| opendevreview | Abhishek Kekane proposed openstack/glance master: Modify download API reference https://review.opendev.org/c/openstack/glance/+/969353 | 05:23 |
|---|---|---|
| opendevreview | Pranali Deore proposed openstack/glance master: Switch tests to native threading https://review.opendev.org/c/openstack/glance/+/975118 | 09:45 |
| opendevreview | Merged openstack/glance master: Fix flaky test_import_copy_bust_lock functional test https://review.opendev.org/c/openstack/glance/+/993670 | 10:26 |
| *** rosmaita1 is now known as rosmaita | 12:01 | |
| opendevreview | Markus Hentsch proposed openstack/glance master: Standardization of encrypted images https://review.opendev.org/c/openstack/glance/+/926295 | 13:59 |
| croelandt | #startmeeting glance | 14:02 |
| opendevmeet | Meeting started Thu Jun 18 14:02:03 2026 UTC and is due to finish in 60 minutes. The chair is croelandt. Information about MeetBot at http://wiki.debian.org/MeetBot. | 14:02 |
| opendevmeet | Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. | 14:02 |
| opendevmeet | The meeting name has been set to 'glance' | 14:02 |
| croelandt | #link https://etherpad.openstack.org/p/glance-team-meeting-agenda | 14:02 |
| croelandt | #topic roll call | 14:02 |
| sakumbha | o/ | 14:02 |
| mklejn | o/ | 14:02 |
| croelandt | I think Abhishek is out today | 14:02 |
| mhen | o/ | 14:02 |
| pdeore_ | o/ | 14:03 |
| abhishekk | o/ | 14:03 |
| croelandt | maybe rosmaita and dansmith are here and somewhere else at the same time | 14:03 |
| croelandt | abhishekk: hope you're feeling better! | 14:03 |
| dansmith | sorry, here | 14:03 |
| croelandt | #topic Periodic jobs | 14:04 |
| rosmaita | o/ | 14:04 |
| croelandt | everything green \o/ | 14:04 |
| croelandt | #topic glance-gaps | 14:04 |
| abhishekk | croelandt: stomach issue but I can manage few minutes here | 14:04 |
| croelandt | We have just merged two patches in the SDK and have some SDK/OSC patches | 14:04 |
| croelandt | so good work everyone, ideally we can get a few more merged before M2 | 14:04 |
| rosmaita | abhishekk: hope you feel better soon | 14:04 |
| croelandt | after M2 I think the SDK/OSC cores will be a bit reluctant to merge patches | 14:04 |
| abhishekk | thanks | 14:05 |
| croelandt | #topic eventlet removal | 14:05 |
| croelandt | The flaky tests seem to have been tamed thanks to abhishekk | 14:05 |
| croelandt | so we can resume reviewing and merging :) | 14:05 |
| abhishekk | fingers crossed | 14:05 |
| croelandt | #topic Open Discussion | 14:05 |
| croelandt | I see mklejn and mhen here so I think we might have topics | 14:05 |
| mhen | o7 | 14:06 |
| mhen | hi | 14:06 |
| Luzi | o/ | 14:06 |
| mhen | just a quick update regarding image encryption | 14:06 |
| mhen | there was some silence recently due to people being absent (vacation etc.), incl. me | 14:06 |
| mhen | I have now picked up again | 14:06 |
| mhen | and rebased most of the patchsets to current master | 14:07 |
| mhen | and tested on a freshly deployed devstack | 14:07 |
| croelandt | do you have a link to that patch series? I think dansmith is eager to read all these patches | 14:07 |
| mhen | they are part of https://review.opendev.org/q/topic:%22LUKS-image-encryption%22 | 14:07 |
| dansmith | not eager, but I don't want to be the weak link :) | 14:07 |
| mklejn | crude data/control plane split etherpad with draft ideas, that needed to be heavily distilled from internal drafts, because multiple ideas/assumptions were incompatible with upstream :) https://etherpad.opendev.org/p/A5ENoJ_0lbZnoBN7HuIC | 14:07 |
| mklejn | that's about it for my topic today | 14:08 |
| croelandt | mhen: so many different projects | 14:08 |
| mhen | there was one thing raised in the cinder patchset | 14:09 |
| mhen | https://review.opendev.org/c/openstack/cinder/+/926298/comment/469d680a_7982c539/ | 14:09 |
| mhen | about some image resize function we need to look into | 14:09 |
| mhen | might be a blind spot in our current implementation | 14:10 |
| mhen | cannot say yet | 14:10 |
| croelandt | mklejn: this looks huge, this will required a PTG talk, and coordination with teams in charge of k8s I would say | 14:10 |
| mhen | aside from that we have some longer-running open threads in https://review.opendev.org/c/openstack/glance/+/926295?tab=comments still open involving mostly dansmith and croelandt | 14:11 |
| dansmith | mhen: I mean, it's pretty much what you'd expect, right? images are rarely the same size as the container you're going to put them in | 14:11 |
| mklejn | croelandt: I agree that this is huge. that's exactly why we decided to first approach community before making downstream changes that would be hard to maintain afterwards | 14:11 |
| dansmith | mhen: https://wiki.archlinux.org/title/Resizing_LVM-on-LUKS#Resize_LUKS_volume | 14:11 |
| dansmith | basically, lay it down on the volume then extend it to fill as if you expanded an LVM | 14:12 |
