21:00:24 <notmyname> #startmeeting swift
21:00:25 <openstack> Meeting started Wed Jun 22 21:00:24 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.
21:00:30 <openstack> The meeting name has been set to 'swift'
21:00:31 <notmyname> who's here for the swift team meeting?
21:00:34 <mattoliverau> o/
21:00:35 <m_kazuhiro> o/
21:00:37 <kota_> here
21:00:37 <cschwede> o/
21:00:40 <timburke> o/
21:00:40 <dmorita> hi
21:00:41 <hosanai> o/
21:00:49 <kei_yama> o/
21:00:51 <mathiasb> o/
21:00:54 <jrichli> .
21:01:05 <acoles> here
21:01:30 <notmyname> welcome, everyone
21:01:36 <tdasilva> hi
21:01:38 <notmyname> agenda for this week is at
21:01:40 <notmyname> #link https://wiki.openstack.org/wiki/Meetings/Swift
21:02:13 <notmyname> because it's already midnight for oshritf_, we'll start with container sync first
21:02:19 <notmyname> #topic container sync patches
21:02:22 <oshritf_> :)
21:02:38 <notmyname> this week, oshritf_ brought up these patches (again)
21:02:43 <notmyname> patch 210099
21:02:43 <patchbot> notmyname: https://review.openstack.org/#/c/210099/ - swift - Add process level concurrency to container sync
21:02:49 <notmyname> patch 225338
21:02:49 <patchbot> notmyname: https://review.openstack.org/#/c/225338/ - swift - Add thread level concurrency to container sync
21:03:07 <notmyname> and also oshritf_ has some performance results!
21:03:14 <mattoliverau> cool
21:03:15 <notmyname> #link https://ibm.app.box.com/s/bf43b9vhupjg3mz2f0pik1fjj5ugd4ca
21:03:40 <notmyname> summary is basically "land the patches. it's a *lot* faster":-)
21:04:12 <notmyname> oshritf_: do these patches apply cleanly on top of the crypto work (the crypto-review branch) or are there merge conflicts?
21:04:40 <oshritf_> right, thanks John! We spent the last few weeks on performance measurements, and the results looks good - so wanted to discuss it with you guys again
21:05:34 <oshritf_> I worked on applying it on crypto branch today, pushed the first patch-set not long ago
21:05:41 <notmyname> my opinion is the same as it was in bristol: please let's land these quickly. that's slightly confounded at this very moment by the crypto merge process going on
21:06:00 <notmyname> oshritf_: were there merge conflicts?
21:06:27 <mattoliverau> oshritf_: an initial glance at the numbers, wow, that looks great, nice work!
21:06:41 <notmyname> mattoliverau: I know, right?!
21:06:56 <oshritf_> yes, I resoved them. Merged it all in one patch - processes + threads together
21:07:09 <notmyname> oshritf_: oh, interesting. ok
21:07:27 <cschwede> impressive stuff!
21:07:57 <notmyname> I don't think proposing the patch to the feature/crypto-review branch is a good idea. but I'm trying to think quickly here about a different process that might work
21:08:21 <notmyname> maybe proposing to master with merge conflicts resolved but with the depends-on tag int he commit message?
21:09:14 <notmyname> but i'm not sure if resolving the conflicts would allow it to be proposed to master if the crypto stuff isn't on master
21:10:14 <notmyname> oshritf_: the plan is to get the crypto stuff landed soon. we'll be taking up that topic later this meeting. and if you have one merged patch that has the conflicts resolved, I'd love to see that land right after crypto
21:10:15 <mattoliverau> hmm, yeah, the fact that there were things to resolve might mean waiting for crypto to land
21:10:45 <mattoliverau> +1
21:10:47 <notmyname> but I don't know if there a way to propose the resolved patch before crypto lands or not. the alternative would be to keep it local or publish it somewhere else until crypto lands
21:10:48 <acoles> oshritf_: where were the conflicts? looking at those patches I don't immediately see where they would be
21:11:20 <notmyname> hmm...I wonder if they were with crypto or with other stuff that's landed on master
21:11:43 <acoles> internal_client maybe? can't remember if we touched that
21:11:45 <oshritf_> I wasnt expecting any conflicts, I wonder if I applied the right git commands
21:12:29 <acoles> if we trust gerrit it does not report any conflicts (top right corner) but not sure if it is checking for conflicts across feature branches
21:12:37 <oshritf_> only sync.py and test container sync prob
21:13:06 <oshritf_> no conflicts with other patches
21:13:19 <notmyname> ugh. hope I didn't miss anything. I just managed to kick my computer plug out of the wall with my feet
21:13:28 <notmyname> crypto isn't touching those, is it?
