21:04:00 <timburke> #startmeeting swift 21:04:00 <opendevmeet> Meeting started Wed May 29 21:04:00 2024 UTC and is due to finish in 60 minutes. The chair is timburke. Information about MeetBot at http://wiki.debian.org/MeetBot. 21:04:00 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 21:04:00 <opendevmeet> The meeting name has been set to 'swift' 21:04:06 <timburke> who's here for the swift meeting? 21:04:15 <kota> o/ 21:04:38 <mattoliver> o/ 21:05:15 <timburke> i neglected to update the agenda again 21:05:29 <timburke> but we've got a few bodies of work moving forward 21:05:53 <timburke> #topic cgi removal 21:06:32 <timburke> i realized just today that we actually have something already in-tree that's close to what we want for a cgi.parse_header replacement 21:06:55 <timburke> so i updated p 887908 to use utils.parse_content_type 21:06:55 <patch-bot> https://review.opendev.org/c/openstack/swift/+/887908 - swift - Stop using cgi.parse_header - 4 patch sets 21:07:33 <mattoliver> oh, well that's handy :) 21:08:45 <timburke> there was one spot where it didn't work -- if the parameter had a slash in it, it'd lose everything between the slash and the next semicolon 21:10:12 <timburke> i considered just allowing slashes all the time, but given the attempts to maintain RFC compliance, i figured i ought to add a `strict` arg, and only allow slashes when not strict 21:10:42 <mattoliver> Do we need to think about renaming the function. parse_content_type doesn't sound quite right in the new places 21:12:31 <timburke> fair question -- i don't have strong feelings either way 21:13:33 <timburke> we can discuss more on the patch, i suppose 21:13:49 <timburke> next up 21:13:54 <mattoliver> yeah, happy to move it there. I'll have a closer look and play with it first anyway :) 21:14:17 <timburke> #topic expirer 21:14:50 <timburke> clayg's been pushing in a few different directions 21:14:59 <timburke> #link https://review.opendev.org/c/openstack/swift/+/919960 21:15:00 <patch-bot> patch 919960 - swift - object-expirer: refactor creating InternalClient - 3 patch sets 21:15:19 <jianjian> I have looked into a few of them. :-) 21:15:28 <timburke> #link https://review.opendev.org/c/openstack/swift/+/919638 21:15:29 <patch-bot> patch 919638 - swift - tests: realistic task_container names - 4 patch sets 21:15:49 <timburke> #link https://review.opendev.org/c/openstack/swift/+/912496 21:15:49 <patch-bot> patch 912496 - swift - add bytes of expiring objects to queue entry - 19 patch sets 21:16:12 <timburke> #link https://review.opendev.org/c/openstack/swift/+/920452 21:16:12 <patch-bot> patch 920452 - swift - configurable object_expirer_task_container_per_day - 3 patch sets 21:16:47 <mattoliver> woah, just a few different directions :P 21:17:45 <jianjian> "add bytes of expiring objects to queue entry" this one is running on prod already, maybe we should get it merged? 21:17:47 <timburke> i'm still torn about how deeply to look at p 914713 when there's some hope that p 918366 will make it largely unnecessary 21:17:47 <patch-bot> https://review.opendev.org/c/openstack/swift/+/914713 - swift - DNM expirer: new options to control task iteration - 18 patch sets 21:17:49 <patch-bot> https://review.opendev.org/c/openstack/swift/+/918366 - swift - Parallel distirbuted task container iteration - 3 patch sets 21:18:37 <timburke> jianjian, agreed -- it's on my list to take another look 21:19:15 <timburke> the check job failures looked like some transient mirror issues 21:21:46 <timburke> meanwhile... 21:22:04 <timburke> #topic skipping inline updates 21:22:25 <jianjian> I took a look at "Parallel distirbuted task container iteration" yesterday, left a few comments, I feel we could have those different strategies for users to choose from, like "legacy", "random" or "parallel". 