21:00:11 <timburke> #startmeeting swift 21:00:11 <opendevmeet> Meeting started Wed May 28 21:00:11 2025 UTC and is due to finish in 60 minutes. The chair is timburke. Information about MeetBot at http://wiki.debian.org/MeetBot. 21:00:11 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 21:00:11 <opendevmeet> The meeting name has been set to 'swift' 21:00:13 <timburke> who's here for the swift meeting? 21:00:24 <mattoliver> o/ 21:02:27 <timburke> i don't have too many topics, mostly just follow-up from a couple weeks ago 21:02:40 <timburke> as usual, the agenda's at 21:02:41 <timburke> #link https://wiki.openstack.org/wiki/Meetings/Swift 21:02:49 <timburke> first up 21:03:04 <timburke> #topic aws-chunked and additional checksums 21:03:29 <jianjian> o/ 21:03:38 <timburke> we still need to review & merge checksum validation 21:03:47 <timburke> #link https://review.opendev.org/c/openstack/swift/+/944073 21:04:06 <timburke> but it's been going well since we carried it in prod 21:04:33 <timburke> we've even seen clients trying to use crc64nvme, so i'm glad we had that follow-on ready to go 21:04:48 <timburke> #link https://review.opendev.org/c/openstack/swift/+/946141 21:04:49 <jianjian> -88 +2061, wow, quite a large patch 21:05:03 <acoles> o/ 21:05:33 <mattoliver> oh nice, if people use it let's look at getting that landed too! 21:05:37 <timburke> decent majority of it should be in test coverage, though, so that's good 21:05:50 <acoles> are we still waiting for a py version bump to land the checksum validation? something to do with up-reving boto3 for tests?? 21:06:13 <timburke> nope -- i decided to just split the test expectations based on version of boto3 available 21:06:20 <acoles> ok 21:06:35 <acoles> I'll try to revisit that review then 21:07:10 <timburke> fwiw, i think that was only needed in test_different_checksum_requested 21:08:29 <timburke> there's also a follow-up for s3token to ensure that aws-chunked works even when the secret key didn't come out of cache 21:08:33 <timburke> #link https://review.opendev.org/c/openstack/swift/+/949671 21:09:21 <timburke> ...which would be nice to have before our next upstream release (not that i'm in a rush for that, but good to mention early) 21:10:34 <timburke> next up 21:10:41 <timburke> #topic swiftclient release 21:10:52 <timburke> we had one! 4.8.0 is out and available 21:12:01 <timburke> it also dropped support for py36, so we're (slowly) getting closer to openstack's support matrix ;-) 21:12:26 <mattoliver> \o/ 21:12:59 <jianjian> great! 21:13:06 <timburke> sorry we didn't get more landed (i know there was some hope of getting retries for bytes literals merged) 21:13:08 <timburke> #link https://review.opendev.org/c/openstack/python-swiftclient/+/918701 21:13:22 <timburke> but there's always more releases we can make! 21:13:35 <jianjian> 🙂 will wait for next release then 21:14:10 <timburke> next up 21:14:23 <timburke> #topic conditional PUTs for s3api 21:14:31 <timburke> #link https://review.opendev.org/c/openstack/swift/+/949680 21:14:42 <timburke> it merged! thanks for your help, jianjian! 21:15:02 <jianjian> \o/ 21:15:23 <timburke> we've got it out in prod, and our preliminary testing shows it working as we'd hoped 21:16:23 <mattoliver> nice work timburke and jianjian 21:16:46 <timburke> (in particular, we had a user interested in lancedb using swift's s3api as a backend, and jianjian's manual testing showed it working -- still waiting to here back from our users) 21:16:51 <timburke> #link https://github.com/lancedb/ 21:17:22 <jianjian> yep, also the s3api support of if-none-match: * works well with aws cli and boto3 SDK as well 21:18:12 <timburke> all right, last topic from me 21:18:15 <timburke> #topic MANIFEST.in removal 21:18:49 <timburke> i merged a change a bit ago because it seemed like this really wasn't necessary anymore 21:18:57 <timburke> #link https://review.opendev.org/c/openstack/swift/+/949368 21:19:20 <timburke> i built wheels before and after the patch, and everything seemed right 21:20:03 <timburke> but the next time we went to build RPMs for our internal release, they were missing the xml