14:00:13 #startmeeting glance 14:00:15 #topic roll call 14:00:25 Meeting started Thu Mar 5 14:00:13 2020 UTC and is due to finish in 60 minutes. The chair is abhishekk. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:00:26 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:00:28 The meeting name has been set to 'glance' 14:00:34 #link https://etherpad.openstack.org/p/glance-team-meeting-agenda 14:00:38 o/ 14:00:53 o/ 14:01:18 looks like only two of us, lets wait 2-3 minutes for jokke_ 14:01:40 ok 14:02:12 o/ 14:02:22 smcginnis, o/ 14:02:42 We have very small agenda for today 14:03:25 Lets start 14:03:28 o/ 14:03:34 on time 14:03:35 :D 14:03:59 #topic release/periodic jobs update 14:04:24 So we manage to release python-glanceclient on Monday 14:04:56 yeii! 14:05:24 \o/ 14:05:24 We still have 1 month until next and final milestone 14:05:39 we are pretty much on target 14:06:02 I will list down remaining priorities and present at next meeting 14:06:26 Regarding periodic job, 1 or 2 timeouts are still there 14:07:16 Cyril is communicating with the stephan who has proposed fix to subunit failure, no concrete output yet 14:07:27 Moving ahead 14:07:51 #topic S3 support for glance_store 14:08:09 This is one of the priority for U cycle 14:08:15 specs is in good shape 14:08:22 #link https://review.opendev.org/687390 14:08:30 Implementation is also available 14:08:34 * jokke_ has to admit no keeping up to date with that spec, sorry 14:08:38 #link https://review.opendev.org/695844 14:09:02 no worries, we have 2 specs, one is this and other is related to image encryption 14:09:27 I am going to test this patch functionally 14:09:54 Owner has also added steps to configure the s3 backend 14:10:08 #link https://etherpad.openstack.org/p/how-to-test-s3-driver 14:11:16 we need to make sure to include that in the proper documentation if we are going to accept this 14:11:33 jokke_, yes 14:11:57 It's great that he made the effort already at this point to help testing it 'though! 14:12:01 big kudos for that 14:12:08 rosmaita, please have a look at it 14:12:17 jokke_, yes, 14:12:36 ok 14:12:40 Moving ahead 14:12:50 #topic Open discussion 14:13:16 rosmaita, jokke_ I have commented on delete from single store patch 14:13:19 I have few things actually for this 14:13:21 please have a look at it 14:13:30 jokke_, stage is yours 14:14:13 there's few housekeeping things that could od with some love 14:14:47 I think smcginnis made great point on the sheepdog one, would like to hear what you guys thing 14:14:50 #link https://review.opendev.org/#/c/705558/ 14:15:01 looking 14:15:24 small client improvement of life from sfinucan 14:15:27 #link https://review.opendev.org/#/c/709262/ 14:15:47 agree with release note point 14:16:22 added to review list 14:16:23 stable backport for rbd driver: 14:16:26 #link https://review.opendev.org/#/c/663857/ 14:17:13 and initial steps to get rid of v1 bits in client: 14:17:17 #link https://review.opendev.org/#/c/708597/ 14:18:03 noted 14:18:19 in general we have loads of stuff that has 1 +2 already in our review queue 14:18:49 I will have a look at review queue of all 3 14:18:55 thank you for bringing this up 14:19:19 some of them have been ball I've been kicking down the road trying to find time to review, some has already +2 from myself 14:19:38 great 14:19:41 when was the sheepdog driver removed? 14:19:49 this cycle 14:20:12 ok, so it's not too late to figure out a migration strategy? 14:20:36 nope 14:20:46 That's what I raised on that patch just to be told that it's gone from glance_store already ;) 14:20:58 Yeah. And I do think it's not likely anyone is using it based on the state of the project. 14:21:11 So now really just need to figure out how to make sure we don't totally break that poor soul who is still using it somewhere 14:21:28 well, with people still running newton, who knows 14:21:36 jokke_: ;) 14:21:51 glance implemented the upgrade checker, right? 14:21:59 at least the skeleton 14:22:10 I will contact cyril and ask him to add release note as suggested 14:22:16 smcginnis: yeah, I understand that the project is unmaintained and all, but is it literally just broken and none of the releases will run anymore due to some dependency breakages etc.? 14:22:24 rosmaita: Yep - https://opendev.org/openstack/glance/src/branch/master/glance/cmd/status.py 14:22:49 ok, so i think upgrade check plus strongly worded release note should be good 14:23:02 That's what I thought as well 14:23:10 jokke_: That I'm not sure about, but based on experience, bitrot has likely set in so I would think it's broken or will be very soon. 14:23:18 just wanted to hear if we all are on the same opinion about this 14:23:46 jokke_: At least a release note from us could be a warning sign for anyone that actually is using it and not aware of the project status that they really should be looking to migrate to something else. 14:24:18 +1 14:24:29 plus, i think we deprecated it in train 14:24:30 smcginnis: ack, but that just means as we do not know for sure that it hasen't for example been operational for year or something like that someone actually might be using this so the upgrade check and strong reno is appropriate measure to do 14:25:09 rosmaita: I'm pretty sure we did not, we deprecated and threw it away based on the idea that it's just busted 14:25:29 I might be wrong 'though 14:25:38 that seems kind of nonstandard 14:25:41 Pretty sure we deprected it first. 