21:03:18 <notmyname> #startmeeting swift
21:03:19 <openstack> Meeting started Wed Feb 20 21:03:18 2019 UTC and is due to finish in 60 minutes.  The chair is notmyname. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:03:20 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
21:03:22 <notmyname> swift team meeting time!
21:03:23 <openstack> The meeting name has been set to 'swift'
21:03:25 <notmyname> who's here for it?
21:03:28 <timburke> o/
21:03:30 <kota_> o/
21:03:32 <alecuyer> o/
21:03:36 <tdasilva> hi
21:04:24 <clayg> sorry, thanks
21:04:29 <clayg> typey typey
21:04:32 <mattoliverau> o/
21:04:52 <notmyname> kota_: thanks for reminding me of the meeting!
21:05:01 <kota_> :-)
21:05:32 <kota_> no one else at my desk in my home in the morning ;)
21:05:45 <notmyname> oh, there's zaitcev!
21:05:53 <notmyname> I'd like to check in on the py3 work.
21:06:11 <notmyname> zaitcev: this morning timburke told me that with your work on the sharder, we'll soon be at about 60% coverage with py3?
21:06:27 <zaitcev> notmyname: It's only unit tests.
21:06:40 <zaitcev> out of all tests it's about 41%
21:06:54 <zaitcev> Unit tests are about 80% of all tests
21:07:19 <notmyname> that's still good though!
21:07:38 <timburke> once we have things to the point that unit tests pass, i'm pretty sure we'll have services able to start. from there, we could run func tests under py2 against swift on py3
21:08:08 <timburke> ...because porting func tests will be its own special hell, as i recall
21:08:31 <notmyname> that will be a great point to get to.
21:08:32 <zaitcev> Still, it's way slower than I need.
21:09:00 <zaitcev> Cyril disappeared this week. He's helping py3 on other places too... Like, Sahara.
21:09:05 <timburke> zaitcev, how much needs to be ported for RH to consider including it?
21:09:46 <zaitcev> timburke: Basically everything has to be there. Including obscure middleware, and the middleware I gave up upon, such as bulk.
21:11:01 <notmyname> for example, functests can run on a different system on py2 if the services are running under py3
21:11:15 <timburke> yeah, i'm fine with everything under swift/ and bin/... i'm debating about how much value there is to be had in requiring func tests run under py3 if we can get to ^^^
21:11:32 <zaitcev> Well... I don't remmeber, but I think we may be running in-process functests.
21:11:55 <notmyname> hmm.. ok
21:11:59 <zaitcev> So, probe tests is the only thing that can be postponed.
21:12:13 <notmyname> zaitcev: is there anything specific you need
21:12:57 <timburke> we do in-process func tests. we also do "spin up swift, then separately, run func tests".
21:13:00 <notmyname> or... should we update https://wiki.openstack.org/wiki/Swift/PriorityReviews ?
21:13:03 <zaitcev> notmyname: more manpower, I think. Someone needs to adapt a thousand tests by adding b'' to everything.
21:13:18 <zaitcev> I looked at slo, 83 failed
21:13:58 <zaitcev> If we can push container through, it unlocks a bunch of dependencies. Like dlo.
21:14:10 <timburke> i'm actually hesitant to touch the func tests too much ATM -- i think it's going to be tricky verifying that we're still testing the same thing when running them under py2 vs py3
21:14:15 <notmyname> zaitcev: on slo?
21:14:26 <mattoliverau> I'll focus on the container patch today.
21:15:07 <mattoliverau> Got through a bunch of work stuff yesterday to free up more time to take a look today.
21:15:35 <zaitcev> mattoliverau: thanks a lot. As you can see, it started as a "sharder CLI" patch, which snowballed.
21:15:48 <mattoliverau> lol, yeah, I noticed that :)
21:16:09 <mattoliverau> thanks for all that work and also sorry :P
21:16:18 <notmyname> zaitcev: as you go through those patches, please highlight them in IRC and on https://wiki.openstack.org/wiki/Swift/PriorityReviews
21:16:31 <zaitcev> timburke: It is a concern. Worst comes to worst, I can ask people doing the release not to run in-process when building.
21:16:35 <notmyname> keeping them in front of the rest of us, I suspect, is a good way to get faster progress
21:16:42 <timburke> heh, sorry. when i start to review py3 stuff, i tend to start poking at the other bits around it and, well, here we are
21:17:41 <mattoliverau> I just keep asking questions about strings, byte strings and encoding :P
21:18:11 <zaitcev> By all means, ask away.
21:18:40 <zaitcev> I'm happy to repeat all explanations, maybe you'll see a logical inconsistency.
21:19:04 <notmyname> timburke: mattoliverau: if you could highlight the py3 patches, too, as you work on them, that would be great
21:19:12 <timburke> for sure, though, running the func tests is going to be the only way that i'll start to feel really good about declaring support. if it wasn't for those, i never would have found https://github.com/python/cpython/pull/7932
21:20:10 <notmyname> is there anything else about py3 that we need to bring up today in the meeting?
21:20:39 <timburke> found a bug in eventlet that's gonna block us porting ssync: https://github.com/eventlet/eventlet/pull/548
21:20:55 <timburke> good news is, they seem to be on board with fixing it
21:21:39 <notmyname> alecuyer: kota_: rledisez: what's up with LOSF this week?
21:22:10 <alecuyer> Earlier today I looked at grpcio/eventlet again,
21:22:42 <notmyname> was it amazing?
21:22:46 <alecuyer> surprisingly the latest grpcio release (1.18.0) appears not to hang anymore with eventlet - just running the functests on SAIO
21:22:47 <clayg> and the survey says: threads, not so bad after all!
