19:00:01 <notmyname> #startmeeting swift
19:00:01 <openstack> Meeting started Wed Sep 10 19:00:01 2014 UTC and is due to finish in 60 minutes.  The chair is notmyname. Information about MeetBot at http://wiki.debian.org/MeetBot.
19:00:02 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
19:00:04 <openstack> The meeting name has been set to 'swift'
19:00:10 <notmyname> hello, everyone
19:00:14 <notmyname> who's here for the swift meeting?
19:00:20 * portante lurking ...
19:00:23 <acoles> o/
19:00:25 <kota_> o/
19:00:25 <cutforth> o/
19:00:26 <mattoliverau> o/
19:00:29 <hurricanerix> hi
19:00:32 <torgomatic> I'm just here for the free pizza
19:00:39 <torgomatic> though it appears I have been misled
19:01:08 <notmyname> we should do something about htat
19:01:17 <notmyname> #link https://wiki.openstack.org/wiki/Meetings/Swift
19:01:24 <notmyname> there's the agenda for this week
19:01:29 <notmyname> a few new things on it
19:01:35 <notmyname> #topic general
19:01:45 <notmyname> nothing new for general info, so this will go fast
19:02:03 <zaitcev> okay
19:02:04 <notmyname> sign up for the hackathon (don't forget flights and hotels) if you are going
19:02:06 <peluse> hi
19:02:07 <notmyname> #link http://swift-hackathon.eventbrite.com/
19:02:21 <notmyname> likewise for the Paris summit
19:02:47 <notmyname> and continue to fill out and promote the OpenStack user survey to Swift clusters that you know of
19:02:49 <notmyname> #link https://www.openstack.org/user-survey/
19:03:40 <notmyname> the user survey is the way a lot of people measure swift (externally, and especially within the openstack community), so good participation there helps us all (both socially and to help us understand use cases)
19:03:57 <notmyname> any other general things that anyone needs to bring up?
19:04:05 <notmyname> before we move on to more focused stuff
19:04:44 <notmyname> ok, moving on
19:04:59 <notmyname> #topic python-swiftclient release schedule (?!?)
19:05:28 <notmyname> so just this morning I saw on the -dev mailing list some comments about a release deadline for client projects of Sept 16
19:05:49 <torgomatic> don't we just push to pypi when we want to release the client?
19:05:52 <notmyname> this was the first I heard of it, and kinda out of the normal way client projects are released (across all openstack projects)
19:05:54 <notmyname> torgomatic: yes
19:06:07 <notmyname> torgomatic: or actually, I tag it and -infra pushes it automagically
19:06:24 <zaitcev> Well, I'm glad because it makes it easier for me to release Kodiak's fix in Rawhide.
19:06:52 <notmyname> I think the reasons are to make sure new client features don't break integration for the juno release
19:07:43 <notmyname> and the point to all this is that I'll tag a python-swiftclient release on or shortly before the 16th to make sure the latest and greatest is included
19:08:02 <notmyname> yikes. that's actually less that a week away. ok then
19:08:14 * notmyname just looked at a calendar
19:08:25 <zaitcev> There's a ton of fixes over 2.2.0, so it's all good
19:08:51 <notmyname> zaitcev: yeah, I think it seems liek a reasonable time to cut a release anyway
19:09:01 <notmyname> so know that it is coming :-)
19:09:06 <notmyname> any questions or comments on that?
19:09:24 <mattoliverau> should we make client reviews a priority then?
19:09:35 <notmyname> if you have any outstanding patch for swiftclient that should really be in the next release, let's get it on the priority reviews page
19:10:00 <notmyname> mattoliverau: AFAIK there isn't anything huge outstanding, so I'd still prefer to see progress on swift patches
19:10:18 <mattoliverau> k
19:10:19 <notmyname> but, I also just found out about this deadline so I haven't seen what exactly is outstanding
19:10:46 <torgomatic> there's only one proposed client change that passes tests, has no negative reviews, and modifies python code
19:11:09 <notmyname> that's a short list :-)
19:11:33 <notmyname> I'll look at what's open and update the priority reviews page accordingly
19:11:44 <notmyname> ok, let's move on to an ec update
19:11:47 <notmyname> #topic ec update
19:12:06 <notmyname> peluse: last week we identified a patch that needed to land that was blocking the rest
19:12:12 <notmyname> what's the status of it?
