21:02:55 <timburke> #startmeeting swift 21:02:55 <opendevmeet> Meeting started Wed May 14 21:02:55 2025 UTC and is due to finish in 60 minutes. The chair is timburke. Information about MeetBot at http://wiki.debian.org/MeetBot. 21:02:55 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 21:02:55 <opendevmeet> The meeting name has been set to 'swift' 21:03:01 <mattoliver> Wife is sick and I'm on kids, might not be very available this morning 😔 21:03:02 <timburke> who's here for the swift meeting? 21:03:12 <mattoliver> Kinda 21:03:37 <timburke> mattoliver, oh no! hope she feels better soon! my wife and daughter are sick today, too 21:04:26 <timburke> fortunately, she's at an age where having her home sick isn't so much of a disruption -- my son would need a lot more things 21:04:45 <timburke> as usual, the agenda's at 21:04:48 <timburke> #link https://wiki.openstack.org/wiki/Meetings/Swift 21:05:02 <timburke> first up 21:05:09 <timburke> #topic aws-chunked 21:05:18 <timburke> stuff's been merging! 21:05:22 <timburke> thanks mattoliver! 21:05:37 <timburke> so we have support for the new protocol 21:05:44 <timburke> #link https://review.opendev.org/c/openstack/swift/+/836755 21:05:59 <timburke> and crc32c available 21:06:05 <timburke> #link https://review.opendev.org/c/openstack/swift/+/909800 21:06:20 <timburke> next up is to actually start verifying checksums during upload 21:06:25 <timburke> #link https://review.opendev.org/c/openstack/swift/+/944073 21:07:46 <timburke> it needs a bit of work at the moment, though -- i tried bumping up our minimum-required boto3 (to make sure that we get a particular behavior in one of the cross-compat tests), but the new minimum requires py38+ :-/ 21:08:05 <timburke> i'm going to try to address that this afternoon 21:09:03 <timburke> i also got a patch up at the end of the chain to address one of the known-gaps -- s3token support in particular 21:09:18 <timburke> #link https://review.opendev.org/c/openstack/swift/+/949671 21:10:04 <timburke> i don't think there's much to discuss about the chain, just wanted to give a status update 21:10:08 <mattoliver> Yeah it seemed it still needed some work 21:10:21 <timburke> #topic dropping py36 support 21:10:34 <timburke> we're doing it! 21:10:53 <timburke> swift and pyeclib patches have already merged 21:11:01 <timburke> #link https://review.opendev.org/c/openstack/swift/+/949684 21:11:05 <mattoliver> I went a landing spree 21:11:06 <timburke> #link https://review.opendev.org/c/openstack/pyeclib/+/938098 21:11:14 <timburke> yeah you did :-) 21:11:32 <timburke> and i wrote a swiftclient patch when i did the swift one 21:11:38 <timburke> #link https://review.opendev.org/c/openstack/python-swiftclient/+/949685 21:12:10 <mattoliver> Now we just need to remove 3.7 for the chunked thing ;) 21:13:06 <mattoliver> Kk will look at the swiftclient one today 21:13:15 <timburke> i'm hopeful that we can drop py37 (and probably py38) soon-ish -- i know mattoliver and alexander have been trying to get us (NVIDIA) off py37 and onto py310 21:13:47 <timburke> but even just dropping py36 should unblock some other work, like some packaging issues other folks have seen 21:13:54 <timburke> #link https://review.opendev.org/c/openstack/swift/+/899958/ 21:14:29 <timburke> speaking of swiftclient... 21:14:36 <timburke> #topic swiftclient release 21:14:39 <jianjian> recently I was blocked by one failed py36 test case, that's awesome we are going to get it removed! 21:15:34 <timburke> we're getting close to the Flamingo-1 milestone, so the release team is looking for a client library release from us. i'll work on getting an authors/changelog patch up this week 21:16:11 <mattoliver> Kk, let's double check the open client patches as well then 21:16:20 <timburke> i think the only thing i'd really like to see land before we cut a release is that drop-py36 patch -- if anyone has others, let me know 21:18:00 <timburke> oh, maybe there are a few other things we oughta dust off: https://review.opendev.org/q/p:openstack/python-swiftclient+is:open+-age:52w+branch:master 21:18:23 <timburke> i'll try to take a look at the formpost patch 21:18:25 <timburke> #link https://review.opendev.org/c/openstack/python-swiftclient/+/833954 21:18:38 <timburke> i feel like it was close 21:19:06 <jianjian> I will look at this patch again: https://review.opendev.org/c/openstack/python-swiftclient/+/918701 21:20:10 <timburke> the default-encoding thing seems... unfortunate :-( 21:20:11 <jianjian> it broke my testing of very large number of uploads a few times, if there is no pending review issue, let's get it merged. 21:21:27 <mattoliver> Partnum support seems pretty useful too.. so I'll take a look at that. (And any follow ups on the form post one, thanks Tim, I haven't looked at that in ages too) 21:22:30 <jianjian> I mean that issue broke my testing, I did not see the issue happening with the fix "Fix retries for bytes, BytesIO" 21:22:30 <timburke> i mean, what i'd *really* like with the partnum work would be parallel downloads -- but i don't think we could realistically get that together this week 21:22:58 <mattoliver> Yeah kk 21:23:14 <timburke> but yeah, retrying with bytes and BytesIO seems like it ought to be a no-brainer 21:23:37 <timburke> all right, i've just got one more topic this week 21:23:48 <timburke> #topic s3api and conditional writes 21:24:54 <timburke> we had a user report of 501s the other day, and i *think* it comes down to AWS having added support for if-match and if-none-match like a year and a half ago or so 21:26:12 <timburke> i didn't get direct evidence of it, but it sure seems like what's going on is the library they're using wants to send `if-none-match: *` during PUTs to prevent overwrites 21:26:40 <timburke> but swift already supports that! so i put together a patch to have s3api allow the request through 21:26:50 <timburke> #link https://review.opendev.org/c/openstack/swift/+/949680 21:28:30 <timburke> it'd be nice if we could eventually support the full range of conditional write headers, but i worry that we'll want to get full-consistency before trying, or users will be disappointed and confused 21:29:32 <timburke> all right, that's all i've got 21:29:36 <timburke> #topic open discussion 21:29:43 <timburke> anything else we should talk about this week? 21:31:07 <mattoliver> Great I'll take a look (if-none-match) patch 21:31:13 <timburke> thanks 21:31:31 <mattoliver> Nope, I mean I think the so_tomestamping patch is coming along. 21:32:04 <indianwhocodes> o/ 21:32:04 <mattoliver> I've been seeing an intermittent test failure, not sure it's related, will check on master today. 21:32:41 <jianjian> yeah, "if-none-match: *" would be useful to unblock some users to keep their app or database working with swift 21:32:48 <mattoliver> timburke: how do you think ring 2 is looking? 21:32:52 <timburke> yeah! i'm confused by those spotty failures, too, mattoliver 21:33:20 <jianjian> will finish review on that as well 21:34:08 <timburke> afaik, ring v2 is pretty good. i went and took out some unneeded interfaces i'd added to the RingData structure; i should probably poke clayg about re-reviewing 21:35:33 <mattoliver> Kk 21:38:42 <jianjian> clay was also happy with the swiftclient fix byte retries patch, I just sent it to be merged: https://review.opendev.org/c/openstack/python-swiftclient/+/918701 21:39:12 <timburke> gate's not happy tho :-( 21:39:40 <timburke> all right, i'm gonna call it -- sounds like my kid does need some help 21:39:51 <timburke> thank you all for coming, and thank you for working on swift! 21:39:54 <timburke> #endmeeting