21:04:46 <timburke> #startmeeting swift 21:04:46 <opendevmeet> Meeting started Wed Apr 24 21:04:46 2024 UTC and is due to finish in 60 minutes. The chair is timburke. Information about MeetBot at http://wiki.debian.org/MeetBot. 21:04:46 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 21:04:46 <opendevmeet> The meeting name has been set to 'swift' 21:05:05 <timburke> sorry to have been gone so long! 21:05:43 <timburke> sorry, i haven't updated the agenda 21:05:49 <timburke> but first up 21:05:53 <timburke> #topic PTG 21:06:19 <timburke> thanks mattoliver for running things in my absence! 21:06:45 <timburke> though, he's not here right now, either... 21:07:14 <timburke> i'd love to hear from kota and zaitcev about how it went, though! (if you don't mind me putting you on the spot) 21:07:35 <zaitcev> I only attended 1 day because I had to be in Cinder. 21:07:55 <kota> I was in the last 2 days 21:08:27 <opendevreview> Anish Kachinthaya proposed openstack/swift master: expirer: account and container level delay_reaping https://review.opendev.org/c/openstack/swift/+/874806 21:08:53 <zaitcev> So I heard about MPU (yet another SLO but with hidden segments), open-sourcing of auto-sharding, and expirer grace period. 21:09:15 <zaitcev> Oh, wait, it's all here: https://etherpad.opendev.org/p/swift-ptg-dalmatian 21:10:04 <kota> hopefully, the meetpad (new tool) will be changed, it's not so good for me. 21:10:44 <zaitcev> I completely checked out of the discussion of the cooperative tokens. People were going on about McRouter and whatever, and I was just sitting there and thinking, "was Sage right after all? there's no exit? Paxos and the leader?" 21:12:01 <timburke> kota, sorry to hear about the tooling -- zoom was better, i take it? 21:12:23 <zaitcev> The ALO is cute. I'm not fundamentally against yet another xLO. Do you know how many memory allocators Linux kernel had? SLAB, SLOB, SLUB. 21:12:59 <timburke> zaitcev, yeah, we've *really* come to rely on having good caching -- we keep throwing new tools and new ideas at it 21:13:18 <zaitcev> I don't remember, but I think Meetpad worked for me. But I closed the Eitherpad and opened it in a separate window. 21:13:21 <timburke> and our users struggle to no end with managing their own segments :-( 21:13:25 <kota> timburke: i think so. zoom is better IMHO. 21:14:01 <zaitcev> Maybe you can write a watcher that maintains consistency of ALO segments 21:14:14 <zaitcev> Although currently nobody is thinking about it, as I gather. 21:15:10 <timburke> that's almost kinda part of acoles's idea :-) i think as a new daemon instead of using the audit-watcher hooks, though :-( 21:15:21 <timburke> all right, thanks for the highlights! 21:15:44 <opendevreview> Anish Kachinthaya proposed openstack/swift master: support x-open-expired header for expired objects https://review.opendev.org/c/openstack/swift/+/874710 21:15:46 <timburke> about one of them... 21:15:54 <timburke> #topic expirer grace period 21:16:26 <timburke> i've been looking at the patches this week, and i know clayg and acoles have been looking at them, too 21:17:02 <timburke> i think it's getting really close to merging! which is good, as we've only got Anish for a few more days 21:18:22 <timburke> so hopefully we'll see some +As later this week 21:19:45 <timburke> #topic ALO / MPU 21:19:51 <zaitcev> I bridled at it at first, but I made my peace with it when I saw that you're not adding any weird APIs that defeat the expirer and "undelete" objects. It just allows to modify the expires-at if you are prompt enough. 21:20:06 <timburke> #undo 21:20:06 <opendevmeet> Removing item from minutes: #topic ALO / MPU 21:20:42 <zaitcev> That's how I see it. Sorry for being slow. 21:20:43 <timburke> well... you can *kind of* undelete things. like you said, adjust (or remove) the expires-at 21:21:34 <zaitcev> There was some kind of an explanation why users cannot do it in time. Apparently they don't always do it, just sometimes. 21:21:55 <zaitcev> I pay my electric bill late sometimes too. 21:22:11 <timburke> there are still two points that rub me a little, too (per-container config options and no global config option), but i've also made my peace with them 21:23:18 <timburke> all right, other things making progress... 21:23:20 <timburke> #topic ALO / MPU 21:23:33 <timburke> it looks like acoles has been pretty active on the feature branch! 21:23:57 <timburke> #link https://review.opendev.org/q/project:openstack/swift+branch:feature/mpu 21:25:12 <timburke> i'm sure he'd appreciate some reviews to help spread knowledge around, though 21:26:19 <zaitcev> Well... It's a branch and I hate dealing with them. 21:26:24 <zaitcev> But okay. 21:26:48 <timburke> thanks :-) 21:27:24 <timburke> though the three patches currently open have failing tests, so... maybe just keep an eye out for when you can ;-) 21:27:53 <timburke> #topic sparse DB files 21:28:41 <timburke> looks like there's been some recent activity on an old-ish bug 21:28:46 <timburke> #link https://bugs.launchpad.net/swift/+bug/1691648 21:28:47 <patch-bot> Bug #1691648 - empty 5GB container DB (In Progress) 21:29:19 <zaitcev> I saw a vacuum patch by Matt. 21:29:36 <zaitcev> https://review.opendev.org/c/openstack/swift/+/916861 21:29:37 <patch-bot> patch 916861 - swift - db_auditor: add vacuum support - 1 patch set 21:30:11 <timburke> if memory serves, we hoped that by implementing sharding, we could avoid needing to get down to vacuuming 21:31:02 <timburke> but matt took a look at some of our large, sharded containers, and saw that the root DBs have a similar problem 21:31:52 <timburke> i forget the exact numbers, but i want to say there was something like a 500MB db file that vacuumed down to under 1MB 21:33:36 <opendevreview> Anish Kachinthaya proposed openstack/swift master: support x-open-expired header for expired objects https://review.opendev.org/c/openstack/swift/+/874710 21:33:49 <timburke> one last patch i'd to point out 21:34:03 <timburke> #topic logging/statsd client refactor 21:34:07 <timburke> #link https://review.opendev.org/c/openstack/swift/+/914029 21:34:07 <patch-bot> patch 914029 - swift - Refactor utils - 14 patch sets 21:35:16 <timburke> we realized that we probably want to push more on https://bugs.launchpad.net/swift/+bug/2015274 before trying to extend things for labeled metrics 21:35:16 <patch-bot> Bug #2015274 - utils module is too big - break it up (In Progress) 21:36:39 <timburke> not sure what the deal is with the probe test failures yet, but it's another thing to keep an eye out for 21:36:54 <timburke> i think that's about all i've got to bring up 21:36:59 <timburke> #topic open discussion 21:37:09 <timburke> anything else we should mention this week? 21:40:01 <timburke> oh! ubuntu noble should come out tomorrow! pretty sure swift should Just Work 🤞 21:41:09 <timburke> all right, i think i'll call it 21:41:22 <timburke> thank you all for coming, and thank you for working on swift! 21:41:30 <timburke> #endmeeting