19:00:29 <notmyname> #startmeeting swift
19:00:30 <openstack> Meeting started Wed May 21 19:00:29 2014 UTC and is due to finish in 60 minutes.  The chair is notmyname. Information about MeetBot at http://wiki.debian.org/MeetBot.
19:00:31 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
19:00:32 <otoolee> Hello.
19:00:34 <openstack> The meeting name has been set to 'swift'
19:00:40 <notmyname> who's here?
19:00:44 <portante> o/
19:00:46 <elambert> me
19:00:51 <cschwede_> \o/
19:00:55 <otoolee> Me.
19:01:11 <creiht> |-o-|
19:01:24 <acoles> i am
19:01:26 * portante goes and gets his x-wing fighter!
19:01:35 <notmyname> :-)
19:01:53 <creiht> :>o<:
19:02:00 <notmyname> I think most (all?) of you were at the summit last week
19:02:13 <notmyname> mostly I want to cover that. but a couple of other things too
19:02:21 <notmyname> #link https://wiki.openstack.org/wiki/Meetings/Swift
19:02:33 <notmyname> #topic summit follow-up
19:02:51 <notmyname> what worked at the summit? what didn't? what do you want to change for next time?
19:03:10 <notmyname> was the swift project pod useful?
19:03:14 <portante> yes
19:03:17 <cschwede_> yes!
19:03:18 <creiht> yes
19:03:18 <acoles> notmyname: yes, i think so
19:03:22 <notmyname> great!
19:03:30 <creiht> that was one of the greatest improvements
19:03:35 <acoles> i thought the formal sessions went well
19:03:39 <portante> more ops feedback, like to see an ops summit like we have the design summit
19:03:40 <notmyname> creiht: cool
19:04:00 <notmyname> portante: more than just the ops session we had? what about the 2 days of ops track they had?
19:04:16 <notmyname> portante: that is, tell me more about your idea
19:04:34 <portante> make it more comfortable for the ops folks to give feedback
19:04:48 <portante> it seemed like they felt they were doing things wrong at times
19:04:53 <portante> at least in a few of the sessions
19:04:56 <notmyname> hmm
19:05:13 <portante> and instead would like to see them voice more
19:05:27 <notmyname> I think the room layout could have been better for that
19:05:43 <portante> yes, true, it was pretty bad for those kind of sessions
19:05:43 <notmyname> bring more people up front. rather than a big lecture hall format
19:06:04 <portante> yes, like concentric circles from center out
19:06:06 <cschwede_> maybe a quick and easy survey before the next summit for topics ops are interested in?
19:06:21 <cschwede_> so we can ensure we talk about them?
19:06:45 <notmyname> cschwede_: portante: you're thinking about swift-specific stuff? 'cause I know tom is working on that for openstack-wide things
19:07:10 <notmyname> portante: cschwede_: part of that has to do with the openstack user survey. which I'll try to get more input on next time
19:07:12 <portante> openstack wide, but I only attended one ops session for openstack wide and one for just swift
19:07:19 <cschwede_> notmyname: yes, for swift-related stuff. i think it’s a little bit different thant for example nova-related ops
19:07:19 <notmyname> ah ok
19:08:14 <notmyname> I liked the ops session we had, and I got some good feedback from the ops-meetup track that was openstack-wide
19:08:42 <notmyname> things that I got from our ops session are:
19:08:44 <notmyname> ring validator (in swift-recon): eg test that something is running on the ports in the ring
19:08:44 <notmyname> deep health check to test all the way to drives (? or at least storage servers)
19:08:44 <notmyname> better ring deployment inside of swift itself
19:08:46 <notmyname> need internal net sec enforcement
19:09:32 <cschwede_> +2 for better/easier ring deployment!
19:10:08 <notmyname> These are things we should put into LP (or the new specs repo) to track and try to get done. or decide we won't do
19:10:19 <notmyname> any other ops-related comments/feedback from the summit?
19:10:36 <cschwede_> benchmarking?
