21:00:25 <notmyname> #startmeeting swift
21:00:26 <openstack> Meeting started Wed Mar 16 21:00:25 2016 UTC and is due to finish in 60 minutes.  The chair is notmyname. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:00:27 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
21:00:29 <openstack> The meeting name has been set to 'swift'
21:00:40 <notmyname> who's here for the swift team meeting?
21:00:42 <onovy> Hi
21:00:47 <dmorita> o/
21:00:47 <Zyric> Hi
21:00:50 <kota_> hello
21:00:50 <cschwede> o/
21:00:50 <ho_away> o/
21:00:51 <minwoob> o/
21:00:52 <jrichli> hey
21:00:52 <timburke> o/
21:00:53 <tdasilva> eu
21:00:53 <pdardeau> o/
21:00:53 <siva_krishnan> o/
21:00:54 <sgundur> hi
21:00:56 <awelleck> o/
21:00:58 <hurricanerix> o/
21:00:58 <gmmaha> o/
21:01:28 <mattoliverau> o/
21:01:35 <notmyname> welcome, everyone
21:01:45 <notmyname> acoles_ said he'd be a few minutes late
21:02:24 <notmyname> agenda this week is short. there's only one big topic to go over
21:02:26 <torgomatic> .
21:02:26 <notmyname> #link https://wiki.openstack.org/wiki/Meetings/Swift
21:02:42 <notmyname> release planning for the mitaka release
21:02:57 <notmyname> clayg: courtesy ping
21:03:04 <acoles> here, sorry i am late
21:03:07 <clayg> oh, short adgenda today - i like it
21:03:09 <notmyname> so let's get started
21:03:21 <notmyname> clayg: might take awhile though :-)
21:03:27 <clayg> boooo
21:03:29 <notmyname> #topic mitaka release
21:03:51 <notmyname> we're coming up on the mitaka openstack release, so we need to have a swift release
21:04:19 <notmyname> our hard deadline, from a community perspective, is the end of the month (march 31/april 1)
21:04:39 <notmyname> however, I want to do a release next week by march 25
21:04:52 <notmyname> #info release date goal is march 25
21:04:57 <notmyname> that's one week from friday
21:04:57 <cschwede> an easter egg release!
21:05:01 <notmyname> wheee!
21:05:10 <notmyname> cschwede: can you send us all kinder eggs?
21:05:27 <jrichli> i might know somebody who brought some to the US ...
21:05:49 <cschwede> i’ll see what i can do ;)
21:05:59 <notmyname> for the most part, we're looking pretty good for the release. there's not a huge thing that we must land before the release
21:06:13 <notmyname> however, there are several things that *should* go in, and a few more that would be nice
21:06:23 <notmyname> I spent some time this morning putting together that list
21:06:28 <notmyname> #link https://wiki.openstack.org/wiki/Swift/PriorityReviews
21:06:43 <notmyname> I revived the old priority reviews wiki page to help over this next week
21:07:17 <notmyname> the major things at the top are the things I think we really need to land for this release
21:07:27 <notmyname> first, concurrent GETs
21:07:30 <notmyname> https://review.openstack.org/#/c/117710/
21:07:30 <patchbot> notmyname: patch 117710 - swift - Add concurrent reads option to proxy
21:07:35 <notmyname> mattoliverau: where are we here?
21:07:45 <notmyname> and acoles and clayg (you've been looking at it)
21:07:58 <clayg> not so much, but i want to look at it again, this week for sure
21:08:10 <mattoliverau> I push a new patch yesterday (my time).
21:08:29 <acoles> i didn't get tome to look at mattoliverau's new patch today but hope to tomorrow
21:08:35 <mattoliverau> acoles was getting some intermittent test failure that I'm failing to reproduce.
21:08:36 <clayg> even if someone else beats me to the +A I'll probably look at it - just cause it's important and wanna make sure i like it - everyone should look at it - concurrent gets is the bomb
21:08:41 <mattoliverau> acoles: ta
21:08:49 <notmyname> ok, also note that mattoliverau is done with swift on his friday morning this week (at least for the next few weeks)
21:08:54 <acoles> and i think clayg's follow up tests will help a lot
21:09:02 <notmyname> baby time
21:09:32 <clayg> acoles: i never could quite get happy with them - but i did want to setup something where I could run it through *all* the scenarios - maybe you can help me make what I started suck less
21:09:39 <acoles> mattoliverau: i didn't see them fail on my laptop, only on my w/station
21:09:40 <mattoliverau> yeah, I be around, but not doing any work other then not sleeping :P
21:10:08 <acoles> clayg: thats what i want to do i.e. watch how different scenarios unfold
21:10:28 <clayg> mattoliverau: you've done a great thing with concurrent gets - if somehow we can't iterate fast enough on the patch by friday acoles or i will take it over as a baby welcome gift
21:10:34 <mattoliverau> acoles: ok. maybe I'll take a look on other decives. I'll try and recreate today
21:10:36 <notmyname> :-)
21:10:52 <mattoliverau> :)
21:11:09 <acoles> mattoliverau: yep, great job, i'm sure its going to land
21:11:13 <notmyname> next up is https://review.openstack.org/#/c/279440/ which patches the auditor to not be terrible
21:11:13 <patchbot> notmyname: patch 279440 - swift - Skip already checked partitions when auditing obje...
21:11:26 <notmyname> not terrible == closes https://bugs.launchpad.net/swift/+bug/1183656
21:11:27 <openstack> Launchpad bug 1183656 in OpenStack Object Storage (swift) "object auditors don't finish" [Critical,In progress] - Assigned to Christian Schwede (cschwede)
21:11:27 <clayg> oh that sounds good - oh goodness yes we have to land that
21:11:34 <notmyname> this is cschwede's patch
21:11:45 <notmyname> tdasilva: you were looking at it this morning. any updates or progress?
21:11:53 <notmyname> cschwede: anything to raise in the meeting about it?
21:12:00 <tdasilva> not yet, got swamped with meetings this afternoon
21:12:04 <notmyname> it already have one +2 from kota_ (thanks!)
21:12:05 <cschwede> nope, hopefully it’s fine
21:12:29 <notmyname> tdasilva: will you be able to finish your review by this time tomorrow?
21:12:44 <tdasilva> there's also a +1 from briancline which I think counts a lot
21:13:12 <tdasilva> notmyname: I will definetely try to review tomorrow, can't commit that I will finish
21:13:34 <notmyname> ok. if we don't have it landed tomorrow by my lunchtime, I'll start bugging people again about it :-)
21:13:36 <notmyname> tdasilva: thanks
21:13:48 <notmyname> next up is https://review.openstack.org/#/c/238799/
21:13:48 <patchbot> notmyname: patch 238799 - swift - Change schedule priority of daemon/server in config
21:13:58 <notmyname> this is from peterlisak
21:14:09 <notmyname> it's been sitting a while (too long!) with good comments
21:14:32 <notmyname> I cleaned up the merge conflict this morning and I'm looking at it, but we need one more
21:14:54 <notmyname> we've got ops feedback that it's good. it's shown benefit with real numbers. no reason not to land it
21:15:07 <notmyname> well, no reason not to actually give it a review
21:15:18 <notmyname> can anyone else volunteer to review it?
21:15:30 <notmyname> kota_ torgomatic?
21:15:30 <clayg> notmyname: was the code for calling into the priority thing really lifted from some GPL code base?  cause it's way more than I expected (looking mainly at our super thing wrapper over other posix calls (e.g. fallocate)
21:15:51 <clayg> *thin
21:15:53 <kota_> yup
21:16:15 <clayg> kota_: yup as in it steals code form GPL project unattributed?
21:16:19 <notmyname> clayg: ah, I'm not sure. I saw some comments, but I didn't see any new dependencies, so I just assumed it was new code written for this
21:16:33 <kota_> maybe tommorow-ish I can have time to look at
21:16:43 <notmyname> onovy: are you here?
21:16:44 <clayg> I mean we could try to feign ignorence - but why are we copy-pasting that jazz anyway when we have examples in our own code base to crib from?
