14:00:15 #startmeeting glance 14:00:19 Meeting started Thu May 31 14:00:15 2018 UTC and is due to finish in 60 minutes. The chair is rosmaita. 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' 14:00:30 #topic roll call 14:00:41 Somewhat present o/ 14:00:44 o/ 14:00:50 hi everyone 14:00:52 o/ 14:00:57 erno is on vacation this week 14:01:03 hi 14:01:10 how was the summit? 14:01:25 it was good, will update you in a minute 14:01:34 #link https://etherpad.openstack.org/p/glance-team-meeting-agenda 14:01:43 ok 14:02:01 there's a bunch of stuff on the agenda, but it should move pretty quickly 14:02:16 #topic updates - vancouver forum 14:02:33 the summit was good 14:02:46 my feeling was that there were more operators and less sales people 14:02:53 but that could just be who i was hanging out with 14:02:59 I felt the same. 14:03:12 It was definitely smaller, but actually felt more productive that way. 14:03:22 the forum was good, a lot of sessions like the old-fashioned design summit sessions 14:03:28 great 14:03:48 so, as far as glance goes: we received mostly positive feedback 14:03:48 And lots of operators in the room, which was great to see. 14:03:53 ++ 14:04:21 the one thing that came up (again) is quota support 14:04:27 that's good to hear 14:04:49 I might pick that up for next cycle 14:04:49 but people understand that we're waiting on the openstack-wide direction to stabilize before we implement them 14:04:54 cool 14:05:08 as far as specific topics for glance 14:05:21 no problems raised with interoperable image import 14:05:40 the image lifecycle session was well-attended, lots of discussion 14:06:03 consensus was that the hidden images spec is required, then the rest is probably just metadata 14:06:16 but the metadata needs to be somehow standardized for interoperability 14:06:48 i'll probably work on that a bit, possible good use of metadefs 14:07:16 abhishekk: you will be happy to hear that operators are very excited about your multiple backends proposal 14:07:28 awesome 14:07:37 erno outlined it, people seemed to agree that it meets their use cases 14:07:52 we pointed out the spec, i have not looked to see if anyone left comments 14:07:57 great 14:08:02 no comments yet 14:08:19 ok, we will take that as meaning the spec is great! 14:08:36 the edge computing session was also well attended 14:08:36 I am almost ready with the changes glance_store, glance and glanceclient 14:08:43 abhishekk: ++ 14:09:07 good news is it looks like the edge cases that were discussed can be met by current glance 14:09:19 you just need to be creative in configuring it and how you deploy 14:09:50 i think sean cohen at Red Hat said he will drive a proof-of-concept 14:10:04 interesting, I will have a look at the video 14:10:21 so it looks like nothing for us at this point from the glance side (though our RH people may be pulled in, I imagine( 14:10:39 i think that's all ... smcginnis anything I missed? 14:11:13 No, nothing else I can think of. 14:11:14 i'm sure erno will read through and he can add his thoughts at next week's meeting 14:11:17 cool 14:11:26 #topic updates - releases 14:11:37 ok, Rocky-2 milestone is next week 14:11:51 we will try as usual to release late on Wednesday 14:12:04 ok 14:12:09 ++ 14:12:14 spec freeze happens Thursday (i think at meeting time), so please get your final comments on them 14:12:28 and, we have a bug scrub scheduled for Friday 14:12:39 I just noticed there has not been a glance-store release yet. 14:12:44 Or is that glance_store? ;) 14:12:46 spec freeze for rocky or also for stein? (i am new btw) 14:12:54 Luzi : for rocky 14:12:56 Luzi: Welcome! 14:13:17 thank you 14:13:17 and what smcginnis said, Welcome! happy to have you here 14:13:34 ++ 14:13:54 i don't think we've had anything substantive into glance store so far, i will have to look 14:14:32 i think just one patch merged 2 days before related to fix in doc strings 14:14:46 OK, great. 14:14:48 do we need a release soon? i thought the deadline was close to Rocky-3 14:15:38 the big change will be the multihash stuff, not sure if it will be ready by wednesday 14:15:46 There's the membership freeze that says milestone projects need to participate in at least two milestone releases. 14:16:36 Ah, but this is cycle-with-intermediary, not cycle-with-milestones, so I guess we're good. 14:16:42 So nevermind me. :) 14:16:47 cool, thanks for checking! 14:17:04 i think that's the same for glanceclient also? 14:17:32 Correct. 