19:00:02 #startmeeting swift 19:00:03 Meeting started Wed Apr 29 19:00:02 2015 UTC and is due to finish in 60 minutes. The chair is notmyname. Information about MeetBot at http://wiki.debian.org/MeetBot. 19:00:05 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 19:00:07 The meeting name has been set to 'swift' 19:00:11 hello, everyone 19:00:15 o/ 19:00:17 o/ 19:00:18 here 19:00:18 who's here for the swift meeting? 19:00:19 hello 19:00:24 o/ 19:00:27 hello 19:00:29 o/ 19:01:19 😹 19:01:21 o/ 19:01:24 o/ 19:01:26 hola 19:01:37 welcome! 19:01:55 today's agenda has one item: summit session prep 19:02:01 #link https://wiki.openstack.org/wiki/Meetings/Swift 19:02:13 hi 19:02:18 so before we get into the sessions, general info and helpful links 19:02:31 room layout for the technical sessions: 19:02:33 #link https://docs.google.com/spreadsheets/d/1VsFdRYGbX5eCde81XDV7TrPBfEC7cgtOFikruYmqbPY/edit#gid=569963128 19:02:45 the overall schedule: 19:02:46 #link https://www.openstack.org/summit/vancouver-2015/schedule/ 19:03:03 you should have your passport by now. and probably flights and hotels 19:03:44 if you have a sched.org account, you can select the thigns you want to go to and download a calendar for you r phone 19:03:55 makes it easier to know where to be when 19:04:21 looks like it will be a busy week, but I think this will be one of the more productive summits for us 19:04:32 +1 19:04:40 I'm especially looking forward to the new technical track layout 19:04:48 there are 3 kinds of sessions 19:05:18 1) fishbowl: large rooms, similar to previous summits. facilitated discussions with a large audience. we have 6 of these 19:05:55 2) working sessions: small rooms, similar to the hackathons, focused discussions on specific things, write/review code. we have 10 of these 19:06:37 3) "meetup": we have a room all day friday for an open space meetup with no specific schedule in it. similar to friday in paris last november 19:07:18 if you have specific conflicts (eg I think I might be giving a talk during one of the swift sessions), please let me know asap 19:07:28 and we'll see if something can be worked out 19:07:46 so...where are we right now? 19:07:58 at home :P 19:08:09 we've got the RC2 tagged. that will be released tomorrow at around 1400UTC as swift 2.3.0 in openstack kilo 19:08:28 thank you, again, for all your work on it 19:08:37 way cool 19:08:45 awesome 19:09:07 great 19:09:23 ] 19:09:26 which brings us to "what's next", which is a perfect segue to the summit sessions (where we talk abotu what's going on and what's next) 19:09:31 #topic summit sessions 19:09:39 #link https://etherpad.openstack.org/p/liberty-swift-summit-topics 19:09:57 since last week, I've reorganized that etherpad 19:10:04 at the top are the fishbowl sessions 19:10:09 next are the working sessions 19:10:16 and at the bottom are the "overflow" sessions 19:10:31 I've got 6 fishbowls listed and IIRC 7 working sessions 19:10:36 so 2 things: 19:11:14 first, if your name is on one of the sessions, this week (today even!) make a etherpad with a high-level description or outline of the session and add it to this document 19:11:28 hmmm, all I see are blank sections for fish/working at the top and then a lundry list of ideas 19:11:57 peluse: have you tried turning it off and back on? 19:12:27 lol - love that show 19:12:27 oh, never mind :) 19:12:35 the linked etherpads can either be a scratch pad for ideas and we'll work through them as a group in the room. or they can be more of a place to collect thoughts for what needs to be discussed 19:12:38 stuff at the top are definitions and then I see 'type' with each idea 19:13:17 remember that while there is a small about of "presentation", especially for fishbowls, these sessions are more of a conversation or discussion, not a lecture with a huge slide deck 19:13:36 I will be very happy to help anyone with prep for a particular session 19:14:06 notmyname: ok, i will prepare it 19:14:23 ok, the second thing is that I'd like to go over the sessions in the bottom overflow sessions to fill out the working rooms and prioritize 19:14:39 dmorita: thanks! 19:15:12 ok, first one to look at is python-swiftclient status and next steps 19:15:35 I think this is important, but there isn't yet actually any specific thing around it other than "we should do something" 19:15:59 what do you think? should we make sure this gets specific time, or leave it to more informal conversations? 