14:00:07 #startmeeting glance 14:00:08 Meeting started Thu Feb 20 14:00:07 2020 UTC and is due to finish in 60 minutes. The chair is abhishekk|away. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:00:09 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:00:10 #topic roll call 14:00:12 The meeting name has been set to 'glance' 14:00:18 #link https://etherpad.openstack.org/p/glance-team-meeting-agenda 14:00:21 o/ 14:00:23 o/ 14:00:26 o/ 14:00:32 o/ 14:00:33 welcome back! 14:00:38 :D 14:00:55 Lets start 14:01:10 #topic release/periodic jobs update 14:01:31 is something wrong? 14:01:48 Meeting bot dead? 14:01:52 #chair abhishekk 14:01:52 i was wondering how smart the meetbot was 14:01:52 Warning: Nick not in channel: abhishekk 14:01:53 Current chairs: abhishekk abhishekk|away 14:02:06 #topic release/periodic jobs update 14:02:14 \o/ 14:02:15 smcginnis, nope 14:02:38 Glance milestone 2 released on time 14:02:48 thank you for all of your efforts \o/ 14:02:49 good job! 14:03:00 Now we are waiting for python-glanceclient release 14:03:22 we still need to merge multiple imports and copy-image support patches 14:03:44 i haven't looked at those at all 14:03:56 will we able to release this week? 14:04:33 jokke_, as you are busy, do you mind if I work and submit the patch for copy-image? 14:04:55 abhishekk: go for it ... today is still quite hectic for me 14:05:08 jokke_, ack 14:05:29 smcginnis, is it ok if we release glance client by coming monday? 14:05:31 it should be simple patch to add as the multi-store import should have majority of the stuff in place already, just the verifications for the new method needed 14:05:56 jokke_, should I fix those review comments as well? 14:06:13 is this going to be the first py3-only glanceclient release? 14:06:25 yes 14:06:55 i wonder whether it would be worth just cutting it so we have a baseline, and then release again when we get the functionality merged 14:07:04 i don't mean wait for the final release 14:07:12 just a second intermediate release 14:07:21 unless the code really is that simple 14:07:53 i may not be making any sense 14:08:00 it's been a long morning already 14:08:28 I am ok for both the way here 14:08:47 1 release now and then make another release with new functionality 14:08:58 2 release with new functionality 14:09:10 i think that would be good 14:09:25 that way the py3-only version can start getting a workout in the gate 14:09:37 and we don't have to rush reviews of something that might break it 14:09:42 right 14:09:52 abhishekk: I'll try to get them done, but if you don't see them popping up or don't hear from me, feel free to pick it up tomorrow morn 14:10:03 I'll try to find some time this afternoon to fix those 14:10:04 jokke_, ack 14:10:23 abhishekk: but do _not_ spend your night worrying about that ;) 14:10:58 jokke_, do you agree to what we discussed just now? 14:11:09 jokke_, no :D 14:11:34 well, IIUC the client is ran py3 only anyways regadless if we release it or not so I don't think it makes any difference 14:12:00 do we have anything merged since last release apart from those test job changes? 14:12:06 yes 14:12:17 we have removed py2 tempest-full job 14:12:46 exactly, but that does not affect the packaged clients what so ever 14:13:17 I think we should release the client now and don't rush for getting important functionality patches in 14:13:18 the client used in gates are executed under py3 anyways as we don't deplyo py27 anymore and we do have the py3 classifiers there 14:13:32 In this case we will get proper time to test those as well 14:13:37 so just tagging the release as point of time kind of has no value when there is no changes 14:13:46 yeah, but i think the release team is pushing for a release 14:14:00 smcginnis: why is that^^ 14:14:12 rosmaita, jokke_ yes 14:14:19 Libs need to be in two milestones to be included in the cycle. 14:14:30 https://releases.openstack.org/ussuri/schedule.html#u-mf 14:15:12 abhishekk: And sorry, I got pulled away a bit. Yes, Monday as good for the client. 14:15:31 Like version numbers are cheap but it feels very silly to have two different versions of the client with zero delta between them 14:16:29 Definitely shouldn't do a release if nothing has changed. 14:16:37 so lets see the status of patches till monday, if they are in good shape merge it otherwise I will go ahead and give nod to release the client 14:16:54 sounds good 14:17:09 Periodic jobs all green 14:17:22 oslo.config 7.0 issue is resolved 14:17:35 nice 14:17:43 ++ 14:17:53 moving ahead 14:17:55 #topic Multiple import fails if "all_stores" specified as "true" 14:18:10 while testing jokke_ client patch, I have found this issue 14:18:21 reported it on launchpad and provided a fix 14:18:30 #link https://review.opendev.org/#/c/708629/ 14:18:38 rosmaita, jokke_ kindly have a look 14:18:56 will do ... that's the one you flagged last night, right? 14:18:59 i like problems that are already fixed! 14:19:05 jokke_, yes 14:19:14 rosmaita, :D 14:19:15 that was good catch and fortunately simple to fix 14:19:23 yes 14:19:34 just something we need to keep in mind when working on this new multi-store world 14:20:10 I have added a note in docs, just check the wording and place is appropriate or not 14:20:22 moving ahead, 14:20:24 abhishekk: amazing! 14:20:26 thanks 14:20:39 #topic glance-store - Image upload fails if cinder multipath is enabled 14:21:08 this sounds like fun 14:21:16 this is a fix to priority issue, I have reported LP bug for the same 14:21:22 jokke_, ha ha 14:21:33 #link https://review.opendev.org/#/c/708811/ 14:21:39 kindly review 14:21:52 so is this actual glance issue or os_brick issue? 14:22:09 rosmaita, your inputs are valuable here 14:22:27 jokke_, it was glance issue as it is hard-coded in glance store 14:22:35 somehow I'm wondering why we should ever even see if the volume is multipath or not 14:23:24 not sure, but all third party like nova or whoever using cinder are doing same 14:23:33 i think the cinder glance_store driver made a bunch of assumptions when it was implemented 14:23:53 also we need to backport it to previous branches 14:24:03 rosmaita, ++ 14:24:30 rosmaita: ah 14:24:37 I will spend some time in refactoring the cinder driver (but it will take some time, mostly V1 milestone) 14:24:58 rosmaita: well I still remember the fact that it did not work at all in real life while we thought it was all good and supported :P 14:25:16 abhishekk: i think whoami-rajat from the cinder team is interested in working with you on this 14:25:26 rosmaita, yes 14:25:40 jokke_: i think it was only experimental while i was ptl :P 14:25:46 we are sitting in same office 14:25:48 yep i am 14:26:06 great, that will work nicely then 14:26:06 i'm currently looking at the gaps that the latest cinder and glance cinder store have 14:26:16 same office and same time zone! 14:26:39 correct 14:26:41 that sounds like a win win for both teams, and most of all for our users 14:26:44 Great to see that get some focus. Thanks guys! 14:27:09 Kindly review this as early as possible 14:27:20 moving ahead 14:27:29 #Open discussion 14:27:35 #topic Open discussion 14:27:58 I definitely need some sleep :P 14:28:10 i put up some spec-lites about the checksum stuff we discussed last week 14:28:13 #link https://review.opendev.org/#/c/708761/3 14:28:26 rosmaita, saw your specs, need to go through the latest version 14:28:32 no rush 14:28:39 thank you for putting it together 14:28:50 i went through the glanceclient code to make sure it was working the way we thought 14:28:58 apparently, we all have pretty good memories! 14:29:16 So just a bit of feedback. I'm sitting in a room with about dozen of our (RH) field folks. Apparently Glance is pretty decent work with as it's straight forward and just works. So thanks all for working to get us to that point and helping to keep things that way :D 14:29:30 \o/ 14:30:02 \o/ 14:30:03 have got good feedback through the week 14:30:18 That's great to hear. 14:30:22 that is definitely nice to hear! 14:30:28 jinx 14:30:30 multiple people telling that they've never had any issues with glance 14:30:34 Finally sounds like all our efforts has been paid off 14:30:38 mhm 14:31:21 just wanted to share it with ye guys as this kind of feedback does not come granted 14:31:36 I guess most of struggle comes to end once we have new import workflow and new import plugins 14:31:40 Hi 14:31:47 Just a little question for you guy's 14:31:53 alistarle: \o 14:31:54 alistarle, shoot 14:32:18 We check with gregoire that in June took place the PTG at Vancouver, is it PTG only or a classic Summit ? 14:32:43 it will be both i guess 14:32:48 Because we see that the Open infra summit are in Berlin later this year 14:33:06 Oh ok 14:33:18 Not a full summit. 14:33:27 Because we don't know if it worth it to get there if it is only the PTG 14:33:35 There will be minimal presentations, so the focus is more on the PTG type activities. 