21:00:10 <notmyname> #startmeeting swift
21:00:11 <openstack> Meeting started Wed Jun 20 21:00:10 2018 UTC and is due to finish in 60 minutes.  The chair is notmyname. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:00:12 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
21:00:14 <openstack> The meeting name has been set to 'swift'
21:00:18 <notmyname> who's here for the swift team meeting?
21:00:24 <torgomatic_> .
21:00:34 <timburke> o/
21:00:42 <kota_> hi
21:00:51 <tdasilva> hi
21:00:59 <clayg> o/
21:01:04 <mattoliverau> o/
21:01:33 <acoles> hi
21:01:44 <notmyname> welcome, everyone. thanks for coming
21:01:52 <notmyname> agenda, as always, is at
21:01:53 <notmyname> #link https://wiki.openstack.org/wiki/Meetings/Swift
21:02:08 <notmyname> only topic on the agenda is to discuss https://review.openstack.org/#/c/427911/
21:02:09 <patchbot> patch 427911 - swift - Replace MIME with PUT+POST for EC and Encryption
21:02:30 <clayg> nice - short meeting
21:02:37 <notmyname> clayg: let's see :-)
21:02:38 <clayg> is zaitcev here tho?
21:02:45 <rledisez> hi o/
21:02:47 <notmyname> acoles: this is something you wanted to bring up
21:02:47 <zaitcev> o/
21:02:55 <notmyname> acoles: and zaitcev are both here
21:03:09 <notmyname> acoles: you want to bring up the reason for it being on the agenda?
21:03:16 <acoles> oh right
21:03:20 <acoles> yes
21:03:27 <zaitcev> well, there's not much to discuss today, then
21:03:41 <zaitcev> I must apply acoles' comments
21:04:06 <acoles> well, really it was just to discuss if we could move this forwards - and who hs the time available to do that
21:04:26 <zaitcev> or, actually, squash the patch 575512
21:04:27 <patchbot> https://review.openstack.org/#/c/575512/ - swift - PUT+POST: simplify object server POST handler, oth...
21:04:47 <acoles> zaitcev has made a great start, I've done some hacking and its passing func tests, but I think there's some more unit test work likely needed
21:04:49 <zaitcev> Kota promised another one, I have not seen his yet
21:05:08 <zaitcev> brb
21:05:18 <clayg> well I think we should make time, it's one of the higher items on my radar from a architecture stand point - technical investment required to unblock lots of awesome stuff
21:05:29 <notmyname> clayg: +1
21:06:28 <notmyname> do patches need to be squashed first? or should they land as a chain?
21:06:53 <acoles> notmyname: in this case, squash - the parent doesn't pass tests
21:06:56 <clayg> zaitcev: I'm pretty sure if kota_ does have time to review - he'd like to see the best foot forward with all of acoles stuff squashed - so doing some mechanical gerrit stuff is probably the very very next step
21:07:36 <clayg> I also want to see acoles stuff squahed and then try to get it a once over - if there's unittest gaps I'm sure once it's loaded in my brain you'll get some more tests to squash
21:07:43 <acoles> zaitcev: I'm keen to know if any of the simplifications I made actually broke something subtle that you were aware of?
21:07:56 <notmyname> zaitcev: how do you feel about managing the squashing and updating of the patch? would you prefer to be the one to do that, or are you comfortable with someone else pushing over it with updates?
21:10:12 <acoles> also, I'd like to revisit the object server version detection, ideally I'd like to avoid the path mangling, and I have an idea how
21:11:10 <notmyname> I thought zaitcev was looking for a patch. but I think he may have actually stepped away
21:11:46 <mattoliverau> He did say brb.. a while ago
21:11:54 <clayg> acoles: i'm interested if it's significantly better than the path mangling - i have to say as someone who has written a non-trivial amount of uncommited backend cleanup and inspecting/audit scripts - I'm a fan of a reasonably sane object-server api
21:12:06 <clayg> e.g. I regret not making storage policy part of the path!
21:12:35 <torgomatic_> acoles: does your idea have room in it for an explicit version field for the next protocol change?
21:12:40 <clayg> well, acoles how are you going to brain dump your idea for better than path mangling transparent upgrade?
21:12:50 <acoles> clayg: best thing is if I write a patch!
21:12:50 <clayg> torgomatic_: OMG yes please please please
21:13:10 <clayg> acoles: next best thing is to give the idea to someone who can write the patch
21:13:26 <acoles> torgomatic_: in the path? no. In headers, yes.
21:13:36 <torgomatic_> as long as it's in there somewhere :)
21:13:49 <clayg> it's annoyingly difficult to find a middle ground between "sketch" to "working code"
21:14:31 <clayg> as soon as you start to write it seems like "well, I might as well just write the code" - then next thing you know you're fixing unittests...
21:14:31 <acoles> torgomatic_: clayg : its still a bit formative, I'm just flagging up that there's still some stuff I'd like to consider on the PUT+POST
21:15:12 <clayg> torgomatic_: have you looked at the PUT+POST path stuff?  or do you remember it from Dublin?
