21:00:05 <notmyname> #startmeeting swift
21:00:05 <openstack> Meeting started Wed Oct 21 21:00:05 2015 UTC and is due to finish in 60 minutes.  The chair is notmyname. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:00:06 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
21:00:09 <openstack> The meeting name has been set to 'swift'
21:00:21 <notmyname> hello, everyone. who's here for the swift meeting?
21:00:25 <mattoliverau> o/
21:00:25 <kota_> o/
21:00:26 <cutforth> o/
21:00:26 <wbhuber> Me
21:00:26 <minwoob> o/
21:00:27 <ho> o/
21:00:30 <m_kazuhiro> o/
21:00:31 <blmartin> o/
21:00:31 <joeljwright> evening
21:00:31 <peluse> yo
21:00:41 <tdasilva> hello
21:01:22 <torgomatic> hi
21:01:37 <notmyname> I hope everyone is doing well
21:01:46 <notmyname> we've got just a few days until the summit
21:01:50 * torgomatic is disappointed at the general lack of donuts when it comes to IRC meetings
21:01:51 <mattoliverau> \o/
21:02:07 <jrichli> yo
21:02:11 <mattoliverau> torgomatic: now I want a donut
21:02:19 <torgomatic> mattoliverau: join the club
21:02:38 <notmyname> torgomatic: there's Swedish candy over on this side of the office
21:02:57 <jrichli> like ... sweedish fish?
21:02:57 <notmyname> in this meeting I'd like to go over the schedule and do some prep for the summit sessions
21:03:05 <notmyname> jrichli: among many other things
21:03:08 <mattoliverau> torgomatic: maybe we need to find some donuts for next week ;)
21:03:21 <notmyname> jrichli: frankly, most of it is disgusting
21:03:27 <jrichli> lol
21:03:38 <notmyname> #link http://mitakadesignsummit.sched.org
21:03:43 <notmyname> there's the overall summit scheudle
21:03:50 <notmyname> #link https://wiki.openstack.org/wiki/Meetings/Swift
21:03:55 <notmyname> there's the swift schedule
21:04:02 <notmyname> so here's how I scheduled it
21:04:10 <notmyname> based on the experience in vancouver
21:04:25 <notmyname> instead of blocking each 40 minute track for one topic,
21:04:41 <notmyname> this time I grouped the sessions
21:04:50 * acoles apologises for being late
21:04:55 <notmyname> so eg there's 2 sessions, a break, then 3 sessions, then 2 sessions
21:05:30 <notmyname> so I took those session blocks and put them in the schedule as "work session N, where we'll cover Foo, Bar, and Baz"
21:05:38 <notmyname> that way, if we have something that goes long, that's ok
21:05:39 * joeljwright waves hello to acoles
21:05:47 <notmyname> and if we have something that is shorter, that's ok too
21:06:06 <notmyname> pretty much, I think that the summit is looking more and more like the swift hackathons ;-)
21:06:12 <mattoliverau> notmyname: great idea
21:06:26 <tdasilva> notmyname: I think that's a good thing :-)
21:06:27 <notmyname> mattoliverau: we'll see how it works :-)
21:06:34 <notmyname> yeah, i'm hopeful
21:06:55 <notmyname> so today, let's walk thought the blocks of sessions
21:07:17 <notmyname> goal is to have someone assigned homework for the topics mentioned to create an etherpad for it
21:07:43 <notmyname> in the interest of being productive with the time in tokyo, we need to have a goal for each topic
21:07:50 <notmyname> ok, so first one
21:07:56 <notmyname> #topic work session 1
21:08:11 <notmyname> this one is "encryption"
21:08:14 <notmyname> Topics to cover:
21:08:14 <notmyname> * Production Keymaster: https://etherpad.openstack.org/p/swift_production_keymaster_issues
21:08:14 <notmyname> * Ouststanding encryption issues: https://etherpad.openstack.org/p/swift_encryption_issues
21:08:25 <notmyname> good news! jrichli has already done etherpads for them :-)
21:08:45 <notmyname> (so not much to talk about on it right now)
21:08:52 <notmyname> #topic work session 2
21:08:58 <notmyname> this one has 3 topics
21:09:05 <notmyname> Topics to cover:
21:09:05 <notmyname> * container sync [eran]
21:09:05 <notmyname> * hummingbird: unifying the sync protocol [dfg, redbo]
21:09:07 <notmyname> * global clusters [kota]
21:09:15 <notmyname> Topics to cover:
21:09:15 <notmyname> * container sync [eran]
21:09:17 <notmyname> * hummingbird: unifying the sync protocol [dfg, redbo]
21:09:19 <notmyname> * global clusters [kota]
21:09:24 <notmyname> sorry for the double paste
21:09:33 <kota_> no worries
21:09:35 <tdasilva> missing a replica
21:09:39 <notmyname> :-)
21:09:46 <notmyname> doesn't look like eran is here
21:09:58 <notmyname> dfg_: around?
