21:03:55 <timburke_> #startmeeting swift 21:03:55 <opendevmeet> Meeting started Wed Oct 27 21:03:55 2021 UTC and is due to finish in 60 minutes. The chair is timburke_. Information about MeetBot at http://wiki.debian.org/MeetBot. 21:03:55 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 21:03:55 <opendevmeet> The meeting name has been set to 'swift' 21:04:03 <timburke_> who's here for the swift meeting? 21:04:09 <mattoliver> o/ 21:04:18 <seongsoocho> o/ 21:04:55 <acoles> o/ 21:05:02 <kota> o/ 21:06:06 <timburke_> agenda's at https://wiki.openstack.org/wiki/Meetings/Swift 21:06:12 <timburke_> #topic PTG 21:06:39 <timburke_> thank you all for attending the PTG last week! i feel like we had a lot of productive conversations 21:07:02 <mattoliver> +1 21:07:43 <timburke_> the main thing i want to do this week is figure out who's going to champion some of the action items we came up with 21:08:42 <mattoliver> great idea. 21:08:52 <timburke_> i've already offered to provide feedback for interop -- i haven't done it yet, but will try to for next week 21:10:23 <timburke_> who's like to take on "better defaults"? either to look over our current defaults and look for improvements, or to make specific changes/recommendations (like servers-per-port) 21:11:32 <mattoliver> I would, but might be better for someone who has a better swift ops background then me. 21:13:06 <timburke_> you could still make sure servers-per-port is prominently recommended ;-) 21:14:03 <mattoliver> Yes, I could do that :) 21:16:18 <timburke_> one part down -- any takers for the broader review? or, does anyone have other immediately actionable recommendations we'd want to make? 21:16:55 <mattoliver> How about if we create an etherpad of defaults and we can collaborate on what might be better? 21:17:20 <timburke_> π sounds like progress :-) 21:17:45 <mattoliver> OK, I'll put together an etherpad so we can iterate over it over the next few weeks/months 21:18:14 <timburke_> dark data watcher -- acoles am i remembering right that you were planning to circle back and review the outstanding patch? 21:18:31 <acoles> yes I ought to do that 21:18:48 <timburke_> i'll attach your name to it then ;-) 21:19:49 <timburke_> who want's to go abandon some old patches? so many clicky-clicks 21:21:08 <acoles> do we have a handy dashboard to show old inactive stuff, like 'the decade that wasn't' equivalent to 'the fortnight that was' ? 21:21:48 <acoles> probably not hard to sift them out if I put my mind to it 21:21:48 <mattoliver> True, we could make one of those. 21:22:20 <acoles> shall we all start with patches we own? I know I have a few that I should abandon 21:22:20 <timburke_> https://review.opendev.org/q/p:openstack/swift+is:open+age:52w should have all open swift patches that haven't been touched in a year 21:22:34 <acoles> π 21:22:51 <mattoliver> We also have a possibly abandoned part of an old dashboard we used to use.. I probably have it somewhere 21:23:08 <timburke_> we can -- i was debating about recommending we start there -- but then i quickly realized that the bulk of the cruft will be from people not here ;-) 21:24:31 <timburke_> fwiw that "only" goes on for 8 pages >.< 21:24:38 <timburke_> i can pick that up 21:25:06 <timburke_> who wants to orchestrate a bug squash? 21:26:06 <timburke_> my hope is that it'll mostly amount to collecting then divying up the list of bugs to give everybody else some to triage 21:27:39 <mattoliver> Here is a shortened version of a Swift Project dashboard we used to use.. might need to be updated to reflect later changes etc: https://bit.ly/3GsKmXg <-- hopefully that works 21:29:45 <mattoliver> We're writing these action items somewhere prominent right? so we can refer back and a place to comment and make progress on, like maybe the agenda wiki or something? 21:30:11 <timburke_> i plan on keeping them on the agenda (just not bringing them up *every* week) 21:30:24 <mattoliver> cool :) 21:31:29 <mattoliver> Last time we did a bug squash was around a PTG, and the way we ran that I remember being pretty good.. but can't quite remember. Was there an etherpad or something to track progress or something? 