14:05:15 #startmeeting glance 14:05:16 Meeting started Thu Sep 12 14:05:15 2019 UTC and is due to finish in 60 minutes. The chair is rosmaita. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:05:17 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:05:20 The meeting name has been set to 'glance' 14:05:24 #topic roll call 14:05:28 o/ 14:05:48 good morninig abhishekk, may be just you and me 14:06:11 good morning rosmaita, most likely :D 14:06:27 let's skip to release updates 14:06:32 this week is feature freeze 14:06:40 ack 14:06:52 you have patches listed on the agenda 14:06:54 oops 14:06:58 i forgot to change topic 14:07:03 yeah 14:07:03 #topic release update 14:07:16 so D day has come 14:07:23 ok, abhishekk has put together a list of pritority patches 14:07:34 #link https://etherpad.openstack.org/p/glance-train-priority-patches 14:07:59 yes, so priority is to get features in before M3 release 14:08:12 we will target bug fixes in RC1 14:08:20 ok 14:08:27 are we going to do an M3? 14:08:39 or just acknowledge the deadline? 14:09:01 So first three patches are important in the etherpad 14:09:12 rosmaita, that was the plan till yesterday 14:09:26 I don't know what jokke_ has planned 14:09:44 ok 14:09:53 along with 1st 3 and 7th patch is important 14:10:03 looks like i can renew my +2 on https://review.opendev.org/#/c/611253/19 14:10:10 So how do we acknowledge the deadline 14:10:23 rosmaita, yes, it's just a rebase, no code change 14:10:34 i think we just say, we are now under feature freeze 14:10:44 also it will be good if we find smcginnis 14:10:56 Is this done by sending ML? 14:11:15 o/ 14:11:18 no, i think it's just done right now in this meeting, the FF is openstack-wide 14:11:24 hi sean 14:11:32 abhishekk has a question 14:11:38 if we don't do m3 release 14:11:48 how do we acknowledge feature freeze? 14:12:02 i was saying , we just say, "Glance is under FF" in this meeting 14:12:42 Yeah, I think it's just really amongst the team that we know it's now FF and don't approve the addition of new features unless there's been an explicit agreement to allow an exception. 14:13:03 smcginnis, cool 14:13:19 great 14:13:26 rosmaita, let's acknowledge the same :D 14:13:44 another question -- i would like to deprecate an option, do in need a FFE? 14:13:47 If it's something we can still get in today, I think we can allow that. But after the end of today then I think we need to enforce limiting things to just bugfixes and critical things for train. 14:13:55 rosmaita: I wouldn't think so. 14:14:24 smcginnis, if you get some time please have a look at first 4 patches from https://etherpad.openstack.org/p/glance-train-priority-patches 14:14:31 ok, good, i'll get a patch up before RC1 14:14:45 some of them already has at least 1 +2 14:14:55 abhishekk: Sure, I'll try to spend a little time later today to go through those. 14:15:11 smcginnis, cool, appreciate your efforts 14:15:47 so, let's acknowledge that we are in FF phase now 14:16:12 #action everyone Glance is now in feature freeze for Train 14:16:18 ack 14:16:41 o/ 14:16:47 ok, i guess that's all for releases 14:16:49 hi jokke_ 14:17:00 abhishekk: did you do a glanceclient release? 14:17:11 for train, i mean 14:17:18 rosmaita, one patch still needs to merge for glancevclient 14:17:23 hey sorry ... got distracted digging out something and didn't realize tis was back to back with my previous one 14:17:24 ok 14:17:32 #link https://review.opendev.org/658961 14:17:55 This is about backend -> store change 14:18:17 I very much thing that we should tag M3 with the key removal and caching patches 14:18:29 big things that potentially affect a lots of people 14:18:39 Is it possible for someone to put a release patch for client as I will be mostly out 14:18:55 jokke_ has a -1 on that client patch yet. Is it realistic that we can get that in today? 14:18:55 ok, so the idea is release m3 glance, and that will contain the api changes that are needed for https://review.opendev.org/#/c/658961/ 14:19:14 abhishekk: i can do it, i have the releases repo checked out 14:19:32 rosmaita, many thanks :D 14:19:37 I have no problem throwing in release patch as well 14:19:41 np 14:19:47 as long as we're in a shape to release 14:20:04 jokke_, in that case, we need to regenerate config patch, reno patch and then release patch 14:20:19 Do we think anyone would actually pick up and use an m3 glance release, or would we want that just to have a clear checkpoint of where that backend/store change happened? 14:20:59 smcginnis: basically I'd like to tag it so it gets packaged for anyone who wants to test it outside of devstack 14:21:32 I'm not expecting anyone to run anything extensive on it, just encourage testing 14:22:12 ok, well we are up against the deadline for glanceclient today 14:22:19 I am Ok with it, but in reno do we need to mention that multi-store is still experimental? 14:22:20 That's fair enough. We (or at least I) have been discouraging milestone releases unless we know someone wants to actually package and test it, but that seems fair in this case. 14:22:22 so we need to work out exactly what we are doing here 14:22:36 also do we need to bump API version? 