21:14:19 <acoles> looking...
21:14:33 <notmyname> not according to `git diff --name-only master | grep sync`
21:15:22 <oshritf_> https://review.openstack.org/#/c/332985/
21:15:22 <patchbot> oshritf_: patch 332985 - swift (feature/crypto-review) - Add thread level concurrency to container sync
21:15:24 <notmyname> https://review.openstack.org/#/c/332985/ is the one you proposed earlier today to crypto-review
21:15:24 <patchbot> notmyname: patch 332985 - swift (feature/crypto-review) - Add thread level concurrency to container sync
21:15:37 <notmyname> :-)
21:15:59 <notmyname> oshritf_: can you propose that to master tomorrow? (instead of crypto-review)
21:16:11 <acoles> notmyname: git diff --stat master doesnt show us touching those sync files
21:16:36 <oshritf_> it's on master https://review.openstack.org/#/c/225338/
21:16:36 <patchbot> oshritf_: patch 225338 - swift - Add thread level concurrency to container sync
21:16:53 <notmyname> oshritf_: oh, is that one combined with the process concurrency on?
21:16:55 <notmyname> *one
21:17:23 <oshritf_> not combined on master
21:17:31 <notmyname> ok
21:18:19 <notmyname> ok, so from my perspective, I'd like to see the crypto review branch land, I don't think non-crypto patches should be proposed to crypto-review, and I'd like to see the container sync stuff here land right after crypto
21:18:41 <notmyname> also, IMo oshritf_ has done a great job being patient and also giving the great performance info
21:19:26 <notmyname> does that make sense to everyone? or is there anything wrong with any of that?
21:19:42 <mattoliverau> yup, i agree
21:19:51 <notmyname> oshritf_: how does that sound to you?
21:20:03 <jrichli> sounds good!
21:20:41 <oshritf_> I'm pro crypto too, right after crypto sounds good
21:20:50 <notmyname> oshritf_: ok great
21:21:17 <notmyname> thanks again for your work on this and for bringing it up again (I do feel bad that it's taken so long)
21:21:22 <notmyname> let's move on
21:21:33 <notmyname> oshritf_: thanks for staying up so late
21:21:37 <notmyname> #topic hackathon
21:21:55 <notmyname> if you're coming and haven't registered, please do so asap https://www.eventbrite.com/e/openstack-swift-july-2016-hackathon-tickets-25913773795
21:21:58 <acoles> i just cherry-picked patch https://review.openstack.org/#/c/225338/ onto the crypto-review patch chain and it does create conflicts
21:21:59 <patchbot> acoles: patch 225338 - swift - Add thread level concurrency to container sync
21:22:07 <notmyname> acoles: ah, ok. thanks
21:22:33 <notmyname> if you haven't gotten a hotel room yet for the hackathon, do so asap. there's a room block available, and you should register soon
21:22:42 <mattoliverau> acoles: awesome
21:22:45 <notmyname> hurricanerix: any updates or info on logistics for the hackathon?
21:23:32 <notmyname> hmm... hurricanerix may not be online
21:23:38 <mattoliverau> the aloft already offers a shuttle to and from the castle. They can also pick you up from the airport if its at a reasonible hour.
21:23:44 <notmyname> cool
21:23:51 <mattoliverau> no idea what hurricanerix has organised tho
21:23:53 <notmyname> last I knew he was checking on the hotel room block to verify when it closes
21:23:56 <mattoliverau> I've just stayed there a few time :)
21:24:06 <notmyname> yeah, that's one reason he chose that hotel
21:24:32 <mattoliverau> I arrive at stupid o'clock so I'll need to uber.