21:22:51 <alecuyer> well it may be :)
21:22:56 <notmyname> nice!
21:23:00 <alecuyer> So I have to dig into this
21:23:01 <kota_> nice
21:23:08 <clayg> those google guys... I think they're pretty smart
21:23:10 <alecuyer> see where it started working and why
21:23:22 <notmyname> that's expecially good, if tim's putting patches in eventlet for py3 support :-)
21:23:22 <alecuyer> and if we can count on it working on later release or if i got lucky
21:23:36 <timburke> clayg, well, at least one of them is :-)
21:23:41 <timburke> i miss torgomatic...
21:23:49 <clayg> alecuyer: 😍 "just cause it works doesn't mean I trust it enough to ship it"
21:23:58 <clayg> I need to _see__ the fix
21:23:58 <alecuyer> clayg : thing is, they claim specifically that it will not work with eventlet, so im wary
21:24:14 <alecuyer> so anyway I will give more details as soon as I find it
21:24:19 <notmyname> thanks
21:24:34 <notmyname> kota_: it looks like you have been merging from master. thanks for that
21:24:39 <notmyname> rledisez: have you attempted that yet?
21:24:52 <alecuyer> (I think rledisez isn't here, he's on vacation)
21:24:57 <notmyname> ah, ok
21:25:13 <notmyname> ok, what else do we need to talk about in the meeting this week?
21:26:26 <clayg> Does anyone know for sure that they ARE interested in running for PTL?
21:27:00 <clayg> Or do we draw straws, or just make zaitcev do it?
21:27:24 <timburke> i feel like there's three slots: i definitely *don't* want that job, i definitely *do* want that job, and i'd be *willing* to do that job if nobody else feels strongly
21:27:47 <timburke> i'm kinda leaning toward the last one, personally
21:27:47 <clayg> right - if someone was in slot 2 that would be AMAZING
21:28:14 <clayg> like, you don't even have to do it a 100 years like notmyname - we promise to let you out if it's even worse than you imagined
21:28:30 <zaitcev> I'm with timburke on this. I could do it, but then I'll run the whole project into ground.
21:28:42 <clayg> 🤗
21:29:29 <timburke> i firmly *don't* think anyone here would run the project into the ground
21:29:53 <timburke> i think that's part of why i'm in the camp i am
21:29:57 <clayg> this project runs itself, notmyname's be coasting :P
21:30:08 <notmyname> lol
21:30:15 <clayg> LOL 🤗
21:30:37 <timburke> clayg, it's like we're used to thinking in terms of a bunch of independent actors, all trying to bring a system into a truer/more correct state or something...
21:30:57 <clayg> timburke: predisposed to organized chaos
21:31:01 <notmyname> one thing to consider is the required public speaking. if that's not something you're comfortable with, then there are major parts of the job that you will hate
21:31:53 <zaitcev> "Hello, I'm the PTL and I have no clue about my project, but here's what they tell they're working on... some small files or other."
21:33:31 <clayg> hrm.. ok, I was really hoping someone was thinking this might be good - I mean... in reality it probably would be good
21:33:32 <notmyname> ok, let me make a few suggestions :-)
21:33:57 <clayg> ... an opportunity to expand your role - stretch your soft skills, leadership
21:34:01 <mattoliverau> I'm currently in slot 3, Suse let's me spend time on it upstream, when I can. So not sure if PTL would mean they'd give me more time, or I'd be time limited which might not be great for the project.
21:34:11 <clayg> and of course force the rest of us to grow up and not "just make notmyname do it all the time"
21:34:42 <clayg> because realistically I'm not anyone of us thinks we dedicate the time and focus that notmyname has brought to us for all these years
21:34:42 <zaitcev> Right, talk to TC about our go component for LOSF
21:34:45 <notmyname> first, I suggest that there is an ongoing conversation about this. very few people will be comfortable discussing it in IRC (semi-public, logged), much less in an IRC meeting (logged, public, and more "formal")
21:34:54 <mattoliverau> lol, that's true. but when you have a notmyname what did you expect us to do :P He's too damn good at the role
21:35:00 <notmyname> figure out a good way to discuss it, irc or otherwise
21:35:02 <clayg> we don't need some one to "fill the shoes" - that's too daunting
21:35:27 <notmyname> second, be very clear that whoever starts the conversation is NOT volunteering
21:36:45 <mattoliverau> kk, I'll post back on the mailing list to get the ball rolling, we'll have the discussion there.
21:36:59 <mattoliverau> sorry not mailing list..
21:37:04 <timburke> hahaha
21:37:07 <notmyname> whew
21:37:33 <mattoliverau> A group email
21:38:02 <kota_> i see, thx mattoliverau
21:38:12 <clayg> Is there any by-laws that say PTL nominations have to be core?
21:38:14 * mattoliverau hasn't had any coffee so my brain isn't englishing very well :P
21:38:20 <notmyname> clayg: no
21:38:26 <clayg> just checking
21:38:26 <mattoliverau> nope
21:38:27 <clayg> ok
21:38:56 <mattoliverau> torgormatic for PTL!
21:38:59 <clayg> a'ight - i have another thing trying to get my attention... and I need to pee.  Sorry to bring it up!  :P
21:39:06 <kota_> mattoliverau: lol
21:39:17 <timburke> mattoliverau: we'll drag you back here, kicking and screaming!
21:39:19 <notmyname> I think that's a good place to leave it for this meeting :-)
21:39:54 <notmyname> thanks for coming today. and thank you for your work on swift
21:40:02 <notmyname> #endmeeting