19:00:46 #startmeeting swift 19:00:47 Meeting started Wed Apr 22 19:00:46 2015 UTC and is due to finish in 60 minutes. The chair is notmyname. Information about MeetBot at http://wiki.debian.org/MeetBot. 19:00:48 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 19:00:50 The meeting name has been set to 'swift' 19:01:04 who's here for the swift team meeting? 19:01:09 o/ 19:01:11 yo 19:01:12 o/ 19:01:13 o/ 19:01:13 hello 19:01:14 hi 19:01:14 hi 19:01:15 o/ 19:01:17 hello 19:01:19 hi 19:01:21 o/ 19:01:26 \o/// 19:01:26 o/ 19:01:29 here for a bit, then running off 19:01:31 o/ 19:02:03 wow. quite a crowd this week. welcome! 19:02:45 just as an FYI, at the summit I want to talk about a different meeting time. perhaps today, but 2 hours later. or perhaps tomorrow at 2000UTC (one hour later). 19:02:50 something to think about 19:03:19 the full openstack meeting schedule is at 19:03:20 #link https://wiki.openstack.org/wiki/Meetings 19:03:32 so you can see if a meeting room is free at a time you want 19:03:40 also, http://www.timeanddate.com/worldclock/meetingtime.html?iso=20150725&p1=224&p2=136&p3=152&p4=248&p5=70&p6=37 19:03:41 i'm not here 19:03:47 notmyname: so good and I just put the idea to ethepad 19:03:52 :-) 19:04:03 ok, a few things on the agenda this week 19:04:09 #link https://wiki.openstack.org/wiki/Meetings/Swift 19:04:14 first up, kilo release 19:04:19 #topic kilo release (swift 2.3.0) 19:04:36 patches just now landed for the ones we had backported for RC2 19:04:50 current HEAD of stable/kilo is at commit SHA f8dee761bd36f857aa1288c27e095907032fad68 19:04:58 yay 19:05:07 if this should NOT be our final release, please speak up (like right now) 19:05:22 we do not expect a third RC 19:05:49 if there is no objection, then I'll give that SHA to ttx as the final one for RC2 19:06:04 he'll tag it and we'll keep testing to see if anything comes up 19:06:21 only major issues (eg data loss/corruption) would cause an RC3 to happen 19:06:48 any questions? any objections to doing the RC2 at the current commit? 19:07:14 notmyname: what if a failure of both the primary and backup system caused a bug in the secondary backup to delay correction of a previous hardware failure resulting in data loss? 19:07:28 clayg: we'd need to ask Scotty 19:07:38 i'm giving it all she's got! 19:08:07 lol, sorry been alseep so looking at current state of patches :) 19:08:21 lol 19:08:38 no objections...going once... 19:08:53 twice..... 19:09:03 twice and a half... 19:09:08 rofl 19:09:10 this is like my marriage 19:09:15 wait... 19:09:19 acoles: ok 19:09:19 acoles: stop it 19:09:19 no its ok 19:09:24 ok 19:09:40 ok 19:09:45 done 19:09:46 ok, i'm board with kilo now 19:09:46 #info commit f8dee761bd36f857aa1288c27e095907032fad68 is RC2 19:09:49 swift: last to RC1, but first to RC2! 19:09:55 ttx: yay@ 19:10:12 ttx: that's the commit to use. thanks 19:10:28 notmyname: ack 19:10:48 ok, so looking forward with kilo, just some info on some testing that will be happening 19:11:10 I hope that everyone will be able to of course test previous functionality, but also look at EC 19:11:22 we've got a lot to learn there about EC in prod 19:11:55 notmyname: second 19:11:56 over the next few weeks at swiftstack, we'll be using the community QA cluster for testing some EC stuff 19:12:34 filling it up, dropping disks, checking reconstructor times, CPU/mem usage, etc 19:12:41 all that kind of stuff 19:13:10 is anyone else also going to be putting it in a lab to test? 19:13:35 kota_: you have already been looking at deploying EC right? 