21:24:42 <notmyname> any questions from anyone about the hackathon?
21:25:03 <torgomatic> is there gonna be transportation from hotel to office-space, or should i get a car?
21:25:29 <notmyname> that's what mattoliverau was just saying. seems there is a shuttle from aloft to the castle (rax office)
21:25:39 <notmyname> but I'll check with hurricanerix about that
21:25:48 <acoles> my network has been dropping, sorry I may have missed some of container sync the stuff
21:26:02 <notmyname> acoles: we're good. on to hackathon now (and just about to get to crypto)
21:26:14 <mathiasb> how does it work with proposing topics/work items for the hackathon?
21:26:17 <mattoliverau> acoles: if only yhe meeting had eventual consistancy.. oh yeah it does.. eavesdrop :P
21:26:28 <acoles> heh
21:26:39 <timburke> as far as verifying when the hotel block closes, the site says july 4
21:26:53 <notmyname> timburke: right, but that needs to be validated ;-)
21:27:00 <torgomatic> ah, that's what "the castle" refers to :p
21:27:07 <notmyname> mathiasb: we need to set up an etherpad soon to start collecting stuff. but the first agenda item is always "set the topics for the week"
21:27:17 <torgomatic> I don't know, maybe San Antonio has an old medieval fortification for reasons
21:27:51 <tdasilva> holding the hackathon at an old medieval castle would be kinda fun
21:27:53 <notmyname> the neighborhood it's in is named from king arthur stories
21:28:08 <notmyname> tdasilva: we'll have to go visit acoles again for that, I think
21:28:20 <mattoliverau> I'm in
21:28:22 <notmyname> (which sounds totally awesome for me)
21:28:41 <tdasilva> romans baths were also great! ;)
21:28:46 <acoles> notmyname: I may be older than you but calling me medieval is a bit much ;)
21:28:53 <notmyname> lol
21:28:55 <mattoliverau> lol
21:28:55 <joeljwright> :)
21:29:10 <notmyname> ok, one other thing slightly related to the hackathon is some other midcycle stuff
21:29:18 <notmyname> I just found out this today, so I wanted to pass it along
21:29:37 <notmyname> the ops midcycle is in NYC august 25-26
21:30:07 <notmyname> I've been to several of them, and I'm planning on going to this one too. generally good to hear from operators, and there's normally a little swift content there
21:30:08 <tdasilva> oh, right after openstack east
21:30:15 <notmyname> also, that same week is openstack east
21:30:22 <notmyname> which is a like a little mini-conference
21:30:52 <notmyname> not really a full summit thing, but if 6 months between summits is too long to go between vendor booths, this one is for you :-)
21:31:00 <tdasilva> notmyname: do you know if there will be any swift content at openstack east?
21:31:03 <notmyname> but seriously, it could be interesting http://www.openstackeast.com
21:31:16 <notmyname> tdasilva: not according to the current agenda
21:31:35 <notmyname> however, I'm planning on going (and today is the last day for early-bird registration)
21:31:54 <notmyname> it's the same week as the ops summit and a few blocks away, so I'll be in NYC that week
21:32:13 <notmyname> and I'm looking forward to seeing a lot of other people in the community and very likely talking about the golang stuff
21:32:24 <notmyname> I think it will be very good to do that before barcelona
21:32:47 <notmyname> anyway, if you're interested, I wanted to pass along the info on those events
21:32:58 <notmyname> especially because the price for openstack east goes up tomorrow
21:33:52 <notmyname> tdasilva: seems theres a keynote on "openstack project governance". so no direct swift content, but stuff that pertains to us :-)
21:34:05 <notmyname> anyway...let's move on :-)
21:34:14 <notmyname> #topic crypto review status
21:34:27 <notmyname> acoles has been doing a great job managing the patch chain
21:34:30 <notmyname> thanks acoles
21:34:39 <jrichli> +2
21:34:43 <notmyname> and to everyone reviewing, it's been looking great
21:34:44 <kota_> +2
21:34:51 <acoles> thanks for all the review comments
21:34:53 <notmyname> (to everyone not yet reviewing, GET TO IT!)
