21:03:27 <timburke> #startmeeting swift
21:03:27 <opendevmeet> Meeting started Wed Oct 18 21:03:27 2023 UTC and is due to finish in 60 minutes.  The chair is timburke. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:03:27 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
21:03:27 <opendevmeet> The meeting name has been set to 'swift'
21:03:34 <timburke> sorry i'm late
21:03:59 <timburke> who's here for the swift meeting?
21:04:02 <mattoliver> o/
21:04:08 <kota> y/
21:05:02 <timburke> big topic for this week -- PTG's next week!
21:05:30 <timburke> thanks everyone for voting for times on https://framadate.org/liKzqQw5lZQW8Pvf
21:06:54 <timburke> looks like while the block starting around this time would work fairly well for Monday, the rest of the days 1300-1800 UTC might work better
21:07:46 <mattoliver> yeah, well having the ops feed back session at busy time always makes sense.
21:08:15 <kota> sounds ok
21:08:24 <timburke> mattoliver and kota, am i right to think it'd be better to not try to shift the times around? i feel like that'd make for a really long first day for you guys
21:08:25 <mattoliver> But I'm easy.. I just wont be as productive during work during normal hours :P
21:09:10 <kota> any of things is ok because i'm at home those days for some reasons
21:09:44 <timburke> we can also aim for somewhat shorter sessions -- i don't really *want* to keep mattoliver up until 5am
21:09:47 <kota> i got conjunctivitis in the last weeks and i have to stay home until the end of this month.
21:09:56 <mattoliver> 1800UTC is getting a little late, that'll be 5am
21:10:10 <kota> so there is no plan to go outside.
21:10:15 <timburke> oh no! hope you feel better soon kota
21:10:19 <mattoliver> oh yeah, obviously my irc is lagging :P
21:10:37 <mattoliver> oh no kota :(
21:10:39 <timburke> seems like all the more reason to make sure you can get a good night's sleep
21:10:58 <kota> :)
21:12:34 <timburke> i'll start booking rooms after this meeting
21:12:39 <mattoliver> kk
21:13:26 <indianwhocodes> o/
21:13:37 <timburke> and i'll try to flesh out a few of my topics some more on https://etherpad.opendev.org/p/swift-ptg-caracal
21:14:28 <timburke> if you have anything else you'd like to bring up at the PTG, please add it to the etherpad!
21:14:29 <mattoliver> yeah I just added a new one based off discussions with SRE yesterday
21:14:50 <mattoliver> will keep adding as we think of things
21:14:59 <timburke> i saw that! i think i like it
21:16:02 <timburke> i was having some other thoughts around recon recently, but somewhat going the other way -- have recon-cron emit gauges that could get picked up by something like statsd_exporter
21:16:18 <timburke> #link https://review.opendev.org/c/openstack/swift/+/895737
21:16:34 <mattoliver> oh yeah, well seems like a good topic to talk about at the ptg then!
21:19:06 <timburke> it also got me thinking about two directions we could take on recon-cron -- (1) fold it into the object-updater, making it the only place that crawls async dirs, or (2) expand it to also do other monitoring, like counting primary/handoff parts and dropping that in recon/stats, too
21:21:28 <mattoliver> oh, interesting idea. both have merits. I like less things walking the dirs.. but having the primary, handoff, deep handoff stats is so useful I think upstream should get that data
21:22:28 <timburke> absolutely -- though it could also be done in some extra process/thread the replicator/reconstructor spins off
21:23:05 <mattoliver> cough audit watcher cough :P
21:23:08 <timburke> the biggest hesitation i've got about folding it into the updater is losing visibility when you've got your updater paused for whatever reason
21:23:23 <mattoliver> yeah
21:23:34 <mattoliver> that's not great
21:24:58 <mattoliver> random thoughts, link into the kernels inotify and look for asyncs that way, get triggered to run rather then walking?
21:25:13 <timburke> oh, now *there's* a thought... wouldn't help with the async counting today... but maybe it'd be reasonable to have it attempt some kind of audit for those, too?
21:25:14 <mattoliver> ie when a new async it written or removed
21:25:45 <mattoliver> same could we said for an alert to quarantine
21:26:06 <timburke> it also seems to have a lot of overlap with some of clayg's overseer ideas
21:27:45 <timburke> anyway, we can discuss this more at the PTG :-)
21:28:02 <mattoliver> I feel a nerd-snipe coming one.. good thing PTG is soon ;)
21:28:06 <mattoliver> *on
21:29:29 <timburke> that's the only thing i wanted to bring up, and to thank everyone for being so flexible on scheduling
21:29:32 <mattoliver> async counting, when started could walk to get current count. Insert a inotify to the kernel and then start counting/decrementing based on events in the async folder to keep the count up to date.
21:29:57 <mattoliver> periodic walk to keep in line (if we don't trust it).
21:30:23 <mattoliver> Oh yeah, we're so close to PTG, no need to bring many more topics up in this meeting today :)
21:30:33 <timburke> heh
21:30:34 <timburke> #topic open discussion
21:30:42 <timburke> anything else we should discuss this week?
21:30:57 <timburke> seems like there's a good number of patches all making pretty good progress
21:32:24 <timburke> mattoliver, just as a heads-up i proposed (and merged) https://review.opendev.org/c/openstack/swift/+/898440 as a follow-up to https://review.opendev.org/c/openstack/swift/+/897108 (silly yaml)
21:32:59 <indianwhocodes> I was wondering if this patch was in our radar  https://review.opendev.org/c/openstack/swift/+/799866
21:33:26 <mattoliver> kk, thanks timburke
21:33:44 <indianwhocodes> 799866: Let WSGI apps signal that a client connection should be closed currently seems to have a merge conflict which I attempted fixing
21:33:54 <timburke> indianwhocodes, it hasn't been, but probably should be!
21:34:29 <mattoliver> yeah, timburke (not that I've looked) but maybe we should start using priority reviews page again
21:35:11 <mattoliver> I keep getting side tracked with other work. But slowly working through timburke swift-reload patch chain. We've been running it for a while in prod and it's pretty great.
21:36:23 <timburke> priority reviews page is a great idea -- i've been remiss in tending it
21:36:32 <mattoliver> chain starting:
21:36:34 <mattoliver> #link https://review.opendev.org/c/openstack/swift/+/833174
21:37:05 <mattoliver> we all have.. but maybe we should start using it. There are alot of patches out there that could do with some love.
21:37:21 <indianwhocodes> <3
21:37:35 <mattoliver> I myself need to get better at reviewing again.. seems I'm not good at multi-tasking :P
21:38:09 <indianwhocodes> part-num api is wrapping up, so recalibrating my swift-priorities
21:38:36 <indianwhocodes> same here, can expect more reviews from me now.
21:40:09 <timburke> 👍
21:41:09 <mattoliver> I got nothing else seeing as the PTG starts next week, can't wait to talk and see everyone again!
21:41:16 <mattoliver> just wish it was in person
21:41:21 <timburke> at least we did a pretty good job of merging what was on the priority reviews page -- 7 of the ten patches that were listed when i last updated it have now merged
21:41:32 <mattoliver> \o/
21:41:33 <indianwhocodes> +2
21:42:28 <timburke> ditto -- i'd much prefer being jet-lagged but all together to having nearly everyone awake at odd hours while still also needing to do normal life things
21:43:06 <mattoliver> ^ 100x this
21:43:45 <kota> agree
21:44:45 <timburke> otoh, i suppose it's nice that we'll get some kota time! idk that we would otherwise, with the conjunctivitis ;-)
21:46:14 <timburke> all right, i think i'll end the meeting a little early
21:46:32 <timburke> thank you all for coming, thank you for working on swift, and see you all next week!
21:46:35 <timburke> #endmeeting