19:10:52 <cschwede_> i mean how to benchmark and some example benchmarks? -> docs
19:11:16 <notmyname> cschwede_: always a popular topic. we need to move from "here's a new benchmark tool" to "here's the patch that speeds stuff up" :-)
19:11:16 <cschwede_> that said, using tools like getput.py and collectl
19:11:56 <cschwede_> agreed, no new tool, but (better) docs and examples what to benchmark and things to watch for
19:12:02 <notmyname> ya
19:12:06 <notmyname> speaking of...
19:12:12 <notmyname> I think the docs session went ok
19:12:15 <acoles> cschwede_: i think mseger is putting a link to getput into the docs
19:12:18 <notmyname> (creiht)
19:12:21 <notmyname> acoles: already done
19:12:30 <acoles> notmyname: great
19:12:30 <creiht> notmyname: yeah I have a lot on my todo list :)
19:13:00 <notmyname> creiht: my list:
19:13:03 <notmyname> refactor swift.openstack.org
19:13:03 <notmyname> use 3 sections: app devs, deployers/ops, dev contributors
19:13:03 <notmyname> make version docs apply to clear
19:13:05 <notmyname> multinode install: https://review.openstack.org/#/c/93788/
19:13:07 <notmyname> SAIO: add link to vagrant SAIO, different instructions for different distress
19:13:09 <notmyname> undocumented areas: debugging tools, operating tools collection
19:13:11 <notmyname> look in to oslosphinx
19:13:13 <notmyname> idea: use the "file a bug for the docs on this page" js that the docs team has
19:13:40 <notmyname> there was a lot that came up asking for more and better docs I think
19:13:49 <creiht> yeah
19:14:32 <notmyname> in general, there is a big need for education at a higher level too about what object storage is, where it's used, and why you should care
19:14:48 <portante> yes
19:14:55 <portante> and what it's benefits are
19:15:01 <cschwede_> yes, and especially benefits of swift
19:15:05 <notmyname> and I've had a few conversations with different companies about that. so I hope we'll, as a community, make good progress there in the next few months
19:15:10 <notmyname> cschwede_: naturally :-)
19:15:10 <cschwede_> it’s not just another object storage
19:15:56 <notmyname> cschwede_: /topic set in -swift ;-)
19:16:13 <portante> nice!
19:16:19 <creiht> we should have called swift "store all the things"
19:16:33 <portante> so be careful of what you say in this channel!
19:16:38 <cschwede_> notmyname: hehe, cool!
19:16:38 <creiht> heh
19:16:42 <notmyname> lol
19:16:44 <notmyname> so far this week we've made progress on getting a -specs repo set up
19:17:21 <notmyname> I'll check on what needs to happen to get that finished
19:17:33 <notmyname> other follow-ups include what we decided for py3
19:17:50 <notmyname> chmouel will be working on getting a py3pep8 jobs set up as non-voting
19:17:53 <notmyname> for swift
19:18:27 <notmyname> once that passes, we'll fold it in to the existing linter job so it's gating and we don't revert syntax issues
19:18:35 <notmyname> and then we just hold our breath for eventlet
19:19:04 <notmyname> ok, helping new contributors
19:19:22 <notmyname> creiht you said you'd be able to tackle an improved doc for new contributors
19:19:25 <notmyname> still up for that?
19:19:30 <creiht> yeah
19:19:37 <notmyname> great
19:19:52 <elambert> as a new contributor (or potential new contributor) i'd like to help with that too
19:20:00 <creiht> cool
19:20:00 <notmyname> elambert: great! thanks
19:20:26 <notmyname> one other thing we talked about is to (in an informal way) is to make use of a "nit" tag to be clear where a comment is "this would be nice but doesn't matter too much"
19:20:39 <notmyname> so that our comments are clearer to people who haven't been contributing long
19:21:11 <portante> did we settle on mentors or sponsors or something?
19:21:13 <notmyname> and also one other thing there...
19:21:15 <notmyname> core sponsors
19:21:19 <notmyname> portante: ya, that
19:21:20 <notmyname> :-)
19:21:35 <creiht> yeah I was thinking maybe it would make sense to integrate that into the specs stuff
19:21:36 <notmyname> I think we were all interested in the idea
19:21:48 <notmyname> creiht: sponsors?