21:35:29 <notmyname> acoles: from your perspective, where do we stand?
21:35:55 <acoles> I didn't get through the updates to patch 328208 today so no new patch versions today
21:35:56 <patchbot> acoles: https://review.openstack.org/#/c/328208/ - swift (feature/crypto-review) - Enable object body and metadata encryption
21:36:14 <acoles> That is where we have substantial issues to work through ^^
21:36:20 <notmyname> ok
21:36:39 <acoles> on etag encryption strategy, derived ivs etc
21:37:10 <acoles> jrichli: is seting up a call to discuss some of that
21:37:11 <notmyname> jrichli: you wanted to set up a phone call on friday about that, right?
21:37:43 <jrichli> acoles notmyname: yes.  I have sent out an email to an initial list.  If you are interested in attending, please let me know
21:38:17 <kota_> jrichli: friday... morning/evening?
21:38:31 <jrichli> I am thinking 1:45 pm UTC
21:38:36 <notmyname> jrichli: yeah, kota_ should definitely be on that list :-)
21:39:13 <kota_> notmyname: is that midnight in my time, maybe?
21:39:25 <kota_> it's ok though.
21:39:32 <notmyname> close to it I think. it's before 6am my time
21:39:37 <jrichli> I was sort of thinking 1:30 pm UTC.  maybe that is better.
21:40:11 <jrichli> notmyname: yes, that is an issue.  is there any way you would be able to make it?
21:40:30 <notmyname> yeah, I will be there
21:41:02 <notmyname> if oshritf_ can stay up til midnight for this meeting (and cschwede nearly so), then I can get up early to get crypto landed ;-)
21:41:03 <jrichli> great.  I'd love to hear what torgomatic is thinking about all this too
21:41:06 <timburke> notmyname: shortly before 7 for us, no? DST and all
21:41:12 <acoles> I hope we can make some progress on the patch before then in gerrit and irc, I think we know there is a problem to fix with the derived iv. And there are some other comments for me to address.
21:41:42 <notmyname> oh. I can't keep PDT and PST straight. normally I log into a server set to UTC and run `date` and then do the math ;-)
21:41:52 <notmyname> acoles: +1
21:41:54 <mattoliverau> lol
21:42:01 <mattoliverau> word time buddy
21:42:07 <mattoliverau> *world
21:42:35 <joeljwright> it is easier being close to UTC :)
21:42:37 <notmyname> acoles: from a logistics process, anything that needs to change for crypto-review?
21:42:37 <mattoliverau> or set a calandar to iceland (who are UTC and have no daylight savings changes)
21:42:58 <notmyname> acoles: or is it going as well as can be expected?
21:43:30 <acoles> notmyname: for me mostly yes, how about for reviewers?
21:43:55 <notmyname> IMO it's been going well
21:44:28 <kota_> probably, I've not reached the final patch yet :/
21:44:29 <notmyname> jrichli: kota_: mattoliverau: torgomatic: timburke: logistics of crypto-review workign well for you?
21:45:02 <kota_> but, yes, I will be able to walk through by friday.
21:45:08 <torgomatic> so far, so good... I've already produced a 300-line patch that we're not gonna use :p
21:45:10 <mattoliverau> I'm a week behind due to moving, so still loading crypto in my head.. but so far so good :)
21:45:12 <jrichli> going well.
21:45:17 <notmyname> good
21:45:23 <notmyname> last logistics thing I wanted to cover about crypto-review is timing
21:45:35 <acoles> on patch 328204, torgomatic has suggested a change to unify the Putter classes that I have not yet processed
21:45:35 <patchbot> acoles: https://review.openstack.org/#/c/328204/ - swift (feature/crypto-review) - Support for http footers - Replication and EC
21:45:45 <mattoliverau> nice work torgomatic :P
21:46:05 <acoles> patches 328206 and 328207 have had v little review yet
21:46:21 <notmyname> when we started crypto-review, we set an initial 2 week window for soft-freeze of master and a goal of landing it by the end
21:46:32 <notmyname> that initial 2 week period expires on friday
21:46:43 <timburke> seems to be going well. i think first patch is pretty well settled now, with the second (and maybe third) nearly so. the cumulative changes are a little distracting when looking at say the fifth patch and wanting to compare patchsets 3 and 7, but it hasn't gotten to the point of me diffing `git diff` output, so that's good
21:46:49 <notmyname> IMO, it seems highly unlikely we'll get a consensus of positive reviews by friday
21:47:21 <acoles> agree
21:47:21 <notmyname> anyone disagree and think we have a good shot of landing this patch chain this week?