21:16:47 <kota_> clayg: oh, I'm too slow
21:16:58 <onovy> Yep
21:17:00 <clayg> oh, you weren't responding to me :)
21:17:17 <notmyname> clayg: it's a good point. I thought there were new dependencies in an earlier patch set, so I assumed it was something new now that there aren't new dependencies
21:17:42 <notmyname> onovy: ah, good. i want to see the ionice patch land, but clay raises the question about GPL code copied into the patch
21:17:47 <notmyname> onovy: do you have any info on that?
21:18:06 <notmyname> onovy: https://review.openstack.org/#/c/238799/
21:18:07 <patchbot> notmyname: patch 238799 - swift - Change schedule priority of daemon/server in config
21:18:08 <onovy> I think it's just consts from kernel
21:18:14 <onovy> Other code is Peter code
21:18:24 <notmyname> soecifically https://review.openstack.org/#/c/238799/22/swift/common/priority.py
21:18:24 <patchbot> notmyname: patch 238799 - swift - Change schedule priority of daemon/server in config
21:18:25 <notmyname> ?
21:18:49 <notmyname> ok
21:19:09 <onovy> It's inspired but other code, but not copy pasted
21:19:13 <notmyname> so i'll keep looking at it, and I'll re-read the historical gerrit comments to see what the issue might be
21:19:14 <notmyname> ok
21:19:23 <onovy> Ok thanks
21:19:33 <notmyname> clayg: so we're probably ok
21:20:03 <notmyname> kota_: thanks for looking at that one (tomorrow)
21:20:14 <notmyname> next up is the copy middleware patch
21:20:24 <notmyname> https://review.openstack.org/#/c/156923/
21:20:24 <patchbot> notmyname: patch 156923 - swift - Refactor server side copy as middleware
21:20:45 <notmyname> this depends on one more +2 on https://review.openstack.org/#/c/260179/
21:20:45 <patchbot> notmyname: patch 260179 - swift - decouple versioned writes from COPY
21:21:40 <notmyname> I don't want to necessarily try to slam this one into the release at the last minute, but it seems like we might be able to reasonably get it in
21:22:06 <notmyname> and it's definitely required for crypto work (which I'll get to ina  bit
21:22:26 <notmyname> please look at these if you get a chance
21:22:26 <clayg> oh goodness - we're swamped!  :'(
21:22:43 <notmyname> yeah, the other ones are definitely higher priority
21:23:26 <notmyname> ok, for smaller patches that could be nice to land in mitaka
21:23:35 <notmyname> Updated Keystone middleware options: https://review.openstack.org/#/c/293382/
21:23:35 <notmyname> Remove deprecated Keystone option: https://review.openstack.org/#/c/258158/
21:23:35 <notmyname> fallocate as a percent: https://review.openstack.org/#/c/288011/
21:23:35 <patchbot> notmyname: patch 293382 - swift - Make keystone middleware options consistent in docs
21:23:36 <patchbot> notmyname: patch 258158 - swift - Keystone middleware deprecated option is_admin rem...
21:23:37 <patchbot> notmyname: patch 288011 - swift - Allow fallocate_reserve to be a percentage
21:23:56 <notmyname> the first 2 are mostly docs changes and simply need to be verified and landed
21:25:13 <notmyname> these 3 should be pretty easy for someone to review quickly, and they all seem to improve the quality of life for deployers
21:25:36 <notmyname> I know you're not a fan of the fallocate percent one, clayg
21:26:45 <notmyname> ok, next thing
21:26:49 <notmyname> container sync
21:27:16 <notmyname> as we talked at the hackathon, there's been a lot of work in this area
21:27:37 <notmyname> from eran, these are the 3 patches that make significant progress for them with container sync
21:27:44 <notmyname> https://review.openstack.org/#/c/281814/
21:27:44 <patchbot> notmyname: patch 281814 - swift - Per container stat. report
21:27:48 <notmyname> https://review.openstack.org/#/c/210099/
21:27:49 <patchbot> notmyname: patch 210099 - swift - Add process level concurrency to container sync
21:27:53 <notmyname> https://review.openstack.org/#/c/225338/
21:27:53 <patchbot> notmyname: patch 225338 - swift - Add thread level concurrency to container sync
21:28:04 <notmyname> here's my thoughts
21:28:18 <notmyname> 1) it would be nice to have this in mitaka, but it's not a requirement
21:29:03 <notmyname> 2) we know container sync has issues, and the proposed changes do make it better. we've got some people who are dedicated to working on container sync, so if it improves things, even if not perfect, let's land these as soon as possible
21:29:22 <notmyname> clayg: acoles: does my point 2 accurately reflect what we talked about in bristol?