21:10:04 <jrichli> I can pass info along to eran
21:10:17 <notmyname> kota_: will you be able to come up with an etherpad on the global clusters topic?
21:10:20 <notmyname> jrichli: thanks
21:10:36 <kota_> yup, I'll make it after meeting.
21:10:40 <notmyname> great
21:10:45 <mattoliverau> note, the list of etherpads should go: https://wiki.openstack.org/wiki/Design_Summit/Mitaka/Etherpads
21:10:48 <notmyname> oh, and where do we put the ehterpads?
21:10:55 <notmyname> mattoliverau: yeah, there too :-)
21:10:57 <mattoliverau> I'll getup a swift session section
21:11:03 <notmyname> mattoliverau: thanks
21:11:05 <mattoliverau> *setup
21:11:07 <kota_> mattoliverau: thanks
21:12:21 <notmyname> dfg_: if you don't have an etherpad for talking about hummingbird, I'll make one
21:12:27 <notmyname> (you might think that's good or bad)
21:13:14 <notmyname> but I'm hoping we can specifically talk about making replication better by unifying the sync protocol. no reason to have a 3rd way for hummingbird. let's have 1 (well, 1 + rsync, in reality)
21:13:42 <notmyname> #topic work session 3
21:13:47 <notmyname> this one is on swiftclient
21:13:54 <notmyname> Topics to cover:
21:13:54 <notmyname> * Keystone session in swiftclient
21:13:54 <notmyname> * swiftclient docs
21:14:19 <notmyname> would anyone like to volunteer to make an etherpad for one or both of these?
21:14:46 <joeljwright> I'm happy to contribute to a docs etherpad
21:14:57 <notmyname> joeljwright: that would be great
21:15:08 <joeljwright> I can make some notes on the session one too
21:15:14 <joeljwright> just aware that I won't be at the meeting
21:15:17 <notmyname> right
21:15:55 <notmyname> #topic work session 4
21:16:04 <notmyname> Topics to cover:
21:16:04 <notmyname> * rings (data placement) [clayg]
21:16:04 <notmyname> * EC topics [peluse]
21:16:06 <notmyname> * symlinks [hrou]
21:16:29 <notmyname> clayg (where is he?) has been doing some cool stuff with ring updates
21:16:45 <notmyname> we need to talk about the improvements, and he's planning on facilitating that discussion
21:17:07 <notmyname> peluse: will you be able to facilitate an EC discussion on outstanding issues and what's next?
21:17:23 <peluse> sorry, in another meeting at the same time... one sec
21:17:36 <peluse> so sure, yeah
21:17:43 <notmyname> I think I heard "I'll volunteer for everything"
21:17:47 <notmyname> peluse: thanks :-)
21:17:51 <briancline> including donuts?
21:17:52 <peluse> :)
21:17:58 <notmyname> especially donuts
21:18:02 <briancline> great
21:18:05 <peluse> donuts and sake....
21:18:34 <notmyname> and then symlinks
21:18:47 <mattoliverau> +1000
21:18:49 <notmyname> I put hrou's name there
21:18:57 <notmyname> jrichli: where are all your people?
21:19:10 <jrichli> hmm .. donno about hrou
21:19:19 <tdasilva> they are all eating donuts
21:19:24 <notmyname> tdasilva: figures
21:19:28 <jrichli> i can ping him
21:20:00 <jrichli> his status is "in a meeting"
21:20:13 <notmyname> probably with peluse
21:20:28 <peluse> yeah, sure
21:20:30 <mattoliverau> lol
21:20:33 <notmyname> IIRC hrou is really interested in the symlink stuff
21:20:34 <briancline> lol
21:20:39 <jrichli> he'll be on soon
21:20:41 <hrou> notmyname, yes +1 : )
21:20:47 <notmyname> hrou: hi
21:20:55 <hrou> notmyname, and robfran (not online)
21:21:14 <notmyname> hrou: can you work on putting an etherpad together for an outline of the symlink topic?
