13:59:22 <pdeore> #startmeeting glance 13:59:22 <opendevmeet> Meeting started Thu Apr 27 13:59:22 2023 UTC and is due to finish in 60 minutes. The chair is pdeore. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:59:22 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 13:59:22 <opendevmeet> The meeting name has been set to 'glance' 13:59:23 <pdeore> #topic roll call 13:59:23 <pdeore> #link https://etherpad.openstack.org/p/glance-team-meeting-agenda 13:59:28 <pdeore> o/ 13:59:32 <mrjoshi_> o/ 14:00:37 <pdeore> lets wait few more minutes for others to join 14:01:11 <abhishekk> o/ 14:01:30 <croelandt> o/ 14:01:35 <pdeore> Let's start 14:01:42 <pdeore> #topic release/periodic jobs updates 14:01:49 <pdeore> M1 is 2 weeks away from now 14:02:01 <pdeore> Periodic job all green except failure/time out on functional tests due to version conflicts 14:02:08 <pdeore> #link https://review.opendev.org/c/openstack/glance/+/881036 14:02:40 <pdeore> croelandt, did you get any chance to test this failure by trying Brian's comment? 14:02:54 <croelandt> no I haven't had time to look at that change again 14:03:04 <pdeore> ohh ok 14:03:20 <croelandt> there were some overwhelming news lately + some other work :) 14:03:25 <croelandt> I'll try to take a look at it tomorrow 14:03:27 <pdeore> so we need to try it by changing the nodeset to jammy for py310 jobs 14:03:52 <pdeore> ack :) 14:04:13 <pdeore> ok let's move ahead 14:04:19 <pdeore> #topic Specs 14:04:38 <pdeore> We have some important specs up for review, so kindly please have a look 14:04:46 <pdeore> #link https://review.opendev.org/c/openstack/glance-specs/+/880627 - Reproposed Image Encryption Spec 14:04:53 <pdeore> #link https://review.opendev.org/c/openstack/glance-specs/+/880030 - Reproposed New Location API spec 14:05:00 <pdeore> #link https://review.opendev.org/c/openstack/glance-specs/+/881451 - Download Image from Specific Store 14:05:13 * croelandt has to do that too 14:05:41 <pdeore> this 3rd one is submitted by abhishekk for new API to download image from specific store instead depending on location strategy configured by deployer 14:05:51 <pdeore> #link https://review.opendev.org/c/openstack/glance-specs/+/880637 - Bobcat project priorities 14:06:25 <pdeore> i think the last one is easy to review :) 14:07:12 <abhishekk> 3rd one is low priority though 14:07:48 <pdeore> yeah 14:08:00 <pdeore> moving to next 14:08:05 <pdeore> #topic Important Bugs 14:08:16 <pdeore> croelandt, ^ 14:08:42 <abhishekk> just FYI, I have submitted PoC for that spec as well 14:08:51 <pdeore> everyone please have a look at https://review.opendev.org/c/openstack/python-glanceclient/+/880765 14:09:02 <abhishekk> https://review.opendev.org/c/openstack/glance/+/881590 14:09:16 <pdeore> abhishekk, ack Thanks 14:09:24 <abhishekk> np! 14:09:25 <croelandt> yeah so https://review.opendev.org/c/openstack/python-glanceclient/+/880765 is actually needed to unblock the gate 14:09:32 <croelandt> if we wait, then we will have another issue at some point 14:09:36 <croelandt> and that's gonna be a pain to fix 14:09:48 <croelandt> I could also use some help with https://bugs.launchpad.net/python-glanceclient/+bug/1674886 14:10:00 <croelandt> The comment from Prerna seems incorrect to me 14:10:35 <croelandt> It's true that using "glance image-upload" with an invalid image id takes a bit of time, but I don't think it's worth checking the image id beforehand, thus adding one API call every time 14:10:53 <croelandt> and I don't think we actually chunk the image before checking the image id 14:11:04 <croelandt> if anyone is familiar with this code path I could use a confirmation :) 14:11:13 <croelandt> and I think we should deny the feature request 14:11:15 <croelandt> any thoughts? 