19:12:31 <peluse> yup, torogomatic +2'd it yesterdfay.  I need to tweak a few comments then will need another +2 it can land
19:12:52 <notmyname> is that https://review.openstack.org/#/c/106910/
19:13:02 <notmyname> "Early framework for the EC Reconstructor"
19:13:11 <peluse> so I'll drop a note in the channel when its ready again - yes
19:13:17 <notmyname> peluse: thanks
19:13:18 <notmyname> #link https://review.openstack.org/#/c/106910/
19:13:35 <notmyname> thanks torgomatic and mattoliverau for reviewing it :-)
19:13:38 <acoles> i'll try to look over that
19:13:44 <notmyname> acoles: great!
19:14:02 <notmyname> peluse: anything else to report for EC status this week?
19:14:08 <peluse> other than that, tsg has the fotter thing getting close I think, he went home sick so not sure he's on
19:14:14 <peluse> footer
19:14:21 <notmyname> ok
19:14:46 <acoles> peluse: is the footer going with the multipart idea?
19:15:12 <notmyname> I just had a conversation in the office today that could make use of that functionality outside of EC. something to work on after ec :-)
19:15:40 <peluse> acoles, not sure exactly what you mean but will enable a generic 'footer/trailer' to go after a payload, will post link here in a sec
19:16:30 <peluse> acoles, : https://review.openstack.org/#/c/111562/
19:16:34 <notmyname> acoles: yes, it's the multipart mime idea
19:17:13 <acoles> notmyname: ok just confirming my understanding of the comments on it so far
19:17:17 <torgomatic> acoles: it's a little harder to parse than length-value encoding, but it extends to >2 documents much more easily
19:17:21 <notmyname> good to do :-)
19:18:08 <notmyname> anything else on EC or shall we move on?
19:18:10 <torgomatic> we're only doing this because of HTTP's relatively poor support for streaming uploads, so we'd be ill-served to come up with something that doesn't stream well :)
19:18:17 <peluse> not really, slow week :(
19:19:00 <notmyname> #topic tempest response codes patch https://review.openstack.org/#/c/117193/
19:19:02 <notmyname> #link https://review.openstack.org/#/c/117193/
19:19:15 <notmyname> so a _lot_ of us were added as reviewers on that tempest patch
19:19:43 <notmyname> I'm not sure what the actual question is for this meeting, other than as of last night I got kinda frustrated with the conversation in gerrit
19:19:54 <zaitcev> the discussion seems to settle onto a common understanding
19:19:55 <notmyname> I think it would be good for other swift contributors to weigh in
19:20:22 <notmyname> zaitcev: well the last commenter still seems to want to block the patch
19:20:31 <notmyname> for those who haven't looked, here's the summary:
19:20:38 <zaitcev> but, but, but... that means re-reading the api doc to find out what it says about 2xx ;_;
19:21:31 <notmyname> the linked patch changes tempest to look at response classes instead of explicit response codes. some tempest devs seem to be against it. the swift devs (who've commented) seem to be for it
19:22:21 <notmyname> so if you've got nothing to do and a shot of whisky available, feel free to jump in the conversation! ;-)
19:22:56 <notmyname> I'll leave the rest of that for gerrit instead of this meeting
19:23:00 <notmyname> #topic priority reviews
19:23:06 <notmyname> #link https://wiki.openstack.org/wiki/Swift/PriorityReviews
19:23:26 <notmyname> several of the "priorities for juno" patches have landed this week!
19:23:35 <notmyname> acoles: the keystone acls patch got approved!
19:23:55 <acoles> yay!
19:24:02 <chmouel> cool!
19:24:09 <acoles> its like 24 hours away from merge though
19:24:10 <peluse> nice!
19:24:33 <notmyname> acoles: assuming no spurious jenkins failures :-)
19:24:34 <acoles> peluse: cschwede: thanks for reviews
19:24:41 <notmyname> I showed this list to ttx this week and he expressed concern that we'd be able to get all these landed in the next month in time for juno
19:24:44 <torgomatic> also, the gate's at depth 131, which is just comical
19:24:45 <acoles> and others along the way
19:25:29 <notmyname> so let's not let him think we can't get patches approved :-)
19:25:29 <zaitcev> It may get worse as we close on Juno
19:26:12 <notmyname> zaitcev: I hope not, but you're likely to be correct
19:26:24 <peluse> acoles, welcome!@
19:26:28 <zaitcev> bind_port is not a priority... okay. That's fine, we'll get it in eventually.
19:26:43 <notmyname> zaitcev: I took it off the list since it got approved
19:26:59 <notmyname> it was on the list last week
19:27:13 <notmyname> the corresponding triple-o patch landed too
19:27:19 <zaitcev> oh, good
19:28:08 <notmyname> acoles: looks like you're waiting for a response from gvernik about the migration patch
19:28:24 <acoles> notmyname: yes, thats been quiet since last week
19:28:31 <gvernik> i am about to submit a patch...just improving the unittests
19:28:42 <notmyname> gvernik: great!