21:21:33 <notmyname> we need a basic flow of the discussion + specific goals so we can make sure we make good use of the time
21:21:50 <notmyname> (and see the previous stuff in the meeting log to see how the sessions are put together)
21:21:53 <hrou> notmyname, absolutely !  As a quick heads up essentially the idea will be to go over the remaining etherpad issuse, I can give a quick overview of the code (to check its on the right track), that way we can make a lot of progress.
21:22:03 <notmyname> hrou: great. thanks
21:22:07 <hrou> notmyname, thanks a bunch !
21:22:34 <notmyname> in this work sessions (rings, ec, symlinks), I suspect that symlinks will take up the least amount of time
21:22:45 <notmyname> but actually I want to come back to that in just a moment
21:22:56 <notmyname> #topic work session 5
21:23:01 <hrou> notmyname:  I agree, maybe just a 10-15 min will suffice really, unless it leads to an interesting topic.
21:23:03 <notmyname> Topics to cover:
21:23:03 <notmyname> * container sharding [matt]
21:23:03 <notmyname> * fast-POST [acoles]
21:23:13 <mattoliverau> I'll take sharding :)
21:23:15 <notmyname> this last one is container sharding and fast-post
21:23:19 <notmyname> mattoliverau: thansk :-)
21:23:29 <notmyname> acoles: can you make one for fast-POST?
21:23:32 <acoles> sure
21:23:35 <notmyname> thanks
21:23:48 <notmyname> great, so passing out homework is done :-)
21:23:55 <notmyname> #topic overall schedule
21:24:08 <notmyname> on these work sessions, well, hang on....
21:24:33 <mattoliverau> I've started putting together the swift section on the wiki page about etherpads.. It'll be up this morning (my time).
21:24:33 <notmyname> basically, here's the whole view, now that we've gone through it
21:24:34 <notmyname> https://gist.github.com/notmyname/f2f1a1ea89ba6b2d4569
21:24:40 <notmyname> mattoliverau: thanks
21:24:46 <notmyname> ok, take a look at that gist
21:25:12 <notmyname> here's the question for everyone: do the work sessions look reasonable? are the right topics grouped? is there too much in any one block?
21:25:40 <peluse> yeah, sure
21:25:45 <tdasilva> good thing we have friday in case something spills over
21:25:46 <notmyname> peluse: :-)
21:25:50 <torgomatic> seems okay
21:25:58 <notmyname> yup. friday is still available
21:26:03 <notmyname> I think session 4 will be busy, but it's doable
21:26:04 <mattoliverau> looks good to me :)
21:26:16 <jrichli> +1
21:26:20 <kota_> +1
21:26:26 <notmyname> also, if we need to end up with "ok go finish that conversation on that side of the room and we'll do another over here" I think that's fine
21:26:28 <acoles> good plan
21:26:31 <hrou> looks great !
21:26:33 <notmyname> cool. thanks for your feedback
21:26:52 <notmyname> here's the link for the friday etherpad:
21:26:54 <notmyname> #link https://etherpad.openstack.org/p/tokyo-swift-contributors-meetup
21:27:17 <notmyname> add stuff there during the week as stuff overflows, or if there's stuff that needs to be addressed that isn't already scheduled
21:27:34 <notmyname> and last up with scheduling...
21:27:36 <notmyname> #topic fishbowls
21:27:45 <notmyname> we've got 2 fishbowl sessions
21:27:52 <notmyname> * ops feedback
21:27:52 <notmyname> ** https://etherpad.openstack.org/p/tokyo-swift-ops-feedback
21:28:15 <notmyname> this one is normal. if you've been to a summit before, you know exactly what to expect here
21:28:36 <notmyname> any other deployer/ops quesions (or comments) you have can be added to the etherpad there
21:28:45 <notmyname> so what about the last fishbowl?
