21:00:21 #startmeeting swift 21:00:22 Meeting started Wed Feb 15 21:00:21 2017 UTC and is due to finish in 60 minutes. The chair is notmyname. Information about MeetBot at http://wiki.debian.org/MeetBot. 21:00:24 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 21:00:26 The meeting name has been set to 'swift' 21:00:29 who's here for the swift team meeting? 21:00:32 o/ 21:00:32 . 21:00:32 o/ 21:00:34 evenin 21:00:38 o/ 21:00:41 o/ 21:00:42 o/ 21:00:43 o/ 21:00:43 hi 21:00:45 hi 21:00:46 hi 21:01:23 clayg_: timburke: ping 21:01:27 hi 21:01:28 welcome, everyone 21:01:33 welcome back acoles 21:01:36 i'm here, i'm here! 21:01:41 thanks 21:01:50 agenda this week is at... 21:01:51 #link https://wiki.openstack.org/wiki/Meetings/Swift 21:02:01 hi 21:02:02 I want to cover the release and the PTG 21:02:28 #topic ocata release 21:02:47 as you know, this week (ie tomorrow) is the release deadline for openstack ocata 21:03:02 we'll be releasing swift 2.13.0. spoiler: it's the best one yet 21:03:17 I've got an authors/changelog patch up at https://review.openstack.org/#/c/434081/ 21:03:23 :-) 21:03:24 thanks for all the comments on it 21:04:09 I've addressed the open ones, and I'm currently translating the changelog to yaml for the reno dir tree. then I'll push that patch, and I expect it to be the one that lands (but i'll ask for a few extra eyes for quick checks) 21:04:26 and then when it lands we'll have the SHA for the release and I'll push that to the openstack-release repo 21:04:52 so we are or will be waiting on two things: me, to typey type and the gate queue 21:04:58 are we finally going to get our linky? as i recall, we don't have that yet... 21:05:24 thank you to everything who contributed to swift in this release. there's a lot of good improvements 21:05:51 timburke: I don't think so. I think there's another repos somewhere where we need to set up a job or reference the right thing or something 21:05:58 timburke: would be great though 21:06:24 so two key points about the release 21:06:34 first, please don't land anything else right now :-) 21:06:52 second, note that neither part power increase nor replicated ec fragments will be in the release 21:07:21 ont eh second note, I would have loved to see them in teh release (especially part power increase), but the release is in no way "bad" without them 21:07:40 just means we'll land them soon and the next release will be that much better ;-) 21:08:00 I know there's been a lot of work put in to those, especially over the last couple of weeks. thank you for that 21:08:04 and we'll have a cycle to make them even betterer 21:08:14 yup :-) 21:08:24 and actually have full global ec suppprt? 21:08:30 \o/ 21:08:35 :-D 21:08:43 :) 21:08:57 one interesting thing I noticed today about this release: I don't think there's a single CVE in it. I don't think that's good or bad, it's just interesting 21:09:26 notmyname: i'll have to work on that for next time ;-) 21:09:30 We also go back to the normal release length now. :-) 21:09:30 like not just 2.13.0, but all of ocata (anything after 2.10.0) 21:09:50 jungleboyj: yeah, but as cycle-with-intermediary, it's not a huge change either way 21:10:04 timburke: going to code up some buffer overflows? 21:10:12 Ahh, yeah ... sorry, Cinder head invading. