21:04:03 <timburke> #startmeeting swift
21:04:03 <opendevmeet> Meeting started Wed Oct  4 21:04:03 2023 UTC and is due to finish in 60 minutes.  The chair is timburke. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:04:03 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
21:04:03 <opendevmeet> The meeting name has been set to 'swift'
21:04:54 <timburke> i've just got a handful of follow-ups to last week's meeting
21:05:02 <timburke> #topic vPTG
21:05:16 <kota> o/
21:05:38 <timburke> we still need to list discussion topics at
21:05:40 <timburke> #link https://etherpad.opendev.org/p/swift-ptg-caracal
21:06:07 <uhyeongjo> o/
21:06:27 <mattoliver> Oh yeah, I was on vacation for the week and half. I'll do my best to seed that with some topics
21:06:35 <timburke> and i still need to book meeting times -- if you haven't already (thanks kota and acoles!), please indicate preferred time slots at
21:06:38 <timburke> #link https://framadate.org/liKzqQw5lZQW8Pvf
21:06:44 <mattoliver> *last week and a half
21:06:55 <mattoliver> Kk
21:07:44 <acoles> o/
21:07:51 <timburke> PTG's just over two weeks away -- it'll be good to have some video time with each other :-)
21:08:33 <timburke> #topic py312
21:09:34 <timburke> greenlet has shipped a 3.12-ready release now, at least -- eventlet is still likely to take a bit to get support
21:10:14 <timburke> but zaitcev reviewed the whole chain to let us drop our use of distutils!
21:10:35 <mattoliver> Oh cool
21:10:51 <mattoliver> I hadnt seen that chain. (Still catching up)
21:11:05 <timburke> #link https://review.opendev.org/c/openstack/swift/+/896473
21:11:11 <timburke> is the end of it
21:11:17 <mattoliver> K
21:12:19 <timburke> i might start looking at getting py312 gate jobs up for swiftclient and pyeclib -- my expectation is those should Just Work
21:12:43 <timburke> last bit from last week:
21:12:47 <timburke> #topic stable gates
21:13:10 <timburke> xena and wallaby are now fixed, and the backport to wallaby that kicked this all off has landed
21:14:09 <mattoliver> Thanks for always digging in an fixing the gates timburke
21:14:15 <timburke> victoria was passing when i checked: https://review.opendev.org/c/openstack/swift/+/896858
21:14:28 <timburke> but usurri was not: https://review.opendev.org/c/openstack/swift/+/896859
21:15:12 <timburke> still haven't tried train, but i'd guess it's probably broken
21:15:37 <timburke> (otoh, i would have said the same for victoria... i was rather surprised it passed)
21:16:09 <timburke> stein and earlier have already been EOL'ed
21:18:28 <timburke> looking at active patches...
21:18:38 <timburke> #topic s3api partNumber API
21:18:59 <timburke> there's been a good bit of activity on the chain ending with
21:19:04 <timburke> #link https://review.opendev.org/c/openstack/swift/+/894580
21:20:01 <timburke> clayg has some nice refactoring in slo and its tests
21:21:09 <timburke> then indianwhocodes builds from there, first adding a part-number query param to SLO, and letting s3api do the appropriate translations for S3 clients
21:21:16 <acoles> +1 there's some great cleanup in the refactor
21:21:50 <mattoliver> cool
21:22:09 <timburke> while on the topic of s3 compat...
21:22:28 <timburke> #topic object lock "support"
21:22:38 <mattoliver> yeah this :)
21:23:01 <mattoliver> So it turns out latest version of ansible that using the aws s3 module doesn't work with swift s3api
21:23:01 <timburke> mattoliver pushed up a patch recently to at least let us respond to some lock-related requests
21:23:12 <timburke> #link https://review.opendev.org/c/openstack/swift/+/897108
21:23:30 <mattoliver> and it's because they create a bucket and then attempt to pull down the object-lock-configuration
21:24:01 <mattoliver> which we don't support.. so instead of returning the "not configured" error we treat it as a normal bucket request
21:24:20 <timburke> fwiw, the patch is very similar to our initial "support" for various versioning-related APIs -- i'm all for it
21:24:37 <mattoliver> so basically it was returning '{}' (empty dict) in boto
21:25:05 <mattoliver> this causes ansble to die with a keyerror. Because they're code is brittle.
21:25:34 <mattoliver> the patch, as timburke said basically just returns the expected not configured 404 that s3 returns
21:26:28 <mattoliver> I need to retest the latest version with ansible (will do that today). But boto should now see the specific s3 error and hopefully now work
21:27:01 <mattoliver> If it doesn't the ansible is simply broken because now we're behaving like s3 :P
21:27:56 <timburke> sounds good -- i expect we'll probably get that landed this week
21:28:08 <mattoliver> nice, thanks :)
21:28:27 <mattoliver> I'm not the most familar with s3api side of swift, so been a good refresher :)
21:28:37 <acoles> nice one mattoliver
21:29:27 <mattoliver> thanks for your patience and support timburke
21:29:37 <timburke> 👍
21:29:47 <timburke> all right, i think that's all i've got
21:29:52 <timburke> #topic open discussion
21:29:59 <timburke> anything else we should talk about this week?
21:30:54 <clarkb> yes! did you see the PR I posted a few days ago?
21:31:12 <clarkb> I don't know enough about erasure coding but want to make sure someone sees it since there is potential for data loss maybe?
21:31:22 <clarkb> https://github.com/openstack/liberasurecode/pull/5
21:31:48 <timburke> yes! thanks clarkb! yeah, that looks like a solid catch, and sadly, not the first time that someone's tried to submit a PR about it...
21:32:12 <clarkb> side note, you can subscribe to the github projects to get notifications by email for stuff like this. This is how I end up finding them
21:32:13 <timburke> i've been trying to dust off my C to write some tests to go with it
21:32:36 <clarkb> cool, i just wanted to make sure that didn't fall through the cracks
21:32:48 <mattoliver> thanks clarkb and great idea, I'll make sure I'm subscribed
21:33:14 <mattoliver> Like I mentioned earlier, I was on vacation for the last 1.5 weeks. So I dont' have much to say, still haven't had a chance to catch up. Will try and get back into reviews soon. I also want to catch up with the proxy side of the GET namespaces API.. But nothing to discuss here until I do (maybe next week)
21:33:37 <timburke> 👍 subscribed
21:36:52 <timburke> all right, i think i'll call it then
21:37:02 <timburke> thank you all for coming, and thank you for working on swift!
21:37:06 <timburke> #endmeeting