21:29:17 <notmyname> since there wasn't one particular topic that we had that seemed to require a really big crowd, I wanted to do something like the work sessions
21:29:39 <notmyname> the fishbowls are more widely advertised, so this is the time for cross-project stuff
21:30:03 <notmyname> so I made a fishbowl session for "inbound cross project stuff" with 2 topics listed right now
21:30:13 <notmyname> first off, what does "inbound cross project" mean?
21:30:39 <notmyname> there's cross project stuff where we add a feature or make a change and another project needs to change
21:31:03 <notmyname> eg trove is rewriting the chunking to use SLOs instead of DLOs. and glance could be using service tokens
21:31:22 <notmyname> that's "outbound". we made a change and another project is doing something
21:31:27 <notmyname> inbound is the opposite
21:31:44 <notmyname> so there's a change or feature in keystone, and we need to implement it. stuff like that
21:32:07 <notmyname> https://etherpad.openstack.org/p/tokyo-swift-cross-project
21:32:12 <notmyname> there's the ehterpad for it
21:32:26 <notmyname> so I've got RBAC with oslo.policy on it
21:32:36 <notmyname> ho: I'd love some help on filling out that part of the etherpad
21:32:49 <ho> notmyname: thanks i will do it
21:32:53 <notmyname> thanks
21:33:07 <notmyname> and the other thing I added to this session is dependencies and requirements.txt
21:33:27 <notmyname> to bluntly summarize, what do we do about global requirements updates?
21:34:07 <notmyname> I've also tagged this fishbowl session in the cross-project track. so I hope to have some discussions
21:34:20 <notmyname> so...what do you think? (we can always change)
21:35:37 <notmyname> (this is where peluse is supposed to say "yeah, sure")
21:35:38 <mattoliverau> I think its a good use of time, good way to, hopefully, get some cross project work/networking happening too
21:36:17 <acoles> yes, good idea, e.g. be useful to discuss the implications of the proposed keystone service catalog change
21:36:36 <peluse> yeah, sure
21:36:38 <notmyname> acoles: right. I think there's another session on that
21:36:48 <notmyname> peluse: glad you feel that way :-)
21:36:49 <stevemar_> acoles: theres a whole x-project sesison on it
21:37:08 <acoles> notmyname: stevemar_ right, on tuesday?
21:37:12 <acoles> think i saw that
21:37:28 <stevemar_> acoles: notmyname, there's actually 2 on it: https://mitakadesignsummit.sched.org/event/e9b01e69686a60c5c61d16ee18f01391#.VigFgxCrRsY and https://mitakadesignsummit.sched.org/event/5a48dfba5a31795682120c92473f2f1e#.VigFgxCrRsY
21:37:40 <notmyname> tuesday has a ton of cross project stuff. so I'd definitely recommend participating in those as you see fit
21:37:58 <acoles> yes. i guess i was thinking we (swift) may need to spend some time considering what it means. idk.
21:38:48 <jrichli> integrating Castellan might be something to consider discussing, but I am not sure if we have enough content yet
21:38:49 <notmyname> acoles: I'm actually feeling better about what it means for swift. seems problematic for users, but that's a different question
21:38:59 <notmyname> jrichli: oh? that could be good
21:39:10 <notmyname> of course we only have 40 minutes for this second fishbowl
21:39:11 <jrichli> i will see what we can provide
21:39:15 <stevemar_> notmyname: i'm sure we can come up with something where we don't break the world :)
21:39:21 <notmyname> stevemar_: :-)
21:40:17 <notmyname> I think that's all I have about summit planning?
21:40:21 <notmyname> #topic else:
21:40:22 <acoles> stevemar_: thanks, i have those sessions on my schedule
21:40:40 <notmyname> what else is going on? anything we need to address before we see each other in person in a few days?
21:41:00 <acoles> our tests are broken in places it seems :(
21:41:15 <notmyname> acoles: how'd that happen?
21:41:17 <notmyname> which tests?
21:41:23 <notmyname> when?
21:41:33 <notmyname> TELL ME MORE!!
21:41:33 <acoles> kilo func tests don't run
21:41:50 <notmyname> weird dependency things?
21:42:07 <acoles> there's a patch to fix it but then some tests fail, i think cos something changed in devstack config of keystone users
21:42:32 <acoles> and i aslo notices that gate-swift-docs doesn't fail on rst errors
21:42:47 <acoles> so some of our doc has gone missing
21:42:49 <notmyname> acoles: is anyone but you looking at the kilo errors yet?