14:17:48 though i guess that doesn't matter, i think we did a release from master a few months ago 14:18:08 2.11.0 i think 14:18:18 Yeah, client didn't come up in my list. That's good. 14:18:25 great 14:18:51 ok, that's all for updates 14:19:03 #topic glance project py35 situation 14:19:48 ok, so it turns out that my impression that our py35 functional test failures, associated with ssl, were cert problems, was incorrect 14:19:51 it's actually an eventlet problem 14:19:55 :( 14:20:07 #link https://bugs.launchpad.net/glance/+bug/1769006 14:20:08 Launchpad bug 1769006 in Glance "reconfigure functional-py35 tests" [High,In progress] - Assigned to Brian Rosmaita (brian-rosmaita) 14:20:22 Oh!? 14:20:34 so, if you look at that bug, i've outlined how we're going to handle this short term 14:20:41 (see comments #1 and #2) 14:21:06 when I say "I", i wrote it down, but it was erno's idea 14:21:22 anyway, there are some patches to please review 14:21:26 Thanks for getting this all written down. I haven't been able to keep track of what the issues are, so this really helps. 14:21:32 np 14:22:06 so, it's not all bad, there is a workaround for glance, namely don't let eventlet deal with ssl handshakes 14:22:23 sounds good 14:22:28 so one patch adds that to our docs 14:23:02 the other change is to stop skipping those functional tests ... instead, we are blacklisting them in the regular functional-py35 testenv 14:23:15 and adding a new non-voting job that will run only those tests 14:23:30 that way it will be obvious that there is a problem 14:23:36 and it will be obvious if it is fixed 14:24:10 so, associated with that patch is another patch that zane bitter put up to change how we run the serial tests 14:24:38 when i say "associated with" what i mean is that i will have to rewrite my patch if we merge zane's 14:24:49 which is perfectly ok 14:25:12 but i have some reservations (which honestly may be due to laziness) 14:25:22 #link https://review.openstack.org/#/c/571286/ 14:25:38 so please look it over and think about what makes sense for long-term maintenance 14:26:07 ok, and the final patch is an update to the governance repo explaining our situation 14:26:23 plus side is glance_store and glanceclient have satisfied the py35 goal 14:26:58 any questions about the py35 stuff? 14:27:10 In case folks haven't seen the plan, we have a couple releases to sort out any py3 related things. 14:27:54 no questions, good workaround 14:28:06 smcginnis: is the plan published anywhere yet? 14:28:21 rosmaita: In the works now: https://review.openstack.org/#/c/571011/4/resolutions/20180529-python2-deprecation-timeline.rst 14:28:32 great, ty 14:28:35 But basically, py2 support will be dropped in U. 14:28:41 Unicorn! 14:28:49 So we have a hard deadline to make sure we get things working before then. 14:28:52 :) 14:29:00 ok, that's good 14:29:28 #topic periodic "tips" testing status 14:29:40 the good news is that this is mostly working! 14:29:54 still have a year 14:30:08 what i need people to do is to look at the jobs I've defined and see if I left anything out 14:30:35 as a reminder, the idea is that we will periodically run our tests against the masters of other openstack projects we depend on 14:31:03 and i believe we could also use non-openstack projects, but i haven't tried that yet (woudl be really good to test against warlock for glanceclient) 14:31:25 on that note, has everyone heard about the gitclient vulnerability? 14:31:48 remote code execution when you recursively clone a repo 14:32:00 no, 14:32:20 #link https://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2018-11235 14:32:40 the fix is either do not recursively clone a repo, or update your git client to 2.17.1 14:32:58 ty for the link 14:33:21 anyway, the only reason i brought that up is i'm not sure the infra team is ready for us to start cloning random git repos 14:34:00 to get back on track, please take a look at the .zuul.yaml (i will list the patches in a minute) 14:34:16 i tried to group related projects so that we're not running a million different jobs 14:34:35 and i think i got most of the openstack deps out of requirements.txt 14:34:54 for glance_store: 14:35:03 #link https://review.openstack.org/571305 14:35:12 for python-glanceclient: 14:35:20 https://review.openstack.org/571310 14:35:31 for glance: 14:35:35 #link https://review.openstack.org/569225 14:35:51 i should say something about how to see the results of these. 