19:13:46 notmyname: thiago and me want to do some testing before the summit 19:13:49 hurricanerix: glange: any chance RAX can spend some lab time on it? 19:13:53 cschwede: nice 19:14:09 notmyname: yes, but my lab environment 19:14:18 kota_: where it should be :-) 19:14:51 it would be awesome to have some preliminary results to talk about in vancouver. 19:14:53 notmyname: i don't think we have any plans for EC at this time. 19:14:57 ok 19:15:10 notmyname: mseger has been talking about some perf testing 19:15:28 I wonder if there's a way we can aggregate the lab work different companies are working on 19:15:48 probably the first step is to actually start testing. and then try to be as open as possible with any results found 19:15:49 I'll throw something up in the cloud. /me is probably the biggest user of rackspace Australian cloud :P 19:16:30 ok, let's move on to summit planning 19:16:35 #topic summit prep 19:16:44 * notmyname got plane tickets yesterday 19:16:47 (finally) 19:17:07 #link https://etherpad.openstack.org/p/liberty-swift-summit-topics 19:17:23 that's where we're collecting our ideas for summit topics 19:18:07 as a reminder, the fishbowl sessions are large rooms (200-ish seats?) designed like previous summit sessions 19:18:24 the working sessions are smaller (30-ish seats) and will all have the same title on the schedule 19:18:50 they are designed to be more about smaller groups of people discussion specific copde and specific ideas' design 19:19:07 and then on friday we have a full day of "open" like we did in paris 19:19:22 notmyname: as a reminder, how many slots for each session type? 19:19:31 good question 19:19:42 6 fishbowl and 10 working, IIRC 19:19:56 notmyname: thanks 19:20:00 so we'll be collecting ideas on that etherpad and scheduling from there 19:20:00 * clayg wishes *he* had a good question 19:20:03 you might have said before, but is each session for an hour? 19:20:08 notmyname: confirmed 19:20:19 each session is 40 minutes 19:20:26 ttx: thx 19:20:29 (10 minutes between sessions) 19:20:52 ttx: thanks 19:20:53 coffee, pee, argue, repeat 19:20:54 #link https://docs.google.com/spreadsheets/d/1VsFdRYGbX5eCde81XDV7TrPBfEC7cgtOFikruYmqbPY/edit#gid=569963128 19:20:54 acoles: Donagh may run a Docs working session, so I finally see you in your alter ego :P 19:21:04 lol 19:21:06 lol 19:21:08 * torgomatic has to bail 19:21:25 ttx: that's the right link, right? ^ 19:22:12 mattoliverau: i have a good make-up artist 19:22:28 the summary is: tuesday for ops, wed-friday for swift 19:23:10 ok, let's go over what's in the etherpad 19:23:15 notmyname: looks like it 19:23:20 notmyname: can i get a link to it please? 19:23:33 notmyname: ok, rc2 tag and tarball up 19:23:38 looks like we start fishbowls on wednesday night and continue on thursday morning 19:23:40 pushing announce 19:23:43 scotticus: which one? 19:23:49 etherpad 19:23:53 https://etherpad.openstack.org/p/liberty-swift-summit-topics 19:23:53 https://etherpad.openstack.org/p/liberty-swift-summit-topics 19:24:07 thanks. 19:24:39 ok, looks like I didn't add any detail to any of the ones I rushed to add last week ;-) 19:24:55 but there's a few new ones at the bottom 19:25:14 Did you want me to talk to Dean about doing a fishbowl on tape and swift? 19:25:15 cschwede: metadata search or object notification 19:25:28 jrichli: yes, please. or facilitate an intro 19:25:49 ok, I will talk with him. 19:26:18 for metadata search or object notification. what sort of sessions? 19:26:23 fishbowl or wokring session 19:26:30 notmyname: the metadata/search topic comes up from time to time, and i like to see some progress on it - thus preparing a spec to start some discussions 19:26:36 cool 19:26:46 notmyname: i think working might be better? 