21:43:04 <acoles> notmyname: yes, jordanP spotted it
21:43:15 <notmyname> ok
21:43:17 <acoles> think it is jordanP nick, he's not here
21:43:27 <acoles> and maybe cschwede ?
21:43:31 <notmyname> ok
21:43:42 <notmyname> and the doc errors?
21:43:52 <acoles> the fix is a backport of a cschwede patch but it reveals the failing tests :/
21:44:17 <acoles> doc errors - well i fixed them here https://review.openstack.org/237013 but i don't yet know what to do about the gate job
21:45:17 <notmyname> ok
21:45:35 <notmyname> I wonder what we can do to make the docs failures more noisy
21:45:59 <acoles> notmyname: re func tests, we could patch kilo to skip the 5 failing tests (they test cross domain users and ACLs) so we can at least get the test job to run again??
21:46:12 <acoles> notmyname: does that sound acceptable? ^
21:46:14 <notmyname> acoles: that sounds better than where we are now
21:46:19 <acoles> right!
21:46:27 <acoles> N - 5 >> 0
21:46:29 <acoles> :)
21:46:50 <notmyname> yeah. but it would be good to know what's changed to cause them to stop running overall
21:47:07 <notmyname> why would skipping tests cause the rest to run if nothing runs now?
21:47:13 <acoles> notmyname: i'll look at that tomorrow then. i can see where devstack fails to set things up but i dont know why it fails
21:47:18 <notmyname> ok
21:47:29 <notmyname> thanks
21:47:50 <acoles> notmyname: 1. skip tests 2. fix the other stuff that stops tests running -> tests pass 3) figure out skipping tests
21:48:02 <notmyname> ah, gotcha
21:48:30 <acoles> notmyname: see https://review.openstack.org/#/c/235933/ for the fix to make tests run again
21:48:46 <notmyname> thanks
21:48:49 <notmyname> I'll take a look
21:49:11 <notmyname> acoles: thanks for bringing those up
21:49:16 <notmyname> anything else from anyone?
21:49:18 <cutforth> on the "other" topic, i have one small logistical item for this meeting
21:49:20 <acoles> i wish jenkins would fail when 0 tests are run!
21:49:38 <mattoliverau> https://wiki.openstack.org/wiki/Design_Summit/Mitaka/Etherpads#Swift
21:49:43 <notmyname> cutforth: what's up?
21:49:53 <notmyname> mattoliverau: thanks
21:50:00 <cutforth> just a friendly reminder that the next time this meeting is held, the US will have gone out of daylight savings time
21:50:02 <mattoliverau> ^ That is an initial version, place etherpad links there.
21:50:10 <cutforth> DST stops on Nov 1
21:50:17 <notmyname> cutforth: thanks for the info
21:50:21 <cutforth> unless you are in Arizona (peluse:)
21:50:40 <notmyname> the meeting is set for UTC, so that will mean that it will relatively change for people in the US
21:50:47 <hrou> I was going to say I didn't think all states participated in that ; )
21:50:48 * cutforth remembers missing this meeting in the past
21:50:51 <mattoliverau> cutforth: for those of us in Australia, we went into day light savings last week
21:51:05 <notmyname> so obviously no meeting next week. but the week after. at 21:00UTC
21:51:19 <mattoliverau> :P
21:51:27 <cutforth> mattoliverau: thx
21:51:31 <notmyname> cutforth: thanks for the reminder
21:51:35 <cutforth> np
21:51:44 <notmyname> last call for anything else
21:51:47 <m_kazuhiro> notmyname: I have a question.
21:52:07 <notmyname> m_kazuhiro: go ahead
21:52:25 <m_kazuhiro> notmyname: To discuss tiering, that I should do is writing it on "friday contributors-meetup ethrepad page" ?
21:52:37 <notmyname> m_kazuhiro: yes!
21:53:05 <m_kazuhiro> notmyname: thank you! :)
21:53:33 <hrou> m_kazuhiro, quick heads up we're interested in tiering as well and doing some work on it
21:53:38 <hrou> m_kazuhiro, lets get in touch !
21:54:08 <notmyname> ok, then I declare this meeting over
21:54:15 <notmyname> thanks for working on swift. see you next week
21:54:16 <notmyname> #endmeeting