19:16:53 ...not everyone at once... 19:16:55 no input from clayg on this one? 19:17:13 clayg doesn't seem to be here :-( 19:17:20 it's a long way between "ugh" and "here's how to fix it" 19:17:23 seems like there's always plenty to talk about wrt client - at least last time anyways 19:17:25 hmm, I'm on the fence.. it is important, but do we have 40 minutes of conversation about it.. but it does need time i guess 19:17:49 maybe call it "a bitch session" :) 19:17:52 ok, in my arbitrary scale, let's put it at a 7 19:18:05 ok, fast-post spec 19:18:12 acoles isn't here (he had a work dinner) 19:18:37 do we need a full working session on this, or do we really just need to get the spec landed and work through it? 19:18:47 (same question for all of the spec ones, actually) 19:18:50 I've talked with him about a few of the tiems there, think its al stuff we will do 19:18:57 so probably land it and do it 19:19:13 probably the latter for this one; seems like we know *what* we want, it's just a matter of *how* 19:19:26 ok. priority 4 19:19:33 discussion is more valuable when there are differing opinions on what to do IMO 19:19:44 agreed 19:19:46 global cluster improvements 19:20:06 I think this one is low priority (I added it) 19:20:24 this should be more a response to what we're all seeing in prod, not ideas we come up with 19:20:35 however, if there are some specific things to bring up, that's good 19:20:47 anyone feel strongly about including this one in a working session? 19:21:17 sounds like a beer session ;P 19:21:31 lol 19:21:44 I can't see your face, so typing "no" helps ;-) 19:21:50 when the only tool you have is getting hammered, everything looks like a beer session ;) 19:21:56 lol 19:22:08 mattoliverau,: it's 6 am for you...and you already speak about beer? :) 19:22:11 * torgomatic doesn't care about this one 19:22:19 gvernik: he's australian. it's different there 19:22:25 gvernik: lol, 5:22 actaully :P 19:22:27 priority 1 19:22:30 not different than AZ! 19:22:34 next 19:22:36 * cschwede just arrived - hello! 19:22:37 symlink spec 19:22:53 torgomatic: do you want a working session on symlinks? 19:22:58 cschwede: welcome 19:23:04 notmyname: if we don't get the spec more hashed out by then, yes 19:23:04 I think we have 3 somewhat related sessions, that one, the policy changing and the netapp tier'ing 19:23:20 I could see that one going with policy-change, but not the tiering thing 19:23:31 do we hold all 3 and then spend some time Fri to sort out overlap (or not)? 19:24:01 well, tiering could be the decision engine that then uses policy change... 19:24:03 I'll mark them as potential candidates to combine 19:24:09 not a bad idea 19:24:24 yeah, just so everyone knows there's some potential relation there 19:24:46 ok 19:24:48 well, the difference is that I like symlinks, could tolerate policy changing, and dislike auto-tiering :) 19:24:55 :-) 19:25:19 nice 19:25:32 object undelete spec 19:25:49 cschwede: does this need a formal working session? 19:26:13 notmyname: no, not a formal one. discussion on friday is enough i think 19:26:18 ok 19:26:47 same question for container alias spec 19:26:57 same answer :) 19:27:19 updateable object sysmeta 19:27:27 isn't this related to fast post? 19:27:39 they're certainly playing in the same sandbox 19:28:02 i think it is 19:28:27 ok, I'll propose combining them and bring it up with acoles 19:28:48 next 19:28:58 metadata search / notifications 19:29:19 this is interesting, since it makes me think of zaqar and their recent call for adoption 19:29:33 cschwede: is there a spec yet? 19:29:42 no, no spec yet 19:30:32 would anyone rank it as more important than a swiftclient session? how about more important than the fastpost/sysmeta one? 19:30:35 I wonder if the zaqar folks would be open to a cross-project working session here 19:31:01 if we schedule it, I'll tag it as part of their track too 19:31:12 like I'll tag the ops session for ops 19:31:19 as it stands now, I believe that you have to get all kinds of keystoney to talk to a zaqar instance; there's no signature-based auth or something that'd be nice for use from within Swift 19:31:20 so it shows up in multiple tracks 19:32:03 ok, but where does this one rank? do we choose it over sysmeta/fastpost? 19:32:15 depends? 19:32:25 if zaqar folks are going to show up, yes. otherwise, no. IMO 19:32:57 that makes sense. the summit is a perfect place for cross-project discussions 19:33:15 hey hey o/ 19:33:28 I, in behalf of Zaqar team, I can say yes :) 19:33:28 of next up is kota_'s updates to container listings 19:33:34 vkmc: :-) 19:34:09 vkmc: should I ping you when it comes to scheduling to check for any potential conflicts? 19:34:21 please, feel free to ping me or flaper87 after the meeting or when you schedule the sessions 19:34:22 sure 19:34:31 ok 19:34:43 I will later this week when I start to schedule them 19:34:49 cool 19:35:16 ok, back to container listing updates 19:35:33 kota_: you proposed this one and said a friday discussion might be sufficient 19:35:58 ya, we don't need 40 minutes for this one, I think. 