14:23:03 I can do that with the config refresh 14:23:30 rosmaita, client patch is ready to go, erno has just -1 it as it should not get merge before the glance patch 14:23:43 ok 14:24:06 erno, why don't you put together the M3 release and ping me as needed to review patches 14:24:06 so for client patch also we need a reno patch ( it will be easy as this is the only change in client patch) 14:24:28 I can review it from mobile as well 14:24:46 Just ping me on telegram 14:24:48 abhishekk: what's needed in the glanceclient reno? 14:25:06 this backend to store change we need to mention right? 14:26:07 oh, crap, i was supposed to write that 14:26:27 guess i still can 14:26:57 great 14:26:59 rosmaita: will do 14:27:00 so will this be glanceclient 2.17.0 ? 14:27:09 yes 14:27:37 ok, so erno, you work on the glance M3 stuff, i will write a releasenote for the client release 14:28:10 rosmaita: also mention that from now on multi-store commands are considered stable and we're not fecking around with them anymore 14:28:22 Do we need API version bump? 14:28:31 jokke_: ack 14:29:30 Bonus points if you include the word "fecking" in the release notes. :) 14:29:43 smcginnis: challenge accepted 14:29:45 :D 14:31:25 so for API, CURRENT is 2.7 and 2.8 is EXPERIMENTAL ... do we just make 2.8 CURRENT? 14:31:25 jokke_, are you going to send ML for FFE of cluster awareness? 14:31:55 I guess, multi-store is still experimental 14:32:22 abhishekk: I think that boat has sailed 14:32:39 as we have couple of important patches still needs to get in which can now be done in RC1 phase 14:32:48 jokke_, really sorry for it 14:33:48 So we need to approve https://review.opendev.org/658961 patch ASAP so that we can proceed with client release 14:33:54 so about the API version, I assume we want to have both current cache and multistore current at T release 14:34:19 jokke_, could you please do the needful 14:34:36 how about we do 2.9 EXPERIMENTAL on M3, and promote it directly current if we feel comfortable doing that at RC1 14:34:48 +1 14:35:22 and for M3 we just leave 2.8 EXPERIMENTAL as well, or is there some stupid test again blocking us to do so? 14:35:53 i forget how i wrote those tests 14:36:01 so maybe 14:36:03 I don't think so, that test is only for migration 14:36:31 If you did it, I'd assume you weren't crazy enough of blocking us having multiple Experimentals :D 14:36:40 :D 14:36:49 doesn't sound like you 14:36:53 i remember that issue came up, and i do not remember how we resolved it 14:37:08 i will blame nikhil :D 14:37:11 ok, well I'll see it once I run the tests 14:37:29 cool 14:37:42 so looks like we have all sorted out at the moment 14:38:02 i think the test handles multiples 14:38:25 it checks to make sure the version negotiation code is in sync with the version response 14:38:37 that was a bug that came up at some point 14:38:51 rosmaita: mind to review 680786 (Cyril's reno patch for key deletion), I'll do the same after the meeting 14:38:57 Bug 1609571 14:38:59 bug 1609571 in Glance newton "version negotiation api middleware is NOT up to date to include v2.3 as CURRENT version" [Critical,Fix released] https://launchpad.net/bugs/1609571 - Assigned to Nikhil Komawar (nikhil-komawar) 14:39:30 jokke_: ack 14:39:51 I will try to rush home asap, so I can offer some helping hand 14:40:25 rosmaita: oh sorry, that was your PS 14:40:48 jokke_: yeah, i put up a revision last night 14:40:53 abhishekk or smcginnis: will need you to glance through that reno 14:41:01 jokke_, on it 14:41:03 https://review.opendev.org/#/c/680786/4 14:41:30 would be good for smcginnis to look, it is cinder-related 14:41:45 tyty 14:44:07 looks good to me 14:45:00 Last 15 minutes 14:45:10 I guess we are well set on course of action 14:45:17 jokke_, rosmaita: ack 14:45:57 yeah ... lots of small things and hours of gating ahead, but we should be good 14:48:00 great 14:48:34 ok 14:50:04 so quick recap just that we're on same page 14:50:44 good, please do! 14:51:08 I'll throw up config/API version patch, rosmaita does renos for client side, I'll double check if we have any renos missing for M3 and after those are merged drop release patches for both 14:51:51 API will be 2.7 CURRENT, 2.8 EXPERIMENTAL, 2.9 EXPERIMENTAL for M3 as long as our testing allows 14:52:21 oh, and we need to merge the abhishekk's client backend-> store patch 14:52:38 did I miss anything from the above? 14:53:49 We need reno for cache-prefetcher, Add 'compressed' option to container_format, Change location metadata key 'backend' to 'store', Image deletion returns 500 if 'file' store is not enabled 14:55:45 Also we need to re-enable multiple stores functional tests which we skipped while releasing glance-store 1.0 (I guess we can do it while RC1 phase when we get other multiple store changes in? 14:56:13 yeah, lets flip those on by RC1 14:56:28 ack, I have added it in todo list of mine 14:56:50 it's fine to keep skipping those while we're still claiming the feature experimental 14:57:03 which I want to change by the final T release ;) 14:57:06 cool 14:57:14 last 4 minutes 14:58:25 abhishekk: pretty sure 'compressed' container_format has a reno, not sure if it merged, though 14:58:33 it was separate from the code change patch 14:58:37 like said I go through those 14:59:00 iirc renos should be quick to gate as we hav quite minimal testing on those 14:59:04 rosmaita, that will be great 14:59:21 cool 14:59:42 rosmaita, it is there :D 14:59:45 ok, jokke_ and i will communicate in glance channel as we get stuff done 14:59:47 you have put it 14:59:54 ++ 15:00:01 thank you all 15:00:08 Thanks 15:00:11 Thanks everyone! 15:00:22 #endmeeting