14:33:37 #link https://www.openstack.org/events/opendev-ptg-2020/ 14:34:00 it is something like, morning sessions are for summit and afternoon for PTG 14:34:02 that indicates that it's still combined event of both 14:34:33 Ok, and do you think it is worth it to get there, or must we wait the other summit in Berlin ? 14:34:38 alistarle, it will be good to see you and gregoire there 14:35:26 If you read under the "What to Expect" section, it sounds like the OpenDev portion is similar to the Forum we've had at Summits. 14:35:38 alistarle: it definitely would not hurt us having you there, but I think it's more of matter how you prioritize your time and travel and can you get other value out of it than spending some quality time with us ;) 14:35:41 It would be a great opportunity to get feedback and discussion between the team and others. 14:35:57 alistarle, what just jokke_ said 14:36:47 Ok great ! We will try to get there so 14:36:55 cool 14:37:29 yeah, i think from a developer point of view, we should be able to get a lot done 14:37:32 We would like to work on the quota stuff, and as it is related to keystone, maybe it is better to work on that with them if needed 14:37:54 great 14:38:26 alistarle: i think nova is working on that too 14:38:59 you may be able to pool efforts 14:39:09 Yeah I check that too, but it is different for nova, as they already have a quota system 14:39:17 We have our one patch related to registry config generation is reverted 14:39:44 so they begin to migrate to the unified limit of keystone, it is different here, because we have nothing, so the keystone system must be complete before working on something 14:40:14 or at least we need to synchronize with them about their target ;) 14:40:15 #link https://review.opendev.org/#/c/707759/ 14:40:21 alistarle: ok 14:40:50 alistarle: the good news o that i that for a change we do not need to break everything in the place just to move common way of doing things 14:40:55 oops, looks like we forgot about devstack 14:41:11 I have submitted one patch in devstack but it was put on hold now :P 14:41:26 #link https://review.opendev.org/#/c/708062/ 14:41:37 mhm, after year+ of deprecation they somehow manage still to depend on that stuff 14:42:00 i have a patch to remove GLANCE_V!_ENABLED option that's been sitting for 10 weeks 14:42:01 I don't know how devstack developers work and who is going to verify it 14:42:02 abhishekk: maybe we should remerge the patch and give them initiative to move that patch forward 14:42:33 lol 14:42:36 i think it may be the QA team that's responsible for devstack these days? 14:42:44 I would have thought the tempest job on there would be the validation needed. 14:42:50 Yep, QA. 14:42:52 well that won't make the life any easier at least 14:43:01 I have submitted one patch to enable multiple stores in devstack is sitting there for almost 2 months now 14:43:26 my "Add config to allow Glance not to run under uwsgi" has been up for 14 months 14:43:40 lol 14:43:47 there have been fires in the last two months 14:44:02 my suggestion is to raise the issues during the weekly QA office hours on Tuesday 14:44:22 tosky: good idea 14:44:35 Maybe post something to the mailing list making sure everyone is aware of this happening. 14:44:36 If this continues to happen then we will not able to get rid of any deprecated stuff 14:44:39 smcginnis: you're right, that's weird that the change made it through 14:45:07 I wonder if Jens is just being extra cautious. 14:45:43 but he is the only active core I see these days 14:45:53 for devstack** 14:46:12 abhishekk: you taking about this ? https://review.opendev.org/#/c/689104/ 14:46:40 gmann, yes 14:46:51 this is been reviewed recently on feb 17 14:47:11 gmann, yes 14:47:20 it is more of getting conclusion than lack of review 14:47:42 may be if you can discuss it on qa channel and we can move faster 14:47:52 gmann, ack 14:48:24 I have discussed about this patch duirng PTG as well 14:49:11 yeah.may be adding a job to show the result can be better. but i have checked the review comment yet so let me review it today 14:49:33 cool, thank you 14:50:03 abhishekk: there is also a new devstack core, and also a few others are still around 14:50:04 gmann, also kindly let me know which of the comments I have missed (see the last comment) 14:50:12 sure. 14:50:24 I am not that good with the bash syntaxes :( 14:50:32 tosky, ack 14:51:16 Do we have anything else to discuss? 14:51:21 jokke_, rosmaita smcginnis 14:52:04 I don't think I had anything else 14:52:13 Nothing from me. 14:52:54 cool 14:53:06 Thank you all for your time 14:53:11 thank you gmann tosky 14:53:11 Thanks! 14:53:19 thanks all 14:53:30 #endmeeting