| mhen | dansmith: thanks for the pointers; I haven't had the chance to actually look at the code yet and what its purpose it, that is still a todo for me | 14:13 |
| dansmith | ack | 14:13 |
| croelandt | mklejn: have you implemented anything downstream? | 14:14 |
| mklejn | croelandt: given possible size of this, and effective architecture change, we didn't touch the topic yet | 14:15 |
| croelandt | I wonder if we've had the same kind of issue | 14:16 |
| croelandt | is there an "easy" reproducer to showcaase the performance issue? | 14:16 |
| mklejn | not really, you'd need to fill your DB with >1M image records for starters | 14:17 |
| croelandt | wow | 14:17 |
| dansmith | I'm interested in this topic, but I think it's probably the kind of thing we need to discuss at a PTG, high-bandwidth | 14:17 |
| croelandt | yeah | 14:18 |
| dansmith | I'd really love to make this something glance can do on its own without an external LB | 14:18 |
| croelandt | but I don't want us to get to PTG and be like "yeah we gota one hour session" | 14:18 |
| dansmith | so option 3 (plus 4 maybe) would be my preference, | 14:18 |
| croelandt | especially since we'd need people who handle real installations, imho | 14:18 |
| dansmith | but it requires a lot more from glance itself of course | 14:18 |
| croelandt | ok I guess we'll need to think about it carefully | 14:20 |
| croelandt | Anything else for today? | 14:20 |
| mklejn | dansmith: we also really like option 3 + 4, even though 4 presents multiple challenges | 14:20 |
| dansmith | yeah | 14:20 |
| dansmith | also, I assume you know that import can redirect to the worker that has the actual staged data right? | 14:21 |
| mklejn | sorry, I'm also in high bw meeting concurrently so I had a lag | 14:21 |
| dansmith | that makes it easier to support in all these cases | 14:21 |
| mklejn | "import can redirect to the worker that has the actual staged data" I was actually not aware :) | 14:22 |
| dansmith | yeah, you can basically have one worker that gets the /import say "oh I don't have that staged data, redirect to https://woker123.domain.com/import" | 14:22 |
| dansmith | so you don't have to have shared staging | 14:23 |
| opendevreview | Pranali Deore proposed openstack/glance master: Switch tests to native threading https://review.opendev.org/c/openstack/glance/+/975118 | 14:23 |
| mklejn | croelandt: I'm proven more and more that this really needs to be taken carefully | 14:24 |
| croelandt | mklejn: yeah I would not expect an implementation to pop up in a few months :D | 14:24 |
| croelandt | Also I think that if there is a real performance issue, we'll want to fix it, but you might have to do the heavy lifting | 14:25 |
| dansmith | follow the thread of CONF.worker_self_reference_url .. the docs site is practically unusable for me at the moment else I'd link you | 14:25 |
| dansmith | so there's already precedent for some of what #3 describes | 14:25 |
| mklejn | for point 4 it at least partially will be implemented, because we need it far before "2 PTGs in the future in best case" | 14:26 |
| mklejn | but 3 is extremely promising, especially given the precedent | 14:26 |
| * dansmith nods | 14:27 | |
| mklejn | I have nothing more to add at this point, so please don't let me detain you. I'll come back from time to time as we refine this on our side, but we'd much rather work out a solution, than Just Write A Thing | 14:29 |
| croelandt | so if you end up writing something we can always review and make sure it's upstream so you don't have to maintain downstream patches | 14:29 |
| mklejn | croelandt: ++, thanks. | 14:29 |
| croelandt | ok, thanks everyone for joining | 14:29 |
| croelandt | see you next week | 14:29 |
| croelandt | #endmeeting | 14:29 |
| opendevmeet | Meeting ended Thu Jun 18 14:29:58 2026 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4) | 14:29 |
| opendevmeet | Minutes: https://meetings.opendev.org/meetings/glance/2026/glance.2026-06-18-14.02.html | 14:29 |
| opendevmeet | Minutes (text): https://meetings.opendev.org/meetings/glance/2026/glance.2026-06-18-14.02.txt | 14:29 |
| opendevmeet | Log: https://meetings.opendev.org/meetings/glance/2026/glance.2026-06-18-14.02.log.html | 14:29 |
| opendevreview | Abhishek Kekane proposed openstack/glance master: Use Jammy nodeset for glance-grenade-centralized-cache job https://review.opendev.org/c/openstack/glance/+/993942 | 14:44 |
| opendevreview | Abhishek Kekane proposed openstack/glance master: Fix import tasks stuck in processing after quota failure https://review.opendev.org/c/openstack/glance/+/993959 | 15:13 |
| opendevreview | Claudiu Belu proposed openstack/glance master: tests: enforce autospec=True via oslotest mock_fixture https://review.opendev.org/c/openstack/glance/+/993981 | 17:54 |
Generated by irclog2html.py 4.1.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!