19:26:55 heh. everythign is working sessions ;-) 19:27:27 probably just means we're anti-social and cliqueish 19:28:10 clayg: or focused on getting s… done… 19:28:10 I think it's a great topic, and yeah it would be cool to start work on it 19:28:39 ok, next 19:28:39 notmyname: metadata search? hrmmm... 19:29:00 clayg: well, notifying another system that will actually do the search ;-) 19:29:09 clayg: actually I'd love for that fucntionality to be in glance 19:29:13 notmyname: exactly like that 19:29:23 (another system) 19:29:38 also, that directly relates to the plea from zaqar on the mailing list this week for more adoption 19:29:38 metadata search: sounds like it wont play well with encrypted metadata 19:29:48 and I think this is one of the best places for it 19:30:07 jrichli: you havne't implemented fully-homomorphic encryption yet? ;-) 19:30:12 anyway, moving on 19:30:15 docs 19:30:19 this one is interesting 19:30:26 * clayg has a doubt 19:30:58 well, I think it's interesting because it's an ops person who submitted it rather than a dev 19:31:17 so, basically, make docs better 19:31:28 or rather, make a plan to make docs better 19:31:29 what, acoles is a dev :P 19:31:44 mattoliverau: gotta keep up the facade 19:31:47 * cschwede likes to see a good community-written reference doc on swift 19:31:51 notmyname: where do i put my proposal? at the bottom? 19:31:52 seriously tho, yeah, good to hear from another viewpoint 19:31:53 mattoliverau: rumours rumours... 19:32:02 scotticus: yeah 19:32:04 k 19:32:37 ok, next up 19:32:46 last modified on container listing 19:32:52 sounds like a filesystem atime 19:32:59 kota_: this one's yours 19:33:19 notmyname: yes, this is a first time I put this idea. 19:33:19 can this be covered in the metadata indexer one? is that a viable alternative implementation? 19:33:58 notmyname: for now, so many change we don't need 19:34:04 kota_: oh! this is for a container, not on a per-object basis? 19:34:16 notmyname: yes, on container. 19:34:36 yeah, I've been asked about a "listing of the stuff that has changed since Timestamp(foo)" 19:35:02 notmyname: that, please! 19:35:11 cschwede: typey typey 19:35:18 notmyname: the use case is something like that :) 19:35:23 interesting 19:35:40 so should this be it's own session or should it be combined with another? 19:35:42 thoughts? 19:36:06 ethier is fine but currently no idea 19:36:20 * cschwede starts thinking that we need an „OpenStack *Swift* Summit“ to cover all the interesting topics 19:36:58 cschwede:paris in june? 19:37:26 :-) 19:37:33 acoles: notmyname: sounds good :) 19:38:03 I made a note about combining this one, but let's leave it separate and see what space is available 19:38:07 ok, next up 19:38:10 meeting time 19:38:15 thanks to who put that down 19:38:44 I'll be happy to lead that one, but I think it's just a friday discusion. no need for a 40 minute session on it 19:38:50 any objections? 19:38:54 +1 19:39:01 will that take 40 minutes, so friday discussion makes sense 19:39:16 +1 19:39:17 or in a bar somewhere ;) 19:39:29 +1 19:39:34 mattoliveau: nice! 19:39:55 ok, last on one the list right now 19:40:03 scaling swift with go 19:40:09 :) 19:40:14 this is something I've been talking with scotticus for a little bit 19:40:27 and some of you know that RAX has reimplemented some of swift in go 19:40:31 so they want to talk about it 19:40:38 and I think that's great! 19:40:56 scotticus: will you have some code to show before vancouver? 