21:00:02 <timburke> #startmeeting swift 21:00:02 <opendevmeet> Meeting started Wed Sep 25 21:00:02 2024 UTC and is due to finish in 60 minutes. The chair is timburke. Information about MeetBot at http://wiki.debian.org/MeetBot. 21:00:02 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 21:00:02 <opendevmeet> The meeting name has been set to 'swift' 21:00:09 <timburke> who's here for the swift meeting? 21:00:13 <fulecorafa> o/ 21:01:04 <mattoliver> o/ 21:01:53 <timburke> as usual, the agenda's at 21:01:55 <timburke> #link https://wiki.openstack.org/wiki/Meetings/Swift 21:01:59 <timburke> first up 21:02:05 <timburke> #topic PTG 21:02:11 <timburke> it's only a month away! 21:02:21 <acoles> o/ sorry I'm late 21:02:32 <timburke> we *are* signed up now, at least :-) 21:02:50 <timburke> registration is appreciated 21:02:53 <timburke> #link https://ptg2024.openinfra.dev/ 21:03:00 <mattoliver> oh yeah, need to add topics to the etherpad. I'll try and think of things you haven't added today. 21:03:06 <timburke> and we need to choose some meeting slot times 21:03:15 <timburke> #link https://framadate.org/LQOsGVVWXDhXqQUw 21:03:46 <timburke> i'll aim to start booking slots around this time next week 21:04:31 <timburke> and following up on what mattoliver said, please add topics to the etherpad as you think of them 21:04:38 <timburke> #link https://etherpad.opendev.org/p/swift-ptg-epoxy 21:04:55 <timburke> and i'll aim to flesh out the ones i already added some more 21:06:18 <mattoliver> kk 21:06:25 <timburke> i think that's most everything for the PTG for now 21:06:35 <timburke> #topic labeled metrics 21:07:10 <timburke> we've been running with this in prod for a bit now, and i hear our SRE have even started making some per-tenant dashboards based on it 21:07:28 <timburke> currently two main patches worth looking at 21:07:34 <timburke> #link https://review.opendev.org/c/openstack/swift/+/909882 21:07:35 <patch-bot> patch 909882 - swift - stats: API for native labeled metrics - 36 patch sets 21:07:42 <timburke> #link https://review.opendev.org/c/openstack/swift/+/917711 21:07:42 <patch-bot> patch 917711 - swift - Add labeled metrics to proxy-logging - 19 patch sets 21:08:19 <timburke> we were talking this morning and i think there's a growing consensus that it's about ready to merge! 21:08:36 <timburke> so we probably ought to find some time to review it :P 21:08:48 <jianjian> saw the new labeled metrics on Grafana panels yesterday, looked great! 21:10:12 <jianjian> SRE also said there were some issues with Grafana, probably not related to the patches, sorry forgot what it is 21:10:36 <mattoliver> kk, then I'll try and get around to review it. although I'll be off on vacation next week.. so have to see where I get to. 21:10:57 <timburke> i don't know that i have much more to say about it, just wanted to call it out since we haven't mentioned it for a bit and ask for some review bandwidth :-) 21:11:29 <timburke> #topic pyeclib wheels 21:11:49 <timburke> mattoliver's been taking a look at the start of the chain! thanks, mattoliver! 21:11:58 <timburke> #link https://review.opendev.org/c/openstack/pyeclib/+/817498 21:11:59 <patch-bot> patch 817498 - pyeclib - Add Dockerfile to build manylinux wheels - 18 patch sets 21:12:01 <jianjian> i will try to take a look of those two patches 21:12:15 <timburke> thanks, jianjian! 21:13:17 <timburke> i rechecked the rest of the patches in the chain; they'd failed with some transient repo error last week 21:14:32 <timburke> once we have those (and maybe mattoliver's p 930171) i think i'd like to get a fresh pyeclib release out 21:14:32 <patch-bot> https://review.opendev.org/c/openstack/pyeclib/+/930171 - pyeclib - Call liberasurecode_instance_destroy in destructor - 3 patch sets 21:16:31 <timburke> oh, and i made a pyeclib/liberasurecode dashboard for myself, but maybe other people would find it useful, too 21:16:35 <timburke> #link https://review.opendev.org/dashboard/?title=Erasure+Code+Patches&liberasurecode%20(Active)=project:openstack/liberasurecode+is:open+-age:12w&pyeclib%20(Active)=project:openstack/pyeclib+is:open+-age:12w&liberasurecode%20(Landed)=project:openstack/liberasurecode+is:merged+-age:2w&pyeclib%20(Landed)=project:openstack/pyeclib+is:merged+-age:2w 21:17:00 <mattoliver> yeah, I'll add it to my gerrit thanks tim! Nice 21:17:39 <mattoliver> yeah, I'm on an pyeclib wave atm, so want to finish the wheels chain. 