21:22:26 <timburke> mattoliver is still pushing on p 919667 21:22:26 <patch-bot> https://review.opendev.org/c/openstack/swift/+/919667 - swift - obj: skip to async_pending when root is busy - 7 patch sets 21:22:34 <opendevreview> Yan Xiao proposed openstack/swift master: tests: Check proxy-logging stats on the wire https://review.opendev.org/c/openstack/swift/+/896967 21:22:35 <opendevreview> Yan Xiao proposed openstack/swift master: stats: API for native labeled metrics https://review.opendev.org/c/openstack/swift/+/909882 21:23:39 <mattoliver> yup. refactored some annoyingly similar tests. Also used it as an example of how to gerrit for the qatar students (whom should be mostly in channel but sleeping now) :) 21:23:52 <timburke> nice! 21:24:14 <jianjian> "obj: skip to async_pending when root is busy" nice work! I am going to take a look at it too. 21:25:17 <mattoliver> thanks jianjian 21:25:25 <timburke> elsewhere 21:25:32 <timburke> #topic stats/logger separation 21:26:15 <timburke> shreeya's pushing on p 919444 and p 915483 21:26:15 <patch-bot> https://review.opendev.org/c/openstack/swift/+/919444 - swift - Move StatsdClient config parsing from get_logger t... - 4 patch sets 21:26:15 <patch-bot> https://review.opendev.org/c/openstack/swift/+/915483 - swift - HybridMultiObjectDispatcher - 12 patch sets 21:27:12 <timburke> whichlooks like the second one's still got some test failures, though 21:27:48 <timburke> but again, we're interested in trying to get it sorted out ahead of 21:27:57 <timburke> #topic labeled metrics 21:28:36 <mattoliver> nice segway! 21:28:51 <timburke> which yan's been keeping chained on top in p 896967 and p 909882 21:28:52 <patch-bot> https://review.opendev.org/c/openstack/swift/+/896967 - swift - tests: Check proxy-logging stats on the wire - 10 patch sets 21:28:53 <patch-bot> https://review.opendev.org/c/openstack/swift/+/909882 - swift - stats: API for native labeled metrics - 15 patch sets 21:29:43 <timburke> next up 21:29:50 <timburke> #topic DB vacuuming 21:30:11 <timburke> #link https://review.opendev.org/c/openstack/swift/+/916861 21:30:11 <patch-bot> patch 916861 - swift - db_auditor: add vacuum support - 5 patch sets 21:31:08 <timburke> i saw zaitcev took a look recently -- thanks! i think there's some good points there, though we might still try it out in our prod cluster and see how things go 21:31:36 <mattoliver> yup, I need to loop around and respond to those. 21:32:10 <mattoliver> I personally like having 2 options/triggers at once. But still more testing is required on it. 21:33:03 <timburke> and last up 21:33:25 <mattoliver> We might (once we feel satifided) just run it with vacuum turned off to get the metrics it generates out to see "how" bad bloated dbs are. 21:33:25 <timburke> #topic aws-chunked support 21:34:45 <timburke> i wrote a custom s3 client that lets me send all the invalid headers i want! i plan on using it to investigate the various failure modes of the aws-chunked protocol 21:35:16 <timburke> and it's already surfaced some differences in how we handle x-amz-content-sha256 mismatches 21:35:22 <timburke> #link https://review.opendev.org/c/openstack/swift/+/920206 21:35:22 <patch-bot> patch 920206 - swift - s3api: Clean up some sha256 validation/mismatch er... - 3 patch sets 21:36:13 <mattoliver> oh nice 21:36:24 <timburke> and i've started to use it with the first additional-checksums patch 21:36:29 <timburke> #link https://review.opendev.org/c/openstack/swift/+/909801 21:36:29 <patch-bot> patch 909801 - swift - s3api: Add support for additional checksums - 15 patch sets 21:37:49 <timburke> next up, i need to refresh my memory about some of the notes i left myself in the MPU patch's commit message 21:37:54 <timburke> #link https://review.opendev.org/c/openstack/swift/+/909802 21:37:54 <patch-bot> patch 909802 - swift - s3api: Additional checksums for MPUs - 14 patch sets 21:38:48 <timburke> then stack up the actual aws-chunked patch and investigate those failure modes 21:38:51 <timburke> #link https://review.opendev.org/c/openstack/swift/+/836755 21:38:52 <patch-bot> patch 836755 - swift - Add support of Sigv4-streaming - 21 patch sets 21:39:16 <timburke> i think that's all i've got 21:39:21 <timburke> #topic open discussion 21:39:29 <timburke> anything else we ought to talk about? 21:39:45 <timburke> there's a *lot* of stuff everyone's working on! 21:41:07 <mattoliver> yeah! great work everyone! I plan to get round and review a bunch more this week. Been a bit of a distracted with other work week (And family all got gastro which didn't help). 21:41:25 <mattoliver> But means I don't have much to update this time round 21:44:16 <timburke> all right, i think i'll call it early then 21:44:28 <timburke> thank you all for coming, and thank you for working on swift! 21:44:31 <timburke> #endmeeting