19:41:02 yes. 19:41:04 good 19:41:13 when could we expect it? this week? next? 19:41:18 i think the team will push the object server and possibly a go based replicator and auditor soon. 19:41:30 within the next couple weeks probably 19:41:34 ok 19:41:37 +1, it'll be awesome to talk about and see :) 19:41:41 obviously we want to vet it 19:42:02 a bit 19:42:12 if we've got the code to talk about, I think it's a great thing to cover :-) 19:42:18 yup 19:42:30 for those of you who have been paying attention, swift now has a feature/hummingbird branch upstream 19:42:45 that was created to give them a place to push that code in the community 19:42:50 i fetch and i fetch 19:43:12 good boy :) 19:43:13 so should this one be a fishbowl or a working session? 19:43:19 I almost think maybe both 19:43:20 a watched branch never commits :P 19:43:43 let's just take over the two-day-nova-fishbowl-fiesta 19:43:49 :-) 19:43:56 nope go-swift or go-home 19:44:53 notmyname: probably working session? 19:44:55 thoughts on having this in a larger session? 19:45:54 I'm happy with either 19:46:03 there's TONS of stuff to go over with the idea of golang stuff in swift. that's the kind of stuff I think would be great in a working sessions. API coverage details, testing, deployments 19:46:38 but the higher-level, "this is what we saw in swift and XYZ is the right answer" might be good in a fishbowl session? 19:46:39 the more the better! 19:47:02 that sounds pretty decent. 19:47:06 notmyname: can we defer until we see the code? 19:47:14 heh ok :-) 19:47:18 i think we can tell everyone what we've found at scale and what we've done to help 19:47:36 and we can def have a session to discuss what the community wants to see out of a go based swift implementation. 19:48:02 so fishbowl+working? 19:48:12 workbowl? 19:48:20 fishing? 19:48:28 lol 19:48:33 let's figure it out based on what we have to talk about. ie the code 19:48:37 lol 19:48:43 k 19:48:46 keep it on the list now with a question mark 19:48:55 decide later 19:49:23 yay, leave it to future us :) 19:49:26 k 19:49:29 ok, that's the (current) bottom of the list on the etherpad 19:49:30 I like them 19:49:41 anything else to bring up in this meeting for the summit scheduling? 19:50:07 ok, last topic then... 19:50:14 #topic swift-bench release 19:50:22 cschwede: you've got the floor 19:50:31 Sometimes I just want to run swift-bench locally for a quick check if everything works 19:50:32 Unfortunately swift-bench from pypi is broken, probably since a longer time :/ 19:50:33 Lots of commits since the latest tagged version 19:50:35 There is also a minor bug, fixed here: https://review.openstack.org/#/c/175285/ 19:50:36 It would be great if we could tag a new version after the fix has been committed, and release a new version on pypi 19:50:37 Doing this would make installing it (hopefully) as easy as "pip install swift-bench" again 19:50:38 What needs to be done to get a new version up on pypi? 19:50:39 tada 19:51:16 "lots"? like 8 19:51:33 cschwede: what needs to be done is to kick me in the pants to get it done. takes about 15 seconds 19:51:50 notmyname: cschwede: well let's plan to do it like next week or something 19:51:51 notmyname: ok, maybe not lots, but useful/helpful fixes 19:52:02 cool, let's land https://review.openstack.org/#/c/175285/ and then release it 19:52:06 notmyname: cschwede: I think I have some local hacks to swift-bench I carry around that I should just see if I need to git review or what 19:52:08 notmyname: clayg: thx, great! 19:52:17 clayg: ok 19:52:42 #action review https://review.openstack.org/#/c/175285/ and release swift-bench 19:52:53 cschwede: anything else there? 19:53:05 notmyname: no, thx! 19:53:08 #topic open discussion 19:53:15 anything else to bring up in the meeting this week? 19:53:25 ok, nm - I think the only think i had left got fixed in I2fca4795fc9b3d1637c72616a52a60482d6e853e 19:54:02 ok, if nothing else... 19:54:13 thanks for working on swift and thanks for coming today! 19:54:17 swift isnt' possible without you 19:54:20 #endmeeting