14:11:21 <abhishekk> will have a look 14:11:40 * dansmith stumbles in 14:11:41 <abhishekk> it's mitaka bug :o 14:11:43 <croelandt> yeah yeah 14:11:49 <croelandt> but sometimes old bugs are still relevant :) 14:12:09 <croelandt> the traceback shows Python 2.7, I was young at the time 14:13:55 <abhishekk> I think this is some environmental issue, because we do call get image before now 14:14:15 <croelandt> yeah 14:14:20 <croelandt> and it fails immediately 14:14:33 <croelandt> from my tests it seems to take a bit longer with a 400MB Debian image than with a fake empty image 14:14:47 <croelandt> I wonder ifthat's because of the file pointer we pass to requests.put() 14:14:53 <croelandt> maybe requests does something with it 14:14:53 <abhishekk> https://github.com/openstack/glance/blob/master/glance/api/v2/image_data.py#L132 14:15:12 <croelandt> but anyway, it does not feel worth checking the image ID every time with a requests from the client 14:15:38 <croelandt> abhishekk: yeah that's what I had in the meeting etherpad as well, and I checked we raised an exception here :) 14:17:09 <abhishekk> hmm, so client side is doing some processing I guess 14:17:46 <croelandt> yeah but getting a file pointer is not costly 14:18:06 <croelandt> I'd have to debug more but I think it's really request.put() which does something with the file pointer 14:18:11 <croelandt> probably reads the whole file or something 14:18:31 <croelandt> now the question is: do we want to add a check for *all* calls just to fail faster when a wrong image id is provided? 14:19:21 <abhishekk> I think ideally that should be case 14:19:27 <croelandt> really? 14:19:35 <croelandt> That means we had a delay for *all* operations 14:19:45 <croelandt> we add* 14:21:14 <abhishekk> that is my opinion :d 14:21:27 <croelandt> I mean it's easy to do 14:21:31 <croelandt> I'd have to time that call 14:21:53 <abhishekk> So this only reproducible with --progress, right? 14:22:03 <croelandt> no, without progress as well 14:22:06 <croelandt> I thought so too 14:22:12 <abhishekk> strange 14:22:18 <abhishekk> then needs to dig into it 14:22:21 <croelandt> yeah maybe I should remove the call to put or something 14:23:07 <abhishekk> get_data_file this function ? 14:23:15 <croelandt> yeah it just gets a pointer 14:24:02 <abhishekk> it has a file open call 14:24:05 <croelandt> y 14:24:08 <croelandt> but that's fast :p 14:24:15 <abhishekk> :D 14:24:40 <croelandt> ok I'll do more diggin and we'll take about it next week :) 14:24:48 <abhishekk> I think at this moment we should dig more 14:24:53 <croelandt> ok ok 14:25:02 <pdeore> +1 14:25:11 <abhishekk> let me know if you need any help here 14:25:18 <croelandt> ok thanks! 14:26:02 <pdeore> let's move to open discussions :) 14:26:12 <pdeore> #topic Open Discussions 14:26:31 <rajiv_> hi 14:26:39 <rosmaita> croelandt: did you see my comments on your patch that adds py310 testing? 14:26:45 <abhishekk> We need to get update about fips, rbd deffered deletion and image encryption 14:26:47 <rajiv_> i read the SLURP info, but want to confirm again. Should we upgrade to Antelope or not ? or is there any upcoming fixes ? 14:26:51 <croelandt> rosmaita: yes 14:26:55 <croelandt> haven't had time to look into this 14:26:55 <rosmaita> :) 14:28:35 <abhishekk> I think there might be some fixes coming in, not sure 14:31:44 <pdeore> abhishekk, ack I will try to take the updates 14:32:04 <abhishekk> ack, thanks 14:32:09 <pdeore> anyone has anything else to disucss 14:32:59 <abhishekk> when will be 1st review party? 14:33:22 <pdeore> ohh ohh i completely forgot that :X 14:33:30 <pdeore> it's next week 14:33:51 <pdeore> Thanks abhishekk for the reminder :) 14:33:57 <abhishekk> np! 14:34:27 <pdeore> so next week we will have the video call meeting as usual for review party ! 14:34:45 <abhishekk> yey 14:35:04 <pdeore> Monday is holiday , right 14:35:11 <abhishekk> for s 14:35:13 <abhishekk> for us 14:35:15 <croelandt> Join the protests :) 14:35:30 <abhishekk> I am going to start one :D 14:35:38 <pdeore> :D 14:36:35 <abhishekk> I am done 14:36:59 <pdeore> ok, let's wrap up then ! 14:37:10 <pdeore> Thanks everyone for joining ! 14:37:29 <pdeore> #endmeeting