21:32:50 <acoles> maybe etherpad with list, people put nickname against a bunch and then strikethrough as they are done 21:33:20 <acoles> any that are hard to decide on can be put back into the pool and/or discussed if we do it in real time 21:33:31 <mattoliver> that rings a bell. 21:33:34 <timburke_> yeah, i think it was something like that. i think tdasilva did the collection 21:34:07 <mattoliver> We should get tdasilva to do it again, and drag him back into swift work :P 21:34:49 <mattoliver> I'll see what I can do about at least getting a start on the etherpad for this too. 21:34:58 <timburke_> my gut sense is that we ought to close most things we can't repro, but with a message that makes it clear that they can re-open if it's still an issue 21:35:03 <timburke_> great, thanks mattoliver! 21:35:05 <acoles> is the idea to do this sync or async? If sync, could we use one weekly meeting slot? or should we set up two times to suit US/EU and AP timezones? 21:35:29 <acoles> maybe first pass can be async 21:35:58 <acoles> we need timburke_ to hand out cookies to reward us :) 21:36:09 <mattoliver> yeah, start async, but sync definitely helps makes progress, so maybe both, let's see how we go. 21:36:14 <timburke_> or apple cider ;-) 21:36:26 <acoles> hehe, hic! 21:36:28 <mattoliver> oh apple cider.. I'll work for booze :P 21:36:57 <acoles> so, async via etherpad, then sync to discuss the trickier ones 21:37:14 <mattoliver> +100 21:37:23 <timburke_> idk -- depending on how my first batch goes, maybe you'd prefer it be nonalcoholic :P 21:37:57 <mattoliver> didn't qualify how good my work would be :P 21:39:05 <timburke_> sounds like a good plan -- maybe get the etherpad together for one meeting (next week, maybe?), ask people to claim a chunk to do sometime that week, with a goal of bringing some specific questionable bug back to discuss the following week 21:39:31 <timburke_> ok, next: drop logging translations 21:40:07 <timburke_> i refreshed https://review.opendev.org/c/openstack/swift/+/749810 last week -- who want's to take on the backend services? 21:41:46 <mattoliver> This was a pet peeve for Pete, should we ask if he's interested in getting out to eraser? :) 21:41:57 <mattoliver> *the 21:42:35 <acoles> I'll take a crack at it 21:45:10 <timburke_> part of me is tempted to allocate the rest to clayg and zaitcev ;-) 21:45:35 <mattoliver> And cschwede ;) 21:46:30 <mattoliver> Looks like seongsoocho already worked out about swift translations 21:47:38 <timburke_> i saw! seongsoocho do you know if there's more i need to look into to get it published? or maybe i should just wait for the first batch and see what happens :-) 21:48:00 <clayg> Yeah sign me up! But not if itβs boring and you really want it done. 21:48:16 <mattoliver> lol 21:48:53 <timburke_> clayg, i feel like you'd be good for either looking through the ops runbook or writing a "why swift" doc ;-) (or both!) 21:49:59 <mattoliver> With the amount of resources available to us do we maybe want to do a doc sprint after we get the bugs and defaults under control. 21:50:27 <timburke_> fair point. maybe we'll leave them for now. 21:50:32 <mattoliver> because that might need more people looking at it async 21:50:37 <clayg> Come join me around the fire and let me tell you of how we USED to run distributes storage. 21:50:49 <timburke_> all right, i *think* that covers the action items from the PTG 21:50:51 <mattoliver> the 'why swift' could be good though. 21:50:59 <timburke_> #topic open discussion 21:51:07 <timburke_> what else should we bring up this week? 21:52:29 <mattoliver> Oh I got a talk approved at the next australian linux conference (virtual again) around Swift tracing 21:52:43 <kota> congrats! 21:52:43 <timburke_> nice! 21:52:54 <mattoliver> so now I just need to clean it up and turn it into some kind of presentation rather then a stream of thought :P 21:54:19 <timburke_> i started playing around with trying to build usable wheels for pyeclib. might have some patches to propose in the not-too-distant future to support it 21:54:43 <mattoliver> cool 21:55:12 <timburke_> it bugs me that you've needed liberasurecode-dev(el) to even get swift installed for so long 21:56:03 <timburke_> then i started thinking about whether it'd make sense to include isa-l in the wheel too, cause that seemed like a useful thing 21:56:46 <timburke_> which in turn made me remember the state of jerasure. i'm thinking it might be time to finally deprecate it: https://review.opendev.org/c/openstack/pyeclib/+/815410 21:58:20 <timburke_> any objections? is it enough to do the standard python DeprecationWarning thing in pyeclib, or should i figure out how to do something vaguely equivalent in liberasurecode, too? 22:00:36 <timburke_> all right, that's about time 22:00:48 <timburke_> thank you all for coming, and thank you for working on swift! 22:00:52 <timburke_> #endmeeting