21:47:30 <mattoliverau> lets see what we can do, but /me thinks this'll definitely stretch into next week
21:48:23 <acoles> notmyname: unfortunately I will be away immediately after the phone conf on Friday
21:48:42 <notmyname> well, me too, but that's because I'll need to shower and eat breakfast ;-)
21:48:44 <mattoliverau> acoles: taking the weekend off!
21:49:14 <notmyname> acoles: when will you be back? monday?
21:49:20 <acoles> yes
21:49:23 <notmyname> ok
21:49:55 <notmyname> here's a proposal. please say if you agree or disagree. I propose we extend the soft freeze through the end of next week, again with the goal of landing crypto-review to master at the end of next week
21:50:09 <acoles> and like I said there's a couple of patches with no review yet
21:50:20 <acoles> agree
21:50:26 <mattoliverau> agress
21:50:32 <mattoliverau> agree even
21:50:33 <jrichli> agree
21:50:43 <torgomatic> argee
21:50:49 <kota_> agree
21:50:52 <torgomatic> eh, you know waht I mean
21:50:54 <torgomatic> ffs
21:51:07 <notmyname> great :-)
21:51:45 <notmyname> acoles: I'd like to look at the proposed docs patches. https://review.openstack.org/#/c/312315/ and https://review.openstack.org/#/c/330070/
21:51:45 <patchbot> notmyname: patch 312315 - swift - Adds migrated API reference files
21:51:46 <patchbot> notmyname: patch 330070 - swift - Add install-guide for swift
21:52:00 <notmyname> acoles: if they look reasonable and don't conflict, then I'd like to consider landing them on master
21:52:14 <notmyname> before crypto. since the docs team is waiting on us for that
21:52:27 <acoles> notmyname: sure, that should be fine, IIRC its all in a new dir?
21:52:27 <notmyname> acoles: but basically, I just want to look at them and talk with you about that
21:52:32 <notmyname> yeah, IIRC
21:52:37 <mattoliverau> notmyname: yes! Lana pinged me about those
21:52:54 <notmyname> mattoliverau: yeah, and I talked witht eh -docs team yesterday (2 days ago?) about it
21:53:02 <mattoliverau> cool
21:53:08 <notmyname> mattoliverau: just tell her "crypto!"
21:53:19 <mattoliverau> I have and she's ok with that.. it seems
21:53:38 <mattoliverau> I said soft freeze, I'll help shepherd them post
21:53:49 <notmyname> anything else about crypto anyone wants to bring up? any questions?
21:54:39 <notmyname> #topic open discussion
21:54:45 <notmyname> anything else to bring up this week?
21:54:56 <acoles> I know I have missed some comments along the way - if I do, please tell me, it's not deliberate!
21:55:08 <notmyname> acoles: you're doing great
21:55:13 * acoles looks to timburke
21:55:55 <timburke> acoles: don't worry, when i look at a new patchset, i go back through all my old comments and carry them forward if they haven't been addressed
21:56:07 <notmyname> ok, then let's adjourn until next week
21:56:08 <acoles> timburke: that's the way
21:56:12 <notmyname> thanks everyone for your work on swift
21:56:21 <loquacities> hi! happy to hold off on those swift patches if it's not too long :)O
21:56:23 <loquacities> :)
21:56:26 <notmyname> and, everyone, please go review crypto-review
21:56:35 <notmyname> loquacities: the docs ones?
21:56:41 <loquacities> yeah
21:56:47 <notmyname> loquacities: ack. thanks
21:56:51 <notmyname> #endmeeting