer that. 14:24:57 jokke_ why won't they work? 14:25:15 they break taskflow 14:25:23 and hence image import 14:25:27 and all other tasks 14:25:29 I thought there was a fix for that 14:25:31 edmondsw: yeah IIRC taskflow won't work under uwsgi 14:25:47 mtreinish around? 14:25:56 didn't you have a fix for that? 14:26:04 edmondsw: no, there is fix for running external command (from taskflow) which will never be executed 14:26:19 because the taskflow won't start 14:26:26 yes, that is the situation 14:26:30 edmondsw, these are some other issues with uwsgi 14:26:31 https://etherpad.openstack.org/p/uwsgi-issues 14:26:33 Any way we can move the taskflow part out of the api? 14:26:43 nope 14:26:47 nothing realistic 14:28:26 so what's the path forward? 14:28:54 i think at least get my patch merged so glance works in devstack by default 14:29:08 the other stuff is mostly a resource issue, i think 14:29:12 that'd be a start 14:29:16 edmondsw: I need to make sure that we get the needed changes merged to devstack so it will deploy glance properly 14:29:27 please :) 14:29:28 edmondsw do you have any pull with the qa team? 14:29:44 not that I know of :) 14:29:51 bummer 14:30:12 that was agreed in Vancouver Summit to be the way forward 14:30:18 just need to follow it up 14:30:27 that's right, it was the summit, not the ptg 14:30:34 I think there was some concern over that having to be a staged process to not break grenade? 14:30:38 is that understood? 14:31:18 also, everyone please read my commit message to make sure i am stating things clearly: https://review.openstack.org/#/c/545483/ 14:31:19 based on my skimming http://eavesdrop.openstack.org/irclogs/%23openstack-infra/%23openstack-infra.2018-04-10.log.html#t2018-04-10T21:09:52 14:31:42 edmondsw: I'm not sure if it's fully understood what it might break from grenade but there was good undertanding of what needs to be done so grenade can be forced to get over it 14:32:00 and looking at my commit message prompts me to ask, do we want to say something about this in the release notes for Rocky? 14:32:15 "Note: The Glance release notes for both the Pike and Queens releases 14:32:15 recommend that Glance not be run under uwsgi in production systems." 14:32:47 we could rush through a release note revision today, and tag RC-3 14:32:54 rosmaita: I'm sure that everyone who does not read releasenotes anyways did ignore it already 14:33:10 so no, we won't be tagging rc3 just to have that repeated there again 14:33:12 yeah, but it has been helpful to say, "hey, it's in the release notes" 14:33:14 I guess we already have mentioned that somewhere 14:33:51 it's in past 2 release notes and it's reinforced in multiple places in the docs (that is ignored by those same people) 14:34:04 so we're good 14:34:14 ok 14:34:18 Do we have it mentioned in the docs? 14:34:31 yes, that's what I just said 14:34:50 https://docs.openstack.org/glance/latest/admin/apache-httpd.html 14:34:52 Oh, sorry. Then I think we're in as good of shape as we can be. 14:34:55 it mentions Queens, though 14:35:05 i think we need to update this (just my opinion, though) 14:35:26 otherwise it is not obvious that it applies to Rocky 14:35:44 everyone always thinks docs are out of date 14:35:50 but having said that, i will now shut up about this 14:36:15 rosmaita: well in this case they would be right :D 14:36:35 i can't help myself, i really think we should add this to Known Issues ... and now i will really shut up 14:36:36 we can remove the queens from there after the release 14:36:45 rosmaita: it's not issue 14:37:01 * rosmaita is keeping mouth shut 14:37:10 apart from the fact that people does not read those notes nor docs and try to do something that is stated not supported 14:37:22 edmondsw: what's up? 14:38:07 was talking about the old glance uwsgi issue 14:38:10 mtreinish: we still haven't got devstack fixed by how it deploys Glance ... that's all 14:38:19 edmondsw: you're looking for this: https://review.openstack.org/#/c/549743/ 14:38:31 which should be the approach to get taskflow running under uwsgi 14:38:43 I think jokke_ was asserting that doesn't work 14:38:51 no one has ever tested it 14:38:53 is that what you were thinking of jokke_ ? 