schema files used by s3api! 21:20:51 <mattoliver> yeah, it seems this is still the way to tell setuptools to include extra files 21:20:52 <timburke> as a result, s3api requests that included xml bodies would all fail (multi-deleted, mpu-completions, probably others) 21:21:05 <timburke> i proposed a revert for it 21:21:09 <timburke> #link https://review.opendev.org/c/openstack/swift/+/950615 21:22:16 <timburke> but i still haven't pinned down where things went wrong -- my assumption is that our build process's old versions of pbr and setuptools were to blame 21:23:01 <mattoliver> I've managed to fix it downstream in our rpm build but specificalyy including them. But that might not help others. So unless there is another way to specify extra files in setuptools we might need to put that back 21:23:23 <timburke> and now that we can actually declare minimum versions of those in pyproject.toml, i think i might prefer to just make sure those minimums are right and keep the MANIFEST.in out 21:23:26 <opendevreview> Yan Xiao proposed openstack/swift master: Object-server: change labeled timing metrics sample rate for debugging requests https://review.opendev.org/c/openstack/swift/+/951042 21:24:41 <timburke> mattoliver, did you explicitly add the files in our spec file, or did you try updating pbr/setuptools? i think clayg tried the latter approach, but ran into other issues because of our old python(/OS?) :-/ 21:24:53 <mattoliver> oh good thinking. Maybe need to play with pbr and setuptools versions then an find the culprit if its just a version then. 21:25:30 <mattoliver> I just added the files, I mean the spec already explicitly grabs the man files etc 21:27:06 <timburke> yeah, i figured i ought to start with trying to build with our lower-bounds, as they're pretty reasonable: https://github.com/openstack/swift/blob/master/pyproject.toml#L2 21:27:26 <timburke> released in '22 and '23 21:27:56 <clayg> yeah, our .spec for setuptools ~41 expected some bootstrap.py script that doesn't exist ~64 21:28:25 <timburke> so i'll plan to try that out in the next day or two and post results in a comment on the revert 21:28:37 <mattoliver> kk sounds like a plan 21:29:18 <mattoliver> clayg: oh great. So setuptools spec needs a bunch of updating then. 21:29:36 <timburke> but i wanted to make sure to bring it up in case cschwede_ or anyone else bumps into a similar issue. if it's more than just us, i'm a lot more inclined to merge the revert 21:29:41 <mattoliver> maybe it's already done in the 3.10 branch 21:30:02 <timburke> i think it might be 21:30:10 <timburke> all right, that's all i've got 21:30:15 <timburke> #topic open discussion 21:30:24 <timburke> anything else we ought to talk about this week? 21:30:51 <acoles> just an fyi, I''m ramping back up on feature/mpu work 21:31:05 <mattoliver> nice 21:31:56 <acoles> picking up where I left off, and pursuing a new approach to reliably detecting obsolete MPUs for orphan cleanup 21:32:39 <mattoliver> I've been a bit distracted with some downstream work, but I looked at the dataclass ringv2 patch tho. Nice we can start using more modern python concepts (even if its from 3.7) :P 21:33:26 <mattoliver> I want to finally push up a rebase of the tracing chain.. otherwise it'll get harder and harder as time goes on. Hopefullly get back to this sooner rather then later. 21:34:03 <mattoliver> mostly got it ready, just need to test it in vsaio to make sure everything is working 21:36:37 <timburke> oh yeah, i should take a look at https://github.com/NVIDIA/vagrant-swift-all-in-one/pull/134 ... 21:38:10 <mattoliver> woah 2022... really should land this one day, maybe this year will be there year :P 21:38:40 <mattoliver> that vsaio probably needs a rebase too 21:38:49 <mattoliver> anyway, that's all I've got 21:39:11 <mattoliver> oh I did remember the meeting last week too, which was nice :P 21:39:44 <timburke> 🎉 21:40:05 <timburke> thanks again for leading that meeting slot 21:41:46 <timburke> all right, i think i'll call it early then 21:41:57 <timburke> thank you all for coming, and thank you for working on swift! 21:42:02 <timburke> #endmeeting