19:21:49 <creiht> so each spec would have a sponsor which would have related reviews
19:21:51 <creiht> yeah
19:22:09 <creiht> might be easier to track that way instead of gerrit
19:22:10 <notmyname> I think that's a great idea to try
19:22:18 <creiht> but that leaves out the usual bug fixes
19:22:19 <notmyname> I talked to the -infra team about it too
19:22:21 <acoles> the specs themselves have to be +2'd, right?
19:22:35 <notmyname> acoles: ya. well, the specs have the same swift-core group as the code
19:22:35 <creiht> but maybe those don't need a sponsor?
19:23:14 <notmyname> so gerrit can add a tag that shows up for a "core sponsor" thing. it's not great (and hard to describe via text). but it's something to consider
19:23:31 <creiht> k
19:23:34 <notmyname> my first idea was to keep a wiki page and start assigning open reviews to different cores
19:23:52 <creiht> and I don't think all reviews need sponsors
19:24:13 <notmyname> creiht: most likely not. I definitely thing all reviews do not need a spec
19:24:17 <notmyname> *think
19:24:19 <creiht> right
19:24:41 <acoles> i was wondering whether +2'ing a spec implies (moral) obligation to review subsequent code?
19:24:44 <portante> sponsors are helpful for keeping complex things on track
19:24:55 <notmyname> but we should encourage people to use specs as much as possible
19:25:04 <notmyname> acoles: I don't know yet. we haven't tried it yet :-)
19:25:23 <notmyname> acoles: I think that's a reasonable start, actually
19:25:45 <acoles> notmyname: might slow down approval rate of specs :)
19:25:53 <creiht> haha
19:25:54 <notmyname> acoles: heh. good point
19:26:06 <notmyname> never mind. that's a horrible idea!
19:26:09 <notmyname> :-)
19:26:34 <creiht> yeah I don't think every spec requires a sponsor either
19:26:41 <notmyname> so the takeaway here, I think, is that I'll set up the wiki and start dividing up existing open patches among existing cores
19:26:54 <notmyname> and that's something we can track in the weekly meetings
19:26:58 <notmyname> sound ok?
19:27:32 <portante> yes
19:27:43 <acoles> notmyname: lets try it and see
19:28:40 <notmyname> ok. I'll do that (number 3-ish on my list, after -specs and storage policies merge start)
19:28:42 <creiht> notmyname: might want to try voluntary first
19:28:56 <notmyname> creiht: no! you get zaitcev's PBE patch! ;-)
19:29:10 <creiht> and figure out how to make it easier for someone to request a sponsor to help
19:29:12 <notmyname> creiht: ya, voluntary start is good and then divide up what's left
19:29:52 <notmyname> ok, that sounds like a good plan
19:30:12 <notmyname> anything more on summit follow-up before we move to the storage policy merge plan?
19:30:52 <notmyname> hey! it's dfg!
19:31:07 <dfg> hey!
19:31:20 <notmyname> #topic storage policies merge plan
19:31:30 <notmyname> it's actually just about almost finally here!
19:31:48 <notmyname> we talked last week about how we are going to get storage policies on to master
19:31:55 <notmyname> here's the rough draft: https://gist.githubusercontent.com/notmyname/7521817bd1027adc35a7/raw/609164665ec6c9ccdb0ee90a69f045df4081ca0a/gistfile1.txt
19:32:53 <notmyname> the important point is that by the end of this week, clayg should be able to propose the patch chain to master, feature/ec will not accept new stuff, and master will be in a freeze
19:33:16 <notmyname> then we all focus on reviewing the storage policy patches, and set them up to merge quickly one after the other
19:33:17 <cschwede_> notmyname: are you going to send this to openstack-dev?
19:33:25 <notmyname> cschwede_: yes, and -operators
19:33:31 <cschwede_> ok, thanks
19:33:39 <notmyname> that's my rough draft. probably add a little to it
19:33:49 <notmyname> are we still all in agreement about this plan?
