21:00:11 <notmyname> #startmeeting swift
21:00:15 <openstack> Meeting started Wed Mar 23 21:00:11 2016 UTC and is due to finish in 60 minutes.  The chair is notmyname. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:00:17 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
21:00:20 <openstack> The meeting name has been set to 'swift'
21:00:28 <notmyname> who's here for the swift meeting?
21:00:36 <gmmaha> o/
21:00:38 <dmorita> o/
21:00:39 <awelleck> hey
21:00:39 <hosanai> o/
21:00:41 <cutforth> hello from snowy Colorado
21:00:41 <takashi> o/
21:00:44 <cschwede> yay! o/
21:00:59 <kota_> hello
21:01:10 <jrichli> hi
21:01:12 <timburke> o/
21:01:17 <tdasilva> hello
21:01:21 <jlhinson_> o/
21:01:43 <notmyname> welcome, everyone
21:01:52 <acoles> hello
21:02:05 <mmotiani> \o/
21:02:05 <ntata> o/
21:02:11 <pdardeau> o/
21:02:13 <minwoob> o/
21:02:21 <notmyname> more people keep joining :-)
21:02:36 <siva_krishnan> o/
21:02:58 <torgomatic> .
21:03:05 <timburke> sign of a healthy, growing community, right?
21:03:29 <notmyname> yup :-)
21:03:45 <notmyname> ok, let's get started
21:03:56 <notmyname> agenda for this week is
21:03:57 <notmyname> #link https://wiki.openstack.org/wiki/Meetings/Swift
21:04:08 <notmyname> #topic new swift core team members
21:04:30 <notmyname> I'm very happy to announce that timburke and hosanai are now swift core team members
21:04:41 <jrichli> Yay!  congrats to both
21:04:42 <notmyname> timburke: hosanai: thank you for all your work on swift
21:04:47 <kota_> congrats!
21:04:48 <tdasilva> cool! congrats timburke and hosanai
21:04:49 <dmorita> Congrats!
21:04:52 <pdardeau> congrats!
21:04:55 <gmmaha> congrats timburke and hosanai
21:04:56 <cschwede> congrats Tim and Hisashi and welcome! great to have you onboard :)
21:05:07 <takashi> congrats! :-)
21:05:11 <siva_krishnan> Congrats guys!
21:05:18 <minwoob> Congrats! :-)
21:05:19 <sgundur> congrats
21:05:27 <hosanai> thanks all :-)
21:05:39 <timburke> thanks :-)
21:06:08 <notmyname> #topic The Next Release (tm)
21:06:14 <acoles> hosanai: timburke ready for the review begging?? :) welcome!
21:06:14 <tdasilva> now can somebody review copy middleware ?
21:06:16 <tdasilva> :D
21:06:18 <tdasilva> jk
21:06:20 <notmyname> lol
21:06:24 <acoles> there you go !! ^
21:06:47 <notmyname> we've got the openstack mitaka release coming up, so that means a swift release too
21:06:59 <notmyname> so far, I feel like this release is one of the smoothest in a long time
21:07:05 <notmyname> we're in really good shape
21:07:17 <notmyname> all of the important things have landed
21:07:23 <notmyname> thank you for your code and reviews
21:07:28 <notmyname> looking at what's left...
21:07:30 <notmyname> #link https://wiki.openstack.org/wiki/Swift/PriorityReviews
21:07:48 <notmyname> ...there's always things that would be nice to include
21:08:03 <notmyname> but none of the open patches should block a release as far as I can tell
21:08:12 <notmyname> do you think that's right?
21:08:15 <acoles> i just added +2 on patch 257502
21:08:15 <patchbot> acoles: https://review.openstack.org/#/c/257502/ - swift - Fix full_listing in internal_client
21:08:18 <notmyname> am I missing anything?