21:15:26 <acoles> clayg: yeah, like i have said before, tests are such an impediment to progress :P
21:15:27 <zaitcev> notmyname: I meant to squash everything that comes, but if someone else is faster...
21:15:32 <torgomatic_> clayg: I've looked at it a little, but I'm kind of fuzzy on it
21:15:41 <zaitcev> notmyname: should only take a day or two
21:16:29 <notmyname> zaitcev: ok
21:16:37 <clayg> zaitcev: I don't think anyone *here*is complaining about YOUR pace - your turn-around time on getting reviews on that patch is mesured in #'s of hack-a-thons
21:16:46 <clayg> :D
21:16:58 <zaitcev> ok. So I proceed then...
21:17:05 <clayg> mattoliverau: can sympathize - but OTOH!?  did anyone mention that sharding is awesome!!!???
21:17:25 <clayg> notmyname: was just saying sharding is awesome...
21:17:42 <clayg> not as awesome as mattoliverau  - but still... pretty cool code
21:17:42 <zaitcev> if sharding works, people will ask for sharded accounts eventually
21:17:52 <clayg> zaitcev: job security!
21:17:56 <zaitcev> brb again
21:18:07 <notmyname> oh yeah, as a footnote to sharding, we've got a customer that's already sharded a bunch of larger containers. it's going really well
21:18:21 <clayg> woot woot!
21:18:31 <mattoliverau> Aww ta, and yeah things can go slow, but then it's awesome when acoles, clayg and timburke get involved to polish into something really awesome
21:18:35 <kota_> cool
21:18:47 <clayg> mattoliverau: it takes a village!
21:18:51 <mattoliverau> Sweet
21:19:01 * clayg hugs
21:19:24 <notmyname> on the PUT+POST patch, acoles can you summarize the next few things that need to happen?
21:20:53 <acoles> 1. assuming zaitcev approves, merge my follow up patches 2. I suspect more unit tests and/or modifying some to exercise the new protocol 3. possibly rethink the version discovery
21:21:33 <notmyname> that sounds like a lot, potentially
21:21:34 <acoles> 4. review and merge!!
21:22:46 <acoles> notmyname: item 2 could be quite a lot of churn , hard to know until we go there
21:22:56 <notmyname> ok
21:23:31 <notmyname> acoles is working on this. clayg you said you'd look, right? sounded like clayg wanted to rope torgomatic_ in too
21:23:53 <clayg> I'm *always* down to rope torgomatic_ into it
21:24:51 <acoles> we'll need to make some current unit tests conditional on proxy using old protocol, plus add new tests for new protocol :/
21:25:05 <clayg> I know I hate the current path mangling for the protocol thing - the problem is "working code wins" - if I had a better idea, I'd be tempted to write it up
21:25:26 <notmyname> torgomatic_: what are your thoughts on the matter?
21:25:33 <clayg> my problem is a lack of a better idea - not motivation to write code that keeps us from getting stuck with the protocol I don't like
21:26:06 <clayg> so a braindump from acoles might be useful to me
21:26:28 <acoles> clayg: if you can hang around in -swift after the meeting I'll share my thoughts
21:26:51 <acoles> like I said, they are formative (aka 'half-baked')
21:27:06 <clayg> acoles: I don't want to keep you up, would it be easier to organize your thoughts in a google doc async (maybe easier to share later as well)
21:27:08 <torgomatic_> the only way I thought to do it without the path mangling was different HTTP verbs, but I never did anything with that
21:27:09 <mattoliverau> Gotta start somewhere :)
21:27:18 <clayg> also, realistically we probably have "some time"
21:27:26 <clayg> zaitcev: said he needs a couple of days just to squash
21:27:37 <torgomatic_> call PUT2, get 405 -> fall back, for example
21:27:39 <acoles> clayg: ok, I'll write something down tomorrow
21:28:12 <clayg> acoles: wfm - is it better than PUT2 (I like PUT2 only a little bit more than the akward path thing that's there now)
21:28:44 <clayg> maybe I could add on my TODO to explain what's there now and zaitcev can correct any parts that are off
21:28:58 <acoles> clayg: well, IDK, you can be the judge of that
21:29:28 <notmyname> do we have other topics to discuss this week?
21:29:49 <clayg> idk, PUT2 isn't so bad ... STAGE?  idk...
21:30:05 <clayg> notmyname: how long until Denver?
21:30:29 <mattoliverau> 3 months
21:30:30 <notmyname> september 10
21:30:38 <notmyname> thanks mattoliverau :-)
21:30:53 <clayg> plenty of time :D
21:31:03 <mattoliverau> It's kinda in my heat being the babies due date :p
21:31:10 <mattoliverau> Head
21:31:23 <notmyname> well I hope we have PUT+POST landed long before Denver
21:31:40 <clayg> WFM!
21:32:37 <tdasilva> grpc everything, just romain
21:32:50 <tdasilva> ;)
21:32:53 <kota_> lol
21:33:22 <notmyname> if we've got no other topics, then let's move deeper discussion back to -swift
21:33:42 <notmyname> thanks for your work on swift
21:33:49 <notmyname> #endmeeting