14:36:18 each of them has an earlier patch set where i ran the jobs that we're defining to be periodic in the experimental queue 14:36:28 so you can look at those logs if you like 14:36:40 the key thing to look at is in the tox directory 14:36:52 there will be a file *siblings.txt 14:37:38 if you look at it, it should list the repos that are being cloned, and will show uninstalling the released package of those projects and building a wheel from master and using that insetad 14:38:02 oh yeah, one more thing 14:38:20 once we merge these patches, the periodic jobs will show up "in zuul" 14:38:31 what i mean is you go to zuul.openstack.org 14:38:37 take the "Builds" tab 14:38:48 and then you can filter at the top of the page 14:39:13 put "periodic" for the pipeline and "openstack/glance" for the project 14:39:34 it doesn't take wild cards, so you have to do it separately for store, client 14:39:50 anyway, you can do that now for glance and you will see some oslo jobs 14:40:07 oslo is already running their masters every night against glance master! 14:40:35 any questions? 14:40:51 got it 14:40:58 i think my advice is let's merge these, and if i missed some deps we should test, we can do it in a follow-up patch 14:41:14 ++ 14:41:16 unless you notice something obvious, we can fix that now 14:41:26 which actually brings me to the next topic 14:41:40 #topic anti-nitpicking-statement 14:41:48 this is just an awareness thing 14:42:05 there was discussion at the summit about not discouraging new contributors 14:42:25 TheJulia volunteered to put together a statement and start a ML discussion 14:42:28 so 14:42:40 I think overall the glance team isn't so bad with nitpicking, but glad it's being discussed. 14:42:40 #link http://lists.openstack.org/pipermail/openstack-dev/2018-May/130802.html 14:42:47 smcginnis ++ 14:42:59 #link https://review.openstack.org/#/c/570940/ 14:43:14 if you're interested in this, you'll probably want to read both 14:43:21 the email thread is really long 14:43:48 yeah, i like to think we're pretty good in the glance community 14:44:04 we're a bit picky about user-facing things, but otherwise i think we're pretty solid 14:44:25 meaning, raising mostly substantive points on reviews 14:44:40 I think that's appropriate to be a little more picky on user facing things. 14:45:04 i think the summary position is that it's fine to point out nits, just don't -1 something merely because of a nit 14:45:10 except maybe for user-facing things 14:45:30 i think that is important to emphasize 14:45:35 i think smcginnis is good about this 14:45:54 you often see him pointing something out and saying, "you can address this if you put up a new patch" 14:45:54 Thanks! 14:46:03 i think that is a really good practice 14:46:15 basically, if you see something incorrect, always point it out 14:46:35 whether it makes you want to block a patch or not ... is a judgement call 14:46:49 I do think it's good to give feedback, but there are a lot of things that can be improved but shouldn't block a legitimate fix or improvement. 14:46:59 All a bit subjective if course. 14:47:03 exactly 14:47:12 *of 14:47:19 nit! 14:47:22 :) 14:47:58 ok, that's all from me 14:48:03 #topic open discussion 14:48:37 any specific etherpads of glance discussion during summit? 14:49:12 yes, i think they're all listed on the wiki page, give me a minute to find the link 14:49:24 thanks 14:49:48 #link https://wiki.openstack.org/wiki/Forum/Vancouver2018 14:50:06 will go through those 14:50:07 hopefully all of ours say "glance" somewhere 14:50:45 oh yeah, just noticed, we had 2 people show up for project onboarding 14:50:59 thats good news 14:51:01 i was one of them 14:51:11 :) 14:51:13 great! 14:51:28 Luzi why don't you introduce yourself 14:51:33 Luzi, let us know if you need anything 14:53:00 abhishekk is our person on IST 14:53:08 jokke_ is on Dublin time 14:53:12 i am EST 14:53:23 well, EDT actually 14:53:29 smcginnis is CDT i think 14:53:38 well, I am CET - Berlin 14:53:47 Yep 14:53:48 so we have pretty good coverage for answering questions 14:53:58 yep 14:54:14 great, don't hesitate to ask any of us questions in #openstack-glance 14:54:15 Luzi: Great, you can show us around at the next Summit. ;) 14:55:14 I work for Secustack - we had a presentation at the summit in Vancouver for securing user data. I wasn't the presenter, but was at many onboarding sessions 14:55:43 good to know 14:55:54 glad you came to the glance meeting! 14:56:12 anything else? 14:56:32 we would like to contribute a few things, and i wanted to write a first spec for osc, but sadly there aren't many people there 14:57:01 Maybe a great place to get involved then. 14:57:56 on a selfish note, could I get some eyes on: https://review.openstack.org/#/c/514114/ ? it would be nice to get it into R-2 14:57:58 yeah, I will come back, either on ml or here to ask questions. Thank you, nevertheless :) 14:58:28 rosmaita: In a tab. 14:58:37 ty! 14:59:21 ok, i guess that's all for today ... have a productive week, everyone! 14:59:32 Thanks! 15:00:01 #endmeeting