21:08:26 <notmyname> acoles: great
21:09:01 <cschwede> acoles: thx :D
21:09:36 <notmyname> patch 296175 needs to land of course, but normally that's the last patch in a release
21:09:36 <patchbot> notmyname: https://review.openstack.org/#/c/296175/ - swift - 2.7.0 authors and changelog updates
21:10:03 <notmyname> thanks acoles and jrichli for looking over that for me. I think the current version is pretty good
21:10:52 <notmyname> so unless something else comes up, I'd like to land the authors/changelog patch tonight, so we'll have a SHA for the release by tomorrow (all times my local timezone)
21:11:22 <notmyname> so if patch 257502 is reviewed by this afternoon, that would be great
21:11:22 <patchbot> notmyname: https://review.openstack.org/#/c/257502/ - swift - Fix full_listing in internal_client
21:12:12 <notmyname> as soon as I have the SHA I'll request the release team to do the release. and then we'll be "done" until backports come up
21:12:21 <notmyname> any questions or concerns about this release?
21:13:44 <notmyname> great :-)
21:13:53 <notmyname> thank you everyone for your work
21:14:13 <notmyname> again, while writing the changelog, I'm impressed with what's landed and excited about the new features and bugfixes available
21:14:24 <notmyname> swift is pretty great, I think, and it's because of your work
21:14:31 <acoles> notmyname: presume you will update changelog for stuff that has recently merged?
21:14:47 <acoles> auditor/rsync temp file cleanup, container sync
21:15:01 <notmyname> acoles: if you think it needs to be specifically called out, then yes I can
21:15:16 <notmyname> oh yeah, the rsync file cleanup should be called out, at least
21:15:23 <acoles> the auditor patch yes, because it introduces a new config option
21:15:27 <notmyname> not sure about the container sync one
21:15:29 <acoles> yep, that ;)
21:15:41 <notmyname> yeah, I'll do that right after this meeting
21:15:59 <acoles> agree, sync patch does not need specific mention
21:16:05 <notmyname> ok
21:16:32 <notmyname> looking ahead to after the release, keep an eye on starred patches in the gerrit dashboard and the community starred patches section of http://not.mn/swift/swift_community_dashboard.html
21:17:04 <notmyname> ok, next topic
21:17:10 <notmyname> #topic rolling upgrade test
21:17:18 <notmyname> this is something that came up yesterday
21:17:37 <notmyname> I don't have a lot to reference or written up to copy/paste, so let me see if I can succinctly describe it
21:17:53 <notmyname> the openstack TC assigns tags to projects
21:18:07 <notmyname> these tags are stuff like "has a diverse community" or "supports stable releases"
21:18:24 <notmyname> one of these tags is "supports rolling upgrades"
21:18:56 <notmyname> now we all know, mostly because we all do this in production clusters all the time, that swift supports rolling upgrades. and it always has
21:19:55 <notmyname> however, the requirements listed on the "supports rolling upgrades" test say that we only qualify for that tag if we have a gating check that passes tests against a configuration that is partly upgraded or in the middle of a rolling upgrade
21:20:19 <notmyname> so since swift doesn't have one like this in the gate, the TC is going to strip this tag from up
21:20:29 <notmyname> so why does this matter?
21:20:54 <notmyname> well, these tags are published as "maturity" measures for people learning about openstack projects
21:21:20 <notmyname> and of course there's a big release coming, and there's a ton of marketing around it. so there will be more eyes on it
21:22:00 <notmyname> so anyone who comes and looks will see that swift has one less point for it's maturity score, and that's bad for everyone, I think (in addition to just being wrong)
21:22:23 <notmyname> so here's the problem: how do we add a gate job to test this to satisfy the letter of the law?
21:22:36 <notmyname> make sense? any ideas?
21:23:24 <acoles> notmyname: do they specify which tests must pass? functional or tempest?
21:23:44 <notmyname> the tag definition is at http://governance.openstack.org/reference/tags/assert_supports-rolling-upgrade.html
21:24:09 <notmyname> specifically, see the 4th bullet point under requirements
21:24:13 <tdasilva> so how do other teams do it? is it literately having a couple of nodes running older code and other nodes running newer code?
21:24:31 <notmyname> tdasilva: don't know. nova seems to be the only one the tc is happy with having that tag
21:25:23 <notmyname> here's the TC resolution that nearly passed yesterday and we were given one week respite on https://review.openstack.org/#/c/292334/
21:25:23 <patchbot> notmyname: patch 292334 - governance - Remove rolling-upgrade tag from swift/ceilometer
21:25:42 <cschwede> notmyname: when does this need to be done? i mean adding a rolling upgrade test?