21:17:44 <jianjian> nice dashboard .... wow, quite a lot of patches 21:17:46 <acoles> I like a dashboard :) 21:17:48 <timburke> :D 21:18:07 * diablo_rojo sneaks in the back of the room 21:18:30 <timburke> fwiw, i modeled it after another of my favorites -- i forget if i shared it here before 21:18:33 <timburke> #link https://review.opendev.org/dashboard/?title=The Fortnight That Was&foreach=-age:2week&Landed (Server)=is:merged project:openstack/swift&Landed (Client)=is:merged project:openstack/python-swiftclient&Active (Server)=is:open project:openstack/swift&Active (Client)=is:open project:openstack/python-swiftclient&Abandoned (Server)=is:abandoned project:openstack/swift&Abandoned (Client)=is:abandoned project:openstack/python-sw 21:18:33 <timburke> iftclient 21:18:42 <timburke> bah! too long! 21:19:04 <timburke> also it botched the spaces :-/ 21:19:40 <timburke> oh well 21:19:43 <diablo_rojo> timburke: chuck it in a paste or an etherpad? 21:20:05 <timburke> thanks :-) doesn't matter too much right now 21:20:23 <timburke> next up 21:20:30 <timburke> #topic s3api bugs 21:20:56 <timburke> sorry, fulecorafa, i think these are still languishing :-( 21:21:09 <fulecorafa> All good hehe :) 21:21:46 <timburke> again, though, i really want to get to them -- at the very least, dig into the traceback 21:21:51 <acoles> I'm waiting for "The fortnight that will be" dashboard ;-) 21:21:55 <timburke> #link https://bugs.launchpad.net/swift/+bug/2081103 21:21:56 <patch-bot> Bug #2081103 - s3api: Deleting the current version of an object can (sometimes?) 500 (New) 21:22:37 <timburke> that's about all i've got 21:22:42 <fulecorafa> timburke if you want any help exploring this bug in the future, feel free to call me 21:22:45 <timburke> #topic open discussion 21:22:57 <diablo_rojo> I've got a topic :) 21:22:59 <timburke> fulecorafa, thanks! 21:23:10 <timburke> diablo_rojo, go ahead! 21:23:28 <fulecorafa> I mean this because I tried replicating the bug with some tens of thousands of buckets in my machine and could not. Possible needed a bigger throughput 21:23:54 <diablo_rojo> A university in North Dakota is looking for projects for students to do for the next semester (starting in January and going till May). Since you mentored for Carnegie Mellon and that seemed to go well I wanted to see if you might be interested in doing it again? 21:24:06 <diablo_rojo> I don't need project descriptions till probably December so you have time. 21:24:13 <diablo_rojo> Just wanted to toss it out there :) 21:24:45 <diablo_rojo> Overall it would be very similar to the last bout, just a different university and timezone. 21:25:13 <diablo_rojo> This is a senior capstone course so the students are in their last 1-3 semesters of their CS degrees. 21:25:25 <diablo_rojo> Any questions? Or immediate thoughts? 21:27:02 <timburke> the CMU students were located in Qatar; do we know whether the north dakota students are there or elsewhere? 21:27:15 <diablo_rojo> They will be in North Dakota this time. 21:27:24 <diablo_rojo> So US Central. 21:27:33 <diablo_rojo> Good question! 21:27:43 <timburke> cool! should simplify some things :-) 21:27:51 <diablo_rojo> Yes for sure :) 21:28:01 <mattoliver> Jan to May, so a longer stint though. And might be mostly during my night 21:28:16 <timburke> i think we're interested -- just need to think of some good project(s) 21:28:19 <diablo_rojo> Yeah it'll be a full semester instead of a summer course. 21:28:36 <diablo_rojo> Maybe you can get acoles in on the party mattoliverau and timburke :) 21:28:46 <timburke> or even clayg :-) 21:28:53 <diablo_rojo> timburke: sweet! When would you like me to bother you again about this? :) 21:29:05 <timburke> yes please! 21:29:06 <diablo_rojo> Ah yes! clayg as well :) 21:29:19 <mattoliver> maybe we get policy migration fully working :P 21:29:42 <acoles> not a huge overlap with my time zone 21:29:45 <diablo_rojo> timburke: lol yes I plan to, but when should I poke again? How much time do you need to get the proposal together? 21:30:14 <timburke> how about we check in again after the ptg? 21:30:22 <diablo_rojo> timburke: you got it :) 21:30:31 <diablo_rojo> Any other questions from anyone? 21:31:06 <diablo_rojo> acoles: the students might not necessarily be working during normal US timezone hours since they are in school - they could be largely nocturnal for all I know. I was as a student :) 21:31:30 * diablo_rojo lets timburke have the floor back and moves to the back of the room 21:32:11 <fulecorafa> I also have a topic 21:32:34 <timburke> fulecorafa, go ahead! 21:32:55 <fulecorafa> So, just to update, we're moving on here on our development to object-locking 21:33:14 <fulecorafa> Just doing a heads up here and if you have any implementation suggestions, I'm all years 21:33:24 <fulecorafa> In the mean time, we're looking at https://wiki.openstack.org/wiki/Swift/ideas/object-lock 21:33:34 <fulecorafa> #link https://wiki.openstack.org/wiki/Swift/ideas/object-lock 21:34:18 <fulecorafa> I've seen this some months ago, we'll be joining the discussion soon. 21:35:07 <mattoliver> You coming to the PTG. Good place to discuss :) 21:35:52 <fulecorafa> Sorry, haven't quite looked at PTG yet. But will certainly bring it if I go 21:36:31 <mattoliver> its remote and free 21:36:49 <acoles> do we know who was the author of https://wiki.openstack.org/wiki/Swift/ideas/object-lock ? 21:36:52 <mattoliver> and we can put some time aside to talk about plans, ideas, or even progress. 21:37:01 <fulecorafa> There is a thread in mailing list 21:37:04 <fulecorafa> Just a sec... 21:37:15 <fulecorafa> #link https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/thread/PGHHFTIUU4DIICVPV4D4CSK2QYIJ5QUV/#UPR35ONG53AZWE4MF7VKLCX22GO63D72 21:37:46 <timburke> the hard parts seem likely to be around eventual consistency/partial failures, and maybe auth. i'd recommend thinking about some good probe tests 21:37:53 <timburke> (though it might be a little tricky given how small the "cluster" is) 21:37:57 <mattoliver> maybe even create a collab etherpad doc to brainstorm ideas async? better then the wiki probably. 21:37:58 <fulecorafa> Thanks mattoliver! I'll lookup some information on how to participate as soon as I'm available 21:38:44 <acoles> thanks fulecorafa 21:39:05 <mattoliver> I'd be interested in helping brainstorm etc. Always better with more heads and upstream from the start ;) 21:39:31 <fulecorafa> Also the etherpad seems like a great idea. Any place of discussion actually. 21:39:38 <acoles> adding any metadata to objects currently replaces all existing metadata 21:40:00 <fulecorafa> I also have some collegues that would like to participate in this conversation, although async, so yeah, 21:40:55 <fulecorafa> acoles yes but we're regularly using it for some other features. Maybe just being careful about changing this could be enough? 21:41:52 <timburke> or we develop some new "sticky meta" namespace, where metadata items are individually reconciled similar to account/container meta 21:42:27 <acoles> yes, it'll likely require inventing an entirely new other-metadata concept 21:42:34 <timburke> i seem to recall us talking about the idea some years ago, though i don't think anyone ever devoted any code to it 21:43:27 <acoles> we sort of have it for the special case of content-type 21:45:23 <fulecorafa> But that's it. Just giving a heads up. We'll be working on it in the next weeks/months 21:46:00 <mattoliver> cool! 21:46:25 <acoles> maybe there's some opportunity around Lock in that it'll be the *last* metadata update on the object, ... ever IIUC 🤷 21:47:27 <timburke> very cool! i'm not sure how best to help, but i'm interested! 21:49:13 <timburke> all right, i think i'll call the meeting a little early 21:49:24 <timburke> thank you all for coming, and thank you for working on swift! 21:49:28 <timburke> #endmeeting