19:01:02 <notmyname> #startmeeting swift
19:01:03 <openstack> Meeting started Wed Jul  2 19:01:02 2014 UTC and is due to finish in 60 minutes.  The chair is notmyname. Information about MeetBot at http://wiki.debian.org/MeetBot.
19:01:05 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
19:01:07 <openstack> The meeting name has been set to 'swift'
19:01:23 <notmyname> who's here for the swift meeting?
19:01:24 <elambert> o/
19:01:27 <peluse_> yo
19:01:27 <mattoliverau> o/
19:01:31 <cutforth> o/
19:01:31 <cschwede> o/
19:01:36 <creiht> kinda
19:01:41 <creiht> :)
19:01:46 <creiht> I have to jet early
19:01:51 <notmyname> creiht: back from vacation?
19:01:57 <creiht> yes
19:02:01 <notmyname> cool
19:02:15 <notmyname> ok, let's get started and see how long this takes :-)
19:02:23 <notmyname> #link https://wiki.openstack.org/wiki/Meetings/Swift
19:02:30 <notmyname> only 2 things on the agenda
19:02:41 <notmyname> #topic swift 2.0 RC period
19:03:00 <goodes> o/
19:03:02 <notmyname> I'm working on getting two patches backported to RC1 so we can cut an RC2
19:03:13 <notmyname> small things that shouldn't invalidate a lot of current testing
19:03:37 <notmyname> fighting with jenkins now, but I hope to have that soon
19:03:44 <notmyname> how' the rest of your testing going?
19:04:03 <notmyname> creiht: ? peluse_: ? cschwede: ?
19:04:03 <creiht> still going
19:04:06 <notmyname> great
19:04:13 <notmyname> anything (major) found?
19:04:21 <cschwede> not from my side
19:04:39 <peluse_> not yet... about an hr left of testing
19:04:46 <peluse_> focused on ssync now
19:05:17 <cschwede> i mostly tried to break things, but i wasn’t very successful, and thats a good sign ;)
19:05:18 <notmyname> we ran a comparison benchmark of 1.13.1 and 2.0 on peluse_'s avoton boxes. actually seems that 2.0 is slightly faster for some workloads.
19:05:23 <notmyname> cschwede: :-)
19:05:24 <creiht> notmyname: I'm sure you will hear from us when we run into anything major :)
19:05:34 <notmyname> creiht: I'm sure I will :-) (and that's good)
19:05:47 <peluse_> notmyname:  that's interesting... cool
19:05:53 <notmyname> so the current plan is to get an rc2 today (please jenkins!) and then cut the final on monday morning
19:05:59 <cschwede> notmyname: interesting, can you share any numbers?
19:06:03 <peluse_> anyone else doing/done rolling upgrade testing?
19:06:29 <cschwede> peluse_: me, but only with my dev vm, and everything worked for me
19:06:59 <peluse_> cool, was curious about the comment I made in the other channel, having to restart memcached and whether that would be a normal part ofthe process anyway
19:07:05 <notmyname> cschwede: not really yet, since there were some odd things separate from swift. like centos locking the cpu frequency lower or networking issues. mostly it's a "in general, the good numbers are slightly bigger"
19:07:39 <torgomatic> oh yeah, meeting :)
19:07:50 <notmyname> peluse_: most prod clusters won't want to flush memcache because it would generally kill clusters (internal and external to swift)
19:07:59 <creiht> haha
19:08:02 <notmyname> peluse_: but it's certainly an ops sort of thing that has to be accounted for
19:08:06 <creiht> yeah we generally try to not do that :)
19:08:23 <notmyname> creiht: make auth cry ;-)
19:08:32 <peluse_> notmyname: OK, I suppose a bunch of errors for things that don't exist in info (storage policy field in container) willeventually go away no their own then?
19:08:40 <notmyname> any question on 2.0 or uncertainty about plans?
19:09:07 <notmyname> peluse_: right, those should be handled as replication gets to them and updates the schema
19:09:43 <notmyname> ok, if no questions....moving on
19:09:47 <peluse_> not the DB, the cached info in memchached that you get from container_info()
19:10:09 <notmyname> peluse_: ah, yes. actually one of clay's patches addresses that
19:10:18 <peluse_> oh, cool
19:10:26 <notmyname> (i think)
19:10:37 <cschwede> peluse_: hmm, need to check that memcache thing, might be restarted in my vm together with swift services
19:10:38 <peluse_> i feel "validated" somehow :)
19:10:50 <notmyname> peluse_: #topic gerrit review dashboard
19:10:53 <notmyname> err
19:10:57 <notmyname> peluse_: https://review.openstack.org/#/c/102747/
19:11:04 * peluse_ looks
19:11:42 <peluse_> OK, I'll try it in my upgrade test w/o restart memcachd, thanks
19:11:49 <notmyname> may be related to what you are seeing. if not, please follow it up :-)
19:12:06 <peluse_> will do
19:12:06 <cschwede> wow, one line of code change, +302 of tests. nice :)
19:12:15 <notmyname> #topic gerrit review dashboard
19:12:20 <notmyname> so speaking of patches...
19:12:25 <notmyname> new gerrti has more tools
19:12:32 <notmyname> so I made http://bit.ly/1iVBigF
19:12:35 <notmyname> #link http://bit.ly/1iVBigF
19:12:47 <notmyname> (go on. click it. I dare you)
19:12:56 <cschwede> this is great, like it!
19:13:03 <peluse_> ditto
19:13:04 <notmyname> I hope it helps keep patches moving
19:13:07 <mattoliverau> The daskboard is awesome and I like the new version that shows your reviews... nice work.