14:25:48 I guess we deprecated it in train 14:25:51 if we did, that was very end of train and I have slept since 14:26:22 Concept of time and calendar has been bit fuzzy lately 14:27:52 https://docs.openstack.org/releasenotes/glance_store/train.html#deprecation-notes 14:27:53 Anything else guys? 14:28:15 so we did warn in Train 14:28:20 rosmaita: Thanks, I couldn't find that and started to question myself. :) 14:28:33 I think I'm good ... I added some comments on Rajat's cinder driver patch and downgraded my -2 to -1 14:28:37 me neither ... i think we need a note in both glance_store AND glance for this one 14:28:47 for sheepdog removal, i mean 14:28:48 We should have it in the glance release notes too so it's not hidden in glance_store. 14:28:56 smcginnis: ++ 14:29:08 rosmaita: thanks for correcting me on that deprecation, good 14:29:15 jokke_, looked at those comments 14:29:20 yeah, i had the same moment of panic as smcginnis 14:29:27 sounds valid 14:29:59 yep! we need to keep that in mind that glance_store is separate and heavily affects glance 14:30:10 ++ 14:30:23 well, the upgrade check will be good 14:31:07 At least the effort was not totally wasted 14:31:13 Oh one more thing! 14:32:12 shoot 14:32:28 Heated discussion about our favorite topic ever, osc vs. python-glanceclient, in the mailing list. Guess what, most people already seems to agree that the right way forward is to make osc yet another community goal like how well it worked out last time! 14:33:19 yes, saw that 14:33:44 I am not sure how they are going to achieve it unless they throw everything on respective teams 14:34:46 well, i will -2 that community goal for sure 14:35:12 smcginnis: rosmaita: I'm sure cinder users will be thrilled as well if they are pushed to use osc in it's current state 14:35:34 well, there's also the osc/sdk merger 14:35:46 yes 14:35:51 rosmaita: thanks, I'm fully planning to do the same. Thus wanted to highlight the thread. Now is the right time to start making noice how bad idea it is 14:35:53 if the osc uses sdk, then it will leave out all sorts of functionality 14:36:13 i think the problem here is that there is data-free analysis going on 14:36:52 rosmaita: wouldn't be first time when I hear something among the lines "But I don't need that feature so I don't care if it's there or not, we just need everyone using this thing X" 14:37:00 i find it hard to believe that someone can remember all 5000 'openstack xxx yyy' commands, but has trouble with 'glance image-list' and 'cinder list' 14:37:27 :D 14:37:39 rosmaita: mhm, I already pointed out that the feedback I personally have received has been very controversial vs. the few loud voices in the ML 14:38:27 yes, and the thing is, this is an empirical question 14:38:36 rosmaita: It's also hard to believe that people who would have those problems even they need to use OpenStack semi frequently would not have cheatsheets for it 14:38:57 yeah, and i think the community would be better off giving horizon some love instead of this 14:39:08 so much that!!!!! 14:39:33 i think serious scripters will either use the client classes directly, or if they use CLI, can handle different names 14:39:36 but that's just me 14:39:55 And if they can't they're probably in wrong business anyways :P 14:40:01 lol 14:40:03 :P 14:41:00 anyway, as a community goal, it needs to be ready to go first week of Victoria cycle 14:41:19 not like this time when goal #2 was finalized this week 14:41:52 and it needs to have details, not be a wish list 14:42:03 and it needs to have goal champions who are going to do the work 14:42:15 rosmaita, ++ 14:42:23 Based on our experience, I'm willing to discuss this matter when osc has demonstrated at least 2 cycles of feature parity and activity fixing the issues raised o we don't need to be telling everyone "Have you tried glance " "Oh it worked, thanks" 14:42:43 jokke_: agree entirely 14:43:08 _Then_ we can discuss endorsing osc again over our native client 14:43:25 agree 14:43:42 Last time we got the demand to change, empty promises of it being fixed and catching up over I think 2 years 14:44:13 I'm not allowing our users to be put back into that situation again as long as I'm core in this project 14:44:35 Yes I remember 14:45:03 fool me once, shame on you; fool me twice, shame on me 14:45:35 anything else? 14:45:54 fool me 3 times, won't get fooled again 14:45:54 Nope, thanks abhishekk & all! 14:46:14 rosmaita, please look at comment on delete single store patch 14:46:23 sure 14:46:28 rosmaita, smcginnis 14:46:42 anything else from you? 14:46:58 #link https://review.opendev.org/698049 14:47:44 Thank you all 14:47:51 have a nice time ahead 14:47:51 o/~ 14:48:02 #endmeeting