13:59:42 #startmeeting glance 13:59:42 Meeting started Thu Jan 19 13:59:42 2023 UTC and is due to finish in 60 minutes. The chair is pdeore. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:59:42 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 13:59:42 The meeting name has been set to 'glance' 13:59:43 #topic roll call 13:59:43 #link https://etherpad.openstack.org/p/glance-team-meeting-agenda 13:59:49 o/ 14:00:51 o/ 14:01:09 o/ 14:01:12 lets wait few minutes for others to join 14:01:25 o/ 14:02:00 let's start, others may join in between 14:02:07 #topic release/periodic jobs updates 14:02:14 Milestone 3 is 4 weeks away 14:02:36 This is Spec freeze week and we have 2 important specs to get in, which we will see in next topic 14:03:00 Periodic job all green except TIME_OUT for fips jobs 14:03:18 moving to next 14:03:19 o/ 14:03:27 #topic Specs 14:03:41 #link https://review.opendev.org/c/openstack/glance-specs/+/863209 - Repropose new location APIs spec 14:03:49 #link https://review.opendev.org/c/openstack/glance-specs/+/868901 - (spec-lite) (cinder) Add suport to allow extend attached volumes 14:04:18 Both these specs looks good to me as per the discussion happened so far on this , so kindly please have a look if you have not yet reviewed the updated one 14:05:47 today is the last day for approving specs, so if I don't see any objection on the specs today, tomorrow I will approve it 14:05:57 ack 14:06:06 after tomorrow no new/old specs will be approved for this cycle 14:06:19 so kindly please have a look 14:07:05 moving to next 14:07:08 #topic Catching up with reviews 14:07:24 Update Migration Constant 14:07:24 #link https://review.opendev.org/c/openstack/glance/+/861752/ 14:07:49 abhishekk, has updated this, kindly please have a look 14:08:02 next is 14:08:05 Pin glance-tempest-plugin with 0.2.0 for wallaby job 14:08:05 #link https://review.opendev.org/c/openstack/glance-tempest-plugin/+/864005 14:08:32 this is blocking other stable/wallaby patches as secure-rbac-protection-functional job is failing for stable/wallaby 14:08:54 this one is important ^^ 14:09:33 yes, please have a look on priority 14:10:10 that's it from me for today 14:10:33 looking at the glance-t-p patch now 14:10:59 rosmaita, thanks ! 14:11:13 shall we move to open discussion ? 14:12:12 yes 14:12:23 #topic Open Discussion 14:12:34 rosmaita, i have got one +2 on service role patch, if you manage to contact keystone guys then we can get it merged and start on glance location api work 14:12:40 https://review.opendev.org/c/openstack/keystone/+/863420 14:12:58 arrrrrrrrgh .... i thought that was taken care of! 14:13:21 As we have quite a bit of time I'd like to discuss a bit of the Rajat's cinder expand proposal. Anyone up for it? 14:14:01 I think rajat will be busy in cinder midcycle stuff 14:15:45 maybe rosmaita can aswer quickly couple of my questions. 14:15:46 jokke_, i think it would be better to comment direct on the patch, I will ping whoami-rajat after meeting for the same 14:16:13 I'm here but not sure what the context is 14:16:25 lite spec 14:16:37 whoami-rajat: ok, cool. Your lite-spec about the cinder extends 14:16:49 sure 14:17:52 happy to answer any concerns 14:17:54 I'm just wondering if my assumptions are correct: 1) we have no programmatic way to know if the volume happens to support the live expansion 2) we cannot shrink the volume afterwards (or what's the reason we're stuck with 1Gb expansions)? 14:19:08 1) no, we don't, it's a backend feature which isn't reported anywhere so the info is only available in our support matrix and the operator using backend is expected to know if their backend supports this capability 14:19:20 having said that, in current master, apart from nfs i think all backends support it 14:19:29 * whoami-rajat quickly checks 14:19:42 Just wondering if 1 is correct, should we do similar what we do with rdb and ramp up the amount we expand and trim once we're done writing. 14:19:54 https://docs.openstack.org/cinder/latest/reference/support-matrix.html#operation_online_extend_support 14:20:04 there are few of them but i haven't seen those drivers used a lot recently 14:20:22 2) no we cannot, cinder only supports extend and not shrink 14:20:30 ohh 14:20:36 I had no idea 14:21:03 yeah, cinder needs to cater for all backend use cases and some backends don't support shrink so it didn't make sense for cinder to do it either 14:21:19 even though some backends support it like rbd, we can't generalize it for all backends 14:21:44 I was worried that the manual config would be error prone, but I guess it's pretty much the only way to improve things then :( 14:22:14 whoami-rajat: well, you do online extends even all backends don't support it either ;) 14:22:18 i know the 1GB expansion isn't great but that's the only way we can get accurate volume size and not overcommit storage 14:23:24 do we reserve correct size volume to start with _if_ we know the image size beforehand (Imports)? 14:24:27 the add method in cinder store does get a size parameter but I've no idea how it gets populated, I haven't seen in my testing the size being used and it always starts with 1 14:24:38 but if it's passed, cinder store should use that as initial volume size 14:25:41 https://github.com/openstack/glance_store/blob/master/glance_store/_drivers/cinder.py#L960-L962 14:26:14 ok, could you quickly add that to the spec as secondary improvement objective (so we don't need another spec for that)? As the volume operations are that heavy I think we should look into reserving correct size volume in the first place if we do have the image file on hand so we know it's size 14:26:35 I'd be happy to +2 that as whole 14:27:58 I can but is it a future effort or expected to be done in the same effort? since it seems like a different issue of size not being passed correctly 14:28:01 oh sorry,, missed that while I was writing 14:28:28 so all we need is to make sure we actually populate that parameter from glance side, it would not need changes in the store 14:28:43 then the proposal is fine as it is as sounds like it's all we really can do 14:28:53 yes, correct 14:28:58 ok i will add that to the spec 14:29:12 thanks for the feedback 14:29:31 amazing, thanks 14:29:53 whoami-rajat, Thanks for joining ! 14:30:13 whoami-rajat: and sory for distracting you from the midcycle efforts 14:30:24 pdeore, no problem, thanks! 14:30:42 jokke_, oh midcycle was yesterday, but I've another meeting to attend 14:30:44 thanks guys! 14:31:03 cheers, I have nothing else for now 14:31:08 anyone has anything else to discuss ? or we can wrap up early 14:31:12 #metoo 14:31:18 cool 14:31:26 Thanks everyone for joining ! 14:31:28 tyty 14:31:33 thank you! 14:31:40 #endmeeting