19:13:48 <portante> +1
19:13:59 <notmyname> sections are specs, your stuff, stuff needing approval (final +2), errored, old, open, with negative feedback, WIP, and finally EC
19:14:12 <clayg> ohai
19:14:18 <mattoliverau> I wonder if a section of patches you haven't reviewed is in order.. I have a gerrit search for that.
19:14:24 <notmyname> unless you're specifically trying to get EC moving forward, you should be able to start at the top and move down, I think
19:14:55 <notmyname> the tool I used to build the long URL is https://github.com/stackforge/gerrit-dash-creator
19:15:08 <notmyname> and I have a patch in gerrit to add the swift dashboard
19:15:19 <notmyname> https://review.openstack.org/#/c/103683/
19:15:37 <notmyname> so you can look at that, add too it, fix it, and share it with everyone! :-)
19:16:29 <notmyname> #topic open discussion
19:16:36 <notmyname> what else is on your mind?
19:17:21 <peluse_> 4 day work week I guess...
19:17:22 <notmyname> I'm working on the 2.0 release stuff this week and I've got the "core sponsor" idea that we talked about in atlanta on my TODO list
19:17:23 <mattoliverau> I think everyone is looking at the dash :P
19:17:27 <notmyname> :-)
19:17:58 <notmyname> torgomatic has some interesting performance stuff to take a look at about removing kernel->user memcopies
19:18:12 <notmyname> https://wiki.openstack.org/wiki/Swift/PriorityReviews is more up to date now
19:18:14 <torgomatic> related to that, does anyone know of folks running Swift on non-Linux systems?
19:18:29 <portante> torgomatic: production only?
19:19:04 <torgomatic> portante: or staging, I guess... basically I'm using Linux-specific syscalls to get zero-copy data movement, and of course that won't work on FreeBSD or whatever
19:19:35 <portante> but does your patch gracefully handle a non-linux box?
19:19:38 * portante thought it doeds
19:19:39 <portante> does
19:19:47 <torgomatic> I think it does ;)
19:21:15 <torgomatic> I'm not overly worried about it, just curious about what's out there
19:21:35 <notmyname> anything else from anyone?
19:21:37 <portante> perhaps worth a mailing list query?
19:22:03 <torgomatic> possibly; once the work gets closer to done I'll ask the ML
19:22:55 <clayg> notmyname: annegentle had asked a while ago about some undocumented container sync config options
19:23:20 <clayg> notmyname: I found recently that run_pause or whatever in the object-replicator is also undoc'd
19:23:40 <clayg> notmyname: as PTL, what's your plan for getting us to make this better?
19:23:51 <clayg> as in, can you make mattoliverau or something do it?
19:23:56 <notmyname> lol
19:24:04 <torgomatic> I looked at a couple of those, and they had help text on them, so I think the doc team's parser just wasn't picking them up
19:24:08 <notmyname> it would be good to audit code->sample->docs to ensure that the right thing is in the right place :-)
19:24:09 <torgomatic> not an exhaustive search though
19:24:30 <clayg> notmyname: maybe we should patch oslo.config to support conf and get on that train?
19:24:33 <mattoliverau> lol, i can take a look if you all want :)
19:24:36 <clayg> support conf.d
19:24:40 <notmyname> mattoliverau: thanks!!
19:25:06 <dhellmann> clayg: oslo.config already supports config directories
19:25:10 <clayg> mattoliverau: well i'm not even sure what I'm asking you to do - torgomatic says it's already as good as it can possibly be
19:25:27 <clayg> dhellmann: oh well I think that was my last hold out!?  maybe we should look again!
19:25:32 <torgomatic> yeah, someone else is looking at it - that's as good as it gets! ;)
19:25:33 <notmyname> clayg: are you proposing getting a lot of swift core devs to add an oslo config dependency? ;-)
19:25:49 <mattoliverau> I can at least looks as I want to work my way through the code anyway :)
19:26:02 <clayg> if it's *better* - I thought we liked dependencies that added value?  is it crappy?
19:26:14 <notmyname> I don't know. new == scary, right? ;-)
19:26:46 <clayg> hrmm.... I think unknown == scary, I apparenlty don't know as much about oslo.config as I thought I used to?
19:26:56 <notmyname> mattoliverau: let me know how I can help guide you for looking at the config options
19:27:16 <mattoliverau> notmyname: k, will do :)
19:27:45 <notmyname> clayg: in general, that's fine. I don't think it's been such a pain point that people have prioritized it. if you look in to it and say it's awesome, I'll believe you
19:27:59 <clayg> notmyname: well see
19:28:57 <clayg> dhellmann: oh... the conf.d support that was tricky was paste.ini configs - psate.loadapp doesn't support the conf.d directive by default
19:29:18 <dhellmann> clayg: ok, that's not oslo.config then
19:29:54 <clayg> dhellmann: oh YEAH?  then how do you document the config options that go in the middleware config sections?
19:30:49 <dhellmann> clayg: I'm not sure exactly what you mean. Are those options passed through paste to the WSGI app?
19:31:36 <clayg> dhellmann: they are
19:32:02 <dhellmann> clayg: how are those being documented now? how are they defined?
19:32:41 <notmyname> does this need to move to #openstack-dev or #openstack-swift?
19:33:08 <notmyname> figure out what the answers are, what the possibilites are, and come back with the answers?
19:33:55 <clayg> notmyname: k
19:34:32 <notmyname> thanks last call for any other topics....
19:35:09 <notmyname> ok
19:35:13 <notmyname> thanks everyone for coming
19:35:16 <peluse_> later on, thanks
19:35:17 <notmyname> see you online...
19:35:20 <notmyname> #endmeeting