14:38:54 it should work fine 14:39:07 I notice it's not mentioned in the etherpad linked above 14:39:23 nor https://review.openstack.org/#/c/545483/ 14:39:38 nope 14:39:52 edmondsw: there is also: https://review.openstack.org/#/c/531498/ which is a related fix 14:41:46 i was confusing the former with the latter 14:41:46 that change actually might have a shot for it ... I did as well 14:42:24 ok, i will make some time to try this out today 14:42:54 rosmaita thanks! 14:43:07 and thanks to mtreinish for tracking this down 14:43:16 yes, thanks mtreinish 14:44:40 ok, continuing general discussion, as far as i can tell the glance channel is no longer being logged 14:44:59 yes, from last week I guess 14:45:06 fungi was looking into it yesterday, but i went off line and can't check the logs to see what he said! 14:45:08 ok, cool 14:45:22 did they run out of space for logs as well? 14:45:22 just mentioning that we may need some followup 14:45:44 no, i think it was the bot didn't have permissions or something afger the registration change 14:45:49 at least that was the first thought fungi had 14:45:53 ahh 14:46:08 the bot logging openstack-unregistered now ;P 14:47:15 So we can say anything we want and there will be no record of it. :) 14:47:42 also I'll try to be there ofr next weeks meeting, won't be on the following 'though (the 6th of Sept or something like that) 14:48:07 smcginnis: I would not count on it :P 14:49:06 telegram etc. reaches if sometihng catches fire, otherwise I try to burn some PTO days before PTG 14:49:27 ok 14:49:35 Should we skip right before the PTG as folks get ready for travel? 14:49:35 hopefully things will be pretty quiet 14:49:42 Or things to discuss to be ready for the PTG? 14:49:49 jokke_, enjoy, have a nice time 14:50:32 Like said earlier I'll go through that ptg planning etherpad by monday and start finilizing the schedule for PTG 14:50:54 ok, good, so we can finalize next week and take week after off 14:51:33 I've been most of this week off and plan to do the same next week and then be totally off grid the week after :D 14:51:57 man, how much PTO do you get? 14:52:10 * rosmaita is jealous 14:52:21 Fri next I will be driving ~850km North for the start of the Moose hunting season :D 14:52:43 won't that put you in the middle of the north sea? 14:53:02 guess you are in finland 14:53:11 rosmaita: including bank holidays (Irish version of the federal holidays you guys got) I think it's somewhere between 30-35 days a year 14:53:11 I am going offline now before change of my mood to work :P 14:53:29 abhishekk: o/ 14:53:52 jokke_ not bad at all 14:53:57 yeah I'm in Funland ... there is still room to get about 1000km north before hitting the sea 14:53:59 smcginnis, \o/ 14:54:08 abhishekk don;'t tell me you are in the mood to hunt some moose! 14:54:41 rosmaita, definitely but I need to travel 5000 km to north for that 14:54:47 LOL 14:54:49 :) 14:55:46 yeah, that's it, thank you all (at least I will hunt a beer not bear in remaining time) 14:56:15 sounds good 14:56:19 rosmaita: we seem to have a race in meetbot exposed by the registration enforcement 14:56:20 before you go, i could use some input on https://review.openstack.org/#/c/594769/ 14:56:33 see my last comment on the patch 14:56:42 thanks all and I'll talk to you next week! 14:56:46 rosmaita: will do 14:56:46 it's a glanceclient thing to use multihash for data validation 14:57:07 fungi thanks for looking ... race conditions are always fun! 14:57:09 will do 14:57:10 basically it starts trying to join channels immediately after identifying, before it gets back confirmation that it's identified, so the first channels it attempts to join don't work... i'm trying to figure out a solution 14:57:41 fungi: easy, just like any great sw project, just hardcode sleep in between :D 14:57:57 heh 14:58:15 99% of the work is finding where to add the sleep call ;) 14:59:23 ok, thanks folks 14:59:27 #endmeeting