19:34:50 <cschwede_> sounds good to me
19:34:55 <notmyname> (If you don't say anything, I'm assuming you agree)
19:35:42 <portante> I agree with that plan
19:35:53 <notmyname> I think there are a couple of patches on feature/ec (like the docs one) that need to be reviewed and merged to feature/ec
19:36:02 <notmyname> today or tomorrow
19:36:04 <portante> yes
19:36:13 <notmyname> now's the time to do that
19:36:36 <notmyname> any other questions about what's happening around storage policies and how it's landing on master?
19:36:56 <notmyname> naturally, I'd expect a formal Swift release after it lands
19:37:45 <elambert> whats the procedure for testing the merge?
19:37:46 <notmyname> ok, let
19:38:06 * elambert apologizes for noob question
19:38:13 <notmyname> elambert: same as normal testing of patches. pull it in to your dev or lab and run tests on it :-)
19:38:27 <elambert> :-)
19:38:29 <notmyname> elambert: we've got a lot of built-in tests. and there are others that some deployers have too
19:38:41 <notmyname> and I hope they'll run those too
19:38:50 <elambert> ok, so it will be run in anger before merge?
19:39:06 <notmyname> I like that :-)
19:39:10 <creiht> hah
19:39:44 <notmyname> ok, let's move on. if there are other questions, feel free to ask me
19:39:51 * elambert nods
19:40:04 <notmyname> I have "Parallel object auditor patch: what else needs to be done?" on the agenda, but I don't know why I added that....
19:40:06 <notmyname> #topic Parallel object auditor patch: what else needs to be done?
19:40:13 <notmyname> anyone?
19:40:14 <otoolee> I added that :)
19:40:27 <otoolee> The patch has been around for a while.
19:40:28 <notmyname> otoolee: ah! what's up? what needs to be done
19:40:36 <notmyname> otoolee: what's the link to the patch?
19:40:57 <otoolee> I just wanted to know if anyone needs anything else done to or with it?
19:41:23 <notmyname> #link https://review.openstack.org/#/c/59778/
19:41:55 <otoolee> Yes.
19:42:08 <otoolee> Sorry, I was logging in and you beat me to it.
19:42:11 <notmyname> creiht: you had some comments a while back on this. would you be able to take another look?
19:42:24 <creiht> well I'm still not fond of the idea :)
19:42:27 <notmyname> heh
19:42:31 <creiht> but as I told otoolee I'm just one dev
19:42:40 <creiht> if others like the idea, by all means :)
19:42:56 <otoolee> Thanks creiht :)
19:43:24 <notmyname> otoolee: so, realistically, unless it gets some attention later this week, nothing will happen to it until after storage policies are merged
19:43:34 <otoolee> That's cool.
19:43:47 <notmyname> looks like torgomatic was commenting on it too
19:43:49 <otoolee> I just wanted to bring it to people's attention.
19:43:52 <notmyname> otoolee: thanks
19:44:01 <notmyname> if only there were a core sponsor for it.... ;-)
19:44:09 <otoolee> I think that torgomatic was fine with it ....
19:44:11 <otoolee> :)
19:44:14 <creiht> sounds like notmyname is volunteering :)
19:44:33 <notmyname> creiht: you only volunteer when you say something is easy :-)
19:44:51 <creiht> lol
19:45:09 <notmyname> so actually that reminds me of one other thing related to patches
19:45:24 <notmyname> creiht said that the priority reviews page has been helpful in the past
19:45:37 <notmyname> I'll get back to updating that
19:45:48 <notmyname> the summit etc kinda got me derailed there :-)
19:46:17 <notmyname> #topic open discussion
19:46:24 <notmyname> anything else on your mind?
19:47:27 <notmyname> note the keystone token size mailing list thread
19:47:37 <notmyname> interesting discussion there to track
19:48:23 <notmyname> last call...
19:48:59 <notmyname> thanks for coming!
19:49:00 <notmyname> #endmeeting