21:00:21 <timburke> #startmeeting swift
21:00:22 <openstack> Meeting started Wed Mar 31 21:00:21 2021 UTC and is due to finish in 60 minutes.  The chair is timburke. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:00:23 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
21:00:25 <openstack> The meeting name has been set to 'swift'
21:00:30 <timburke> who's here for the swift meeting?
21:00:58 <kota_> hello
21:01:14 <seongsoocho> o/
21:01:32 <acoles> o/
21:02:10 <rledisez> o/
21:02:19 <mattoliverau> o/
21:02:45 <timburke> as usual, the agenda's at https://wiki.openstack.org/wiki/Meetings/Swift
21:03:01 <timburke> mostly the same stuff as last week, but there's definitely some progress :-)
21:03:06 <timburke> #topic PTG
21:03:34 <timburke> i finally got around to booking our room slots for next month
21:04:08 <timburke> they're a little more scattered than i'd meant (since i was late in doing it, things were fuller), but i updated the etherpad with the rooms & times
21:04:15 <timburke> #link https://etherpad.opendev.org/p/swift-ptg-xena
21:04:44 <timburke> speaking of the etherpad, we all still need to add discussion topics ;-)
21:04:48 <mattoliverau> cool, thanks for getting that done timburke
21:05:54 <timburke> i suspect there will be a decent bit of replication, sharding, relinker talk -- but if there are other things you're interested in, please add them!
21:06:25 <timburke> speaking of those...
21:06:31 <timburke> #topic relinker
21:07:09 <timburke> so we merged https://review.opendev.org/c/openstack/swift/+/781908 to prevent get_hashes from creating empty partitions...
21:07:49 <timburke> ...then fairly quickly discovered that the new checks for whether a suffix is valid or not were too strict (at least on py2)
21:08:28 <timburke> thanks acoles for running that down yesterday and proposing https://review.opendev.org/c/openstack/swift/+/784108 -- it's working its way through the gate now
21:09:09 <timburke> and if anyone sees a centos7 probe test failure on a patch in the next day or so, that may be why
21:11:14 <timburke> there are a couple other relinker patches i wanted to call attention too, though: https://review.opendev.org/c/openstack/swift/+/769633 (for fanning out subprocesses to relink disks in parallel) and https://review.opendev.org/c/openstack/swift/+/783731 (to track which partitions were actually updated when relinking and only rehash those ones)
21:11:54 <timburke> thanks clayg for figuring out a mostly-sane way to mock out fork/wait for the disk parallelization
21:13:18 <timburke> so, things are moving; no immediate blockers
21:13:25 <timburke> any questions or comments?
21:14:27 <timburke> all right -- next up
21:14:33 <timburke> #topic shrinking and sharding
21:15:17 <timburke> acoles, i know we recently approved the repair and analyze commands patch, which was at the root of a chain or two
21:15:26 <timburke> where are we now?
21:16:03 <acoles> yes the repair command has merged https://review.opendev.org/c/openstack/swift/+/765624
21:16:08 <mattoliverau> \o/
21:16:25 <acoles> thanks mattoliverau and others ´for reviews
21:16:34 <mattoliverau> nps
21:16:44 <acoles> there's a couple of follow ups that I need to rebase, will do that tomorrow
21:17:46 <timburke> sounds good
21:17:48 <mattoliverau> I'm working through the new WIP extra ACTIVE states POC to see how that pans out for solving the quick shard shrink issue. Working through all the existing tests and want to write some more.
21:17:49 <acoles> on shrinking...some progress on reporting tombstones https://review.opendev.org/c/openstack/swift/+/782832 but that's still WIIP
21:19:07 <timburke> i was just about to ask mattoliverau about quick shard/shrinking :-) does it seem like more states are probably the way to go, then?
21:19:19 <mattoliverau> my WIP is: https://review.opendev.org/c/openstack/swift/+/782903 but I haven't pushed up a new version yet. But will today, to at least get it off my computer.
21:19:32 <mattoliverau> it is a little cleaner to be honest
21:20:02 <mattoliverau> because were storing state with more info, which we can encode into timestamp offsets, but states where kinda made for that.
21:20:45 <mattoliverau> If anyone has any thoughts, better ideas then https://etherpad.opendev.org/p/swift-shrinking-active-age
21:20:57 <timburke> cool. and acoles, i like the direction with tombstone counting -- it's nice that we get it pretty much "for free"
21:21:03 <mattoliverau> Or just want more info. I'll make sure that etherpad is updated today
21:21:57 <acoles> timburke: yep, on that one I need to work through all the plumbing
21:22:43 <timburke> great. any blockers or discussion points for any of that work? or questions that others have about it?
21:23:56 <mattoliverau> for anyone who wants to see acoles analyze in action, I have added a --analyze to my hacked up graph viewer script.
21:24:01 <mattoliverau> https://review.opendev.org/c/openstack/swift/+/775066
21:24:26 <mattoliverau> which pipes it through the analyze and then draws that patch on the graph in green.
21:25:32 <timburke> cool. i've been meaning to try that out
21:25:33 <acoles> mattoliverau: nice!
21:26:05 <timburke> those are the main topics i had
21:26:09 <timburke> #topic open discussion
21:26:26 <timburke> what else should we bring up this week?
21:27:51 <timburke> i brought the same retry-failed functionality to func tests that we've had for a bit now with unit and probe tests: https://review.opendev.org/c/openstack/swift/+/783091
21:28:29 <timburke> hopefully that'll further reduce how often we need to recheck, and allow us to merge patches faster/more reliably
21:29:29 <acoles> great
21:29:31 <mattoliverau> Nice
21:30:00 <mattoliverau> I've been doing a bit of recon stuff lately, so have also gone and done this: https://review.opendev.org/c/openstack/swift/+/783997 in the hopes to clean it up a little.
21:31:09 <zaitcev> I was tinkering with recon too, wanted to add --sharding. I'm surprised it didin't exist before, because the middleware has the support, just not the CLI.
21:31:10 <mattoliverau> thanks for the review on that timburke
21:31:52 <timburke> mattoliverau, happy to! i like the idea; it was probably overdue
21:31:56 <timburke> zaitcev, good thought!
21:31:59 <mattoliverau> yeah, I believe I had a patch to plumb through reconstructor somewhere too (from years ago) not sure it ever landed.
21:35:04 <timburke> seems like we covered most of the things people wanted to talk about
21:36:17 <timburke> do people generally feel like they get enough value out of this meeting to block off the time every week? i sometimes feel like i'm mostly repeating myself from the week before, and wonder if an every-other-week cadence would be better
21:36:44 <zaitcev> We don't have another meeting point aside from IRC, don't we.
21:37:21 <zaitcev> So, I had to bring things up in the meeting at times, to the extent of editing the agenda wiki.
21:37:58 <timburke> which is great! by all means, always feel free to add items to the agenda
21:38:49 <acoles> timburke: I'm ok with it, but we shouldn't feel we have to fill an hour - if there is nothing new, but it may still be good for people to have chance to touch base
21:39:52 <acoles> at the same time, its pretty lat ein the evening for me so I'd be happy with a 2 week cadence
21:40:51 <timburke> well, it's a discussion we can keep thinking about. i bet we'll talk about it again at the ptg, at the latest :-)
21:41:11 <timburke> but on that note
21:41:16 <timburke> thank you all for coming, and thank you for working on swift!
21:41:22 <timburke> #endmeeting