21:25:52 <cschwede> ah, so one week?
21:26:15 <tdasilva> interesting how there are so many +1s there, but nobody there works on swift????
21:26:29 <cschwede> can we extend that to two weeks, if we have a proposal in one week?
21:26:38 <notmyname> cschwede: I dont' know. maybe
21:26:45 <notmyname> cschwede: it certainly wouldn't hurt :-)
21:27:04 <cschwede> notmyname: i’ll think about this, and get in touch with you tomorrow
21:27:08 <notmyname> thanks
21:27:15 <cschwede> maybe we can jump on a short conf call then
21:27:17 <notmyname> I talked briefly in the office with timburke about it
21:27:22 <notmyname> cschwede: yeah, that woudl be good
21:27:40 <takashi> I heard that nova has some tests(unit tests?) to make sure they don't add any schema changes with drop supports for live upgrading, but I'm not so sure that keeps upgrading tag.
21:27:53 <notmyname> we thought maybe running old code int he proxy with new code on the storage nodes might be sufficient. but I dont' know how to set that up
21:28:29 <takashi> db schema changes
21:28:48 <gmmaha> notmyname: i am guessing this is something we would like to acccomplish in an SAIO environment?
21:29:24 <notmyname> takashi: I briefly looked at the gate jobs for nova, but there wasn't anyone that was named something obvious like "dsvm-test-rolling-upgrade" or anything
21:29:35 <acoles> notmyname: might be possible using virtualenvs, but I can't say for sure without experimenting
21:29:42 <notmyname> gmmaha: in the CI gate, it's all devstack
21:29:43 <cschwede> notmyname: run functests on a saio using HEAD~1, upgrade to HEAD, run functests again?
21:30:00 <notmyname> that's an upgrade test, not a rolling upgrade test
21:30:03 <gmmaha> notmyname: aah ok. thanks
21:30:06 <torgomatic> head to the Winchester, have a pint, and wait for this all to blow over?
21:30:21 <notmyname> seems like we'd need 2 versions of the code at the same time
21:30:24 <cschwede> notmyname: well, sure, with in-between steps then
21:30:26 <acoles> where is this Winchester place?
21:31:22 <notmyname> as part of this, I did discover that with a very small change, I could take a swift sample config from september 2011 and run current code from it and pass functests :-)
21:31:22 <takashi> notmyname: I see
21:32:01 <kota_> notmyname: awesome
21:32:10 <notmyname> I thought it would be fun to say "yeah we support rolling upgrades. from 5 year old code to now. any other project do that?"
21:32:27 <kota_> lol
21:32:41 <cschwede> nice :)
21:32:47 <notmyname> maybe we could get double maturity points for that ;-)
21:32:58 <notmyname> anyway, so how do we make progress on this?
21:33:13 <notmyname> cschwede: sounded like you had an idea?
21:33:15 <acoles> notmyname: i will attempt to run different servers in different venvs tomorrow, that way we install two versions in different repo clones and venvs but set them up to talk to each other
21:33:40 <notmyname> I think proxy at versionX and a/c/o at versionY should be sufficient
21:33:41 <cschwede> notmyname: yes, but i need to think about it with a fresh mind and a coffee
21:33:48 <notmyname> cschwede: lol
21:33:52 <cschwede> so tomorrow morning (my time) ;)
21:34:52 <notmyname> cschwede: acoles: so how about I'll ping you tomorrow morning my time, and we'll check on status and have a phone call as necessary (of course anyone else is welcome too)
21:35:41 <cschwede> notmyname: sure, let’s do that, and I’ll try to ping acoles in advance, exchaning ideas
21:35:50 <cschwede> exchanging
21:35:56 <acoles> notmyname: cschwede ok
21:35:59 <notmyname> ok, great
21:36:16 <notmyname> torgomatic: can you take care of the pint at the windchester for us?
21:36:38 <torgomatic> if nothing else, I'll get that done
21:36:43 <notmyname> :-)
21:36:49 <notmyname> #topic open discussion
21:36:56 <notmyname> anything else to bring up this week in the meeting?
21:39:38 <notmyname> ok, then :-)
21:39:50 <notmyname> thanks you all for coming. good work on the release. thanks for your work
21:39:54 <notmyname> #endmeeting