19:28:44 <acoles> i think its close
19:28:47 <tdasilva> nice
19:29:13 <acoles> gvernik: hi, will look out for that
19:29:29 <gvernik> acoles: thanks
19:29:49 <notmyname> oh I want to add the new patch for policy descriptions from acoles to the priority list. fairly simple, but I think it's good for the user eperience of storage policies that people have if they first see SP in a juno release
19:30:06 <notmyname> #link https://review.openstack.org/#/c/120424/
19:30:29 <notmyname> done
19:31:10 <notmyname> are there any specific patches or other things you'd like to discuss here in the meeting this week?
19:32:11 <mjseger> real quick 'activity' for the hackathon - pancake breakfast at parker's barn?
19:32:13 <zaitcev> nothing comes to mind
19:32:20 <mjseger> http://www.parkersmaplebarn.com/
19:32:34 <acoles> notmyname: donagh's keen for feedback on his spec for composite tokens
19:32:37 <acoles> https://review.openstack.org/#/c/105228/
19:32:41 * notmyname likes pancakes
19:32:43 <zaitcev> I won't have a car
19:32:44 <tdasilva> just a quick reminder about the hackathon, if you are going and would like a t-shirt, email me the size
19:32:51 <zaitcev> Pancakes are pure carbs.
19:33:16 <zaitcev> I love japanese style crepes though. In secret.
19:33:24 <mjseger> pancakes are bad for you BUT with each order you get your own little pitcher of pure maple syrup
19:33:27 <notmyname> mattoliverau: there's a mailing list thread about auto-abandoning patches. I didn't want to volunteer you, but if you want to talk about your script, that's the place :-0)
19:33:51 <mjseger> thougth it might be nice excuse to go leaf peeping too, something new england is famous for that time of year, but I digress
19:33:54 <notmyname> acoles: thanks for bringk that up
19:33:54 <mattoliverau> notmyname: thanks :) I'll give it a read
19:34:26 <notmyname> I'll look over that service token spec document this week
19:34:36 <mattoliverau> I remember how agaist it they were at the mid-cycle... sigh
19:34:53 <notmyname> as always, the -spec documents are good things to review, just to we can keep feedback going
19:35:07 <notmyname> mattoliverau: yeah, but 134 deep gate queue now
19:35:25 <notmyname> not that abandoning patches directly affects that, but I think they are still related
19:35:48 <torgomatic> if you push it over 150, I think you win a plush teddy bear
19:35:54 <notmyname> wow, it seriously is 23 hours deep right now
19:36:09 <notmyname> #topic open discussion
19:36:17 <notmyname> (we kinda were already there)
19:36:29 <notmyname> anything else to bring up in the meeting this week?
19:36:32 <dfg> oh- anybody look at the error limiter thing at all?
19:36:45 <notmyname> dfg: got a link handy?
19:36:47 <dfg> just curious
19:37:00 <dfg> https://review.openstack.org/#/c/112424/
19:37:13 <dfg> i should have put the link to that middleware in the comments
19:37:43 <dfg> ok just did that
19:37:53 <notmyname> seems clayg really likes it :-)
19:38:12 <goodes> 2cents - requiring people to create an openstack identity in order to fill in the survey will probably discourage 80% of people within the first 20 seconds
19:38:29 <dfg> i was talking to darrel bishop about that too. but don
19:38:30 <notmyname> goodes: ya, I'm not a big fan of that either
19:38:33 <dfg> 't know his nick
19:38:45 <notmyname> dfg: swifterdarrell
19:39:00 <dfg> oh ok
19:39:29 <dfg> ok. well nm then. just wondering if anybody looked.
19:39:44 <notmyname> dfg: so the basic idea is to use a local sqlite db instead of memcache for tracking error-limited?
19:40:04 <dfg> ya
19:40:46 <notmyname> wait. nm. error-limited wasn't in memcached. it was in local proxy server memory
19:40:54 <notmyname> so what's the sqlite db give you?
19:41:02 * notmyname clearly hasn't looked at it :-/
19:41:07 <dfg> all the workers share the data
19:41:18 <notmyname> ah, yes. that would be good
19:41:25 <dfg> if its per worker then you don't get enough data imo
19:43:08 <notmyname> cool. I'll try to take a look this week. I hope others can too
19:43:31 <mattoliverau> I'll give it a review when I wake up :)
19:43:32 <notmyname> dfg: thanks for bringing it up
19:43:49 <notmyname> anything else from anyone, or shall we adjourn early?
19:44:39 <notmyname> ok, let's end early :-)
19:44:42 <notmyname> thanks for coming!
19:44:44 <notmyname> #nedmeeting
19:44:49 <notmyname> #endmeeting