19:36:06 what do we all think? keep it on friday or have a working session? 19:36:25 fri 19:36:50 anyone want to argue for a working session? 19:36:58 not I 19:37:14 ok. let's bring it up on friday 19:37:37 weekly meeting time is already a friday discussion and we havent' seen any hummingbird code so there's nothing to talk about there yet 19:37:47 that's the end of the overflow list. thanks 19:37:56 now... 19:38:06 sorry. I would like to put https://review.openstack.org/#/c/149930/ on a friday discussion. 19:38:38 ho: ok. add it to the etherpad and mention it for a friday session there 19:38:46 of the ones I had higher in the working sessions 19:38:49 large contaienrs 19:38:53 notmyname: thanks! 19:38:54 ec follow-on work 19:38:58 tiering 19:39:00 change policies 19:39:09 encryption 19:39:14 increase part_power 19:39:15 docs 19:39:35 are there any there that should be dropped in favor of another? 19:39:39 So as it stands we have 7 working (what your going through now), plus swiftclient, symlink and messaging making 10. 19:40:03 mattoliverau: sounds good 19:40:05 sounds good 19:40:06 mattoliverau: yeah. i want to make sure the 7 I had aren't taking the place of one that we should have 19:40:20 specifically, the one I'm thinking of is the tiering one 19:40:35 i agree its good to go through and make sure :) 19:41:19 tiering could be replaced with fast post 19:42:14 yeah but fast post isn't very controversial 19:42:19 on the other hand, tiering is one of those things that I hear a lot from customers 19:42:29 yeah. fast post is like "yes, ok just do it" 19:42:38 however 40 minutes of torgomatic rage would be awesome :P 19:42:44 lol 19:42:56 lets leave it then 19:42:58 * peluse thinks he'll just say 'this is stupid' and leave after 5 minutes :) 19:42:58 just withhold coffee that mornign for extra torgomatic smash! 19:43:40 lol, awesome t-shrit idea 19:43:48 torgomatic SMASH! 19:43:57 lol 19:44:18 swift is many things, but it is not many more 19:44:19 ok, unless I hear differently soon (ping me in IRC) I'll keep it pretty much the way it is and how mattoliverau said it above 19:44:37 and I'll start scheduling those later this week 19:44:53 thanks for helping me go through them. I'm looking forward to it 19:44:57 #topic open discussion 19:45:05 anything else to bring up in the meeting this week? 19:45:13 one question about summit. 19:45:14 I want to drop py2.6 support in Swift after the Kilo release 19:45:26 dmorita: you first 19:45:35 general quesion. do you know what equipments are prepared in the room of design summit? 19:46:02 dmorita: generally we'll have a projector and a table and microphones (if the fishbowls are like the previous ones) 19:46:19 but that's a good question 19:46:27 ok. no change from previous summits. thanks. 19:46:31 AFAIK 19:46:42 I'll check with ttx to see if there is anything changed 19:46:46 torgomatic: your turn 19:47:08 basically, any objections to dropping 2.6 support after Kilo? 19:47:13 2.6 is dead, dead, dead 19:47:19 python core team won't touch it 19:47:28 lucid expires tomorrow IIRC. it had py26 19:47:35 what about centos/RH? 19:47:41 I'm not sure 19:47:52 where's the red hat people? 19:47:59 I think the latest one (6?) has 2.7 available, though it's not the default 19:48:13 does anyone here run on py26? 19:48:58 I think it's perfectly reasonable to drop py26 at this point. all of the major distros have py27 or don't even ship py26 anymore 19:49:12 I think we can assume that all new deployers will have py27 19:49:28 this is something that definitely needs to be asked of the operators 19:49:39 plus, writing code for 2.7 + 3.4 is annoying but reasonable 19:49:46 torgomatic: rather than waiting for the summit, can you send an email to the operators mailing list about this? 19:49:46 (or 3.X) 19:49:46 shall we bring it up at the ops session? 19:49:47 rhel6 uses py2.6 by default 19:49:56 cschwede: ah thanks 19:50:01 2.6, 2.7, and 3.X together is awful 19:50:11 cschwede: and I'm guessing it has support until about 2056? 19:50:19 lol 19:50:29 notmyname: close 19:50:31 and now that eventlet supports py3, it's probably getting towards time to get Swift working over there at some point 19:50:34 ;) 19:50:47 i’ll check that, get back to you later 19:50:51 ok, thanks 19:50:56 torgomatic: can you send that email? 19:51:07 notmyname: I'm not on the -operators list 19:51:14 * torgomatic already gets too much email 19:51:16 that's not hard to change ;-) 19:51:27 ok ok. I'll do it 19:51:31 thanks 19:51:50 that's called deligating up :P 19:52:08 :D 19:52:17 ok, anything else to bring up this week? 19:52:27 * torgomatic has nothing 19:53:02 thank you for coming today, and thanks for working on swift! 19:53:05 #endmeeting