21:29:47 <acoles> yes
21:30:12 <notmyname> ok, good :-)
21:30:32 <clayg> right on
21:30:33 <gmmaha> notmyname: i have just started looking into container sync after talking to pdardeau. Would definitely review the patches from eran
21:30:41 <notmyname> gmmaha: great, thanks
21:30:46 <clayg> yeah that's awesome!
21:30:50 <notmyname> it's rather late for eran and oshrit right now, and it doesn't look like they're online right now
21:31:14 <jrichli> I will be sure they are aware
21:31:17 <clayg> notmyname: i'm pretty sure some changes came up like in the last 24 hours or - definately want to loop back again
21:31:32 <notmyname> jrichli: thank you
21:31:33 <acoles> notmyname: i think there is one missing from the sync list - https://review.openstack.org/#/c/270961/
21:31:34 <patchbot> acoles: patch 270961 - swift - Container-Sync to perform HEAD before PUT object o...
21:31:55 <notmyname> acoles: ah, yes
21:32:02 <acoles> oshrit updated a couple of reviews today
21:32:04 <notmyname> I'll add it to that section
21:32:06 <notmyname> good
21:32:22 <clayg> notmyname: acoles has some ec/ssync related patches too yeah?
21:32:29 <notmyname> yeah, eran told me there were several patches that had to be re-pushed today. so it looks like that happened
21:32:31 <acoles> always :)
21:33:01 <notmyname> acoles: can we come back to those in just a minute?
21:33:20 <acoles> notmyname: sure
21:33:37 <notmyname> last big group of patches I called out has to do with crypto. here's the plan we have for that...
21:33:48 <notmyname> we're *not* going to have crypto in mitaka
21:33:59 <notmyname> but we'd like to have a reasonably good feature branch by the summit
21:34:08 <notmyname> then after the summit, we'll do the integration/merge dance
21:34:27 <notmyname> ie, we'll have crypto in swift soon, just not in this release
21:34:39 <notmyname> the patches listed are from jrichli's excellent summary on this etherpad
21:34:44 <notmyname> #link https://etherpad.openstack.org/p/swift-hackathon-feb-2016-encryption
21:35:19 <notmyname> so after all the other stuff...start getting familiar with what's happening there
21:35:34 <notmyname> jrichli: acoles: is that an accurate crypto summary?
21:35:45 <jrichli> sounds good
21:35:55 <acoles> notmyname: yes
21:35:58 <notmyname> great
21:36:09 <jrichli> we will start working on docs to help with ramp-up/ review
21:36:15 <notmyname> perfect
21:36:20 <notmyname> ok, acoles, back to you with ec/ssync patches. what do we need to add to the list for mitaka?
21:37:07 <acoles> notmyname: oh, ok, well https://review.openstack.org/232684 would be nice to have and you and torgomatic reviewed once in Bristol
21:37:16 <acoles> its probably the simplest
21:37:39 <notmyname> ah, right. yes, I'll look at that again
21:37:50 <notmyname> torgomatic: will you be able to review that one again?
21:38:22 <acoles> notmyname: and this is new but hopefully fixes several bugs https://review.openstack.org/293633
21:38:30 <acoles> patch 293622
21:38:30 <patchbot> acoles: https://review.openstack.org/#/c/293622/ - openstack-health - Adjust tooltip position for homepage gauges
21:38:36 <acoles> argh, sorry
21:38:44 <acoles> patch 293633
21:38:45 <patchbot> acoles: https://review.openstack.org/#/c/293633/ - swift - Put correct Etag and Accept-Ranges in EC 304 and 4...
21:38:49 <notmyname> those tooltips though
21:39:02 <clayg> tooltips are key
21:39:05 <mattoliverau> lol
21:39:17 <tdasilva> heh
21:39:17 <clayg> maybe we should review that one...
21:39:35 <notmyname> right. I saw acoles propose that one this morning. looks pretty important
21:39:35 <mattoliverau> add it to the mitaka list :P
21:40:01 <acoles> I'd like to see patch 293633 merge if poss because it fixes a tempest test that fails with EC, which would help with our QA, if not others
21:40:01 <patchbot> acoles: https://review.openstack.org/#/c/293633/ - swift - Put correct Etag and Accept-Ranges in EC 304 and 4...
21:40:18 <clayg> meh
21:40:21 <clayg> well..
21:40:28 <clayg> i mean - if there's a tempest test
21:41:13 <kota_> if possible, one more EC related patch 282578
21:41:13 <patchbot> kota_: https://review.openstack.org/#/c/282578/ - swift - Set backend content length for fallocate - EC Policy
21:42:05 <jrichli> only needs one more +2
21:42:19 <kota_> it has already torgomatic +2 and i'm sure that works but... I'm in the co-author :/
21:42:29 <notmyname> so to summarize, here's what I see for mitaka
21:42:46 <acoles> notmyname: then there is patch 181407 and also optimistic gets patch 215276 but the realist in me says they won't make mitaka
21:42:47 <patchbot> acoles: https://review.openstack.org/#/c/181407/ - swift - EC: Avoid conflicts when ssync'ing fragments
21:42:48 <patchbot> acoles: https://review.openstack.org/#/c/215276/ - swift - Enable object server to return non-durable data
21:43:25 <notmyname> auditor bug, concurrent gets, ionice, the 304 etag one, EC fallocate, and don't ssync when only .durable is missing
21:43:46 <notmyname> I'll update the wiki page after the meeting
21:44:18 <notmyname> then, once those are landed, look at copy middleware, container sync, and the few small patches
21:44:46 <clayg> aww man ec optomistic gets was gunna be awesome :'(
21:44:53 <clayg> we should merge everything
21:44:58 <notmyname> and if you still need something to do, I'll call you a liar ;-)
21:45:07 <kota_> clayg: agree
21:45:09 <mattoliverau> lol
21:45:11 <acoles> clayg: it'll still be there for newton :)
21:45:24 <acoles> notmyname: i need something to do
21:45:31 <notmyname> liar ;-)
21:46:08 <notmyname> anything else to bring up for mitaka? (please say no. we've got a lot on our plate already)
21:46:42 <clayg> bah there's like a week an half - it'll be fine
21:46:48 <notmyname> :-)
21:46:58 <notmyname> #topic open
21:47:07 <notmyname> I don't have anything else this week. does anyone else?
21:47:38 <clayg> i think maybe we should skip the M release - swift is already pretty great - let's just sit this one out
21:47:47 <mattoliverau> lol
21:47:50 <acoles> lol
21:47:59 <notmyname> lol
21:48:17 <notmyname> yeah, but we're making it even better
21:48:18 <clayg> "what's the sha for the RC?!"  "nah, we're good"
21:48:21 <timburke> hey, we've already got an M release. we're good.
21:48:39 <clayg> timburke: oh crap did i get my letters mixed up?
21:48:52 <clayg> jokes probably less funny when I don't even know what release i'm suggesting we skip
21:48:53 <timburke> nope. just forgetting about 2.6.0
21:48:55 <mattoliverau> and watch the other project scramble around clogging up the gate.
21:49:07 <clayg> mattoliverau: yes!
21:49:09 <clayg> rofl
21:49:33 <notmyname> in case you missed it an hour or so ago, someone from hubic/ovh was talking about their 10+PB EC cluster (along with their even larger replicated storage)
21:49:40 <clayg> notmyname: ok, yeah i got nothing
21:49:47 <acoles> notmyname: i think there is no other business :P
21:49:49 <notmyname> yet another really cool use case where swift is being used today
21:50:01 <kota_> great
21:50:26 <notmyname> thanks, everyone, for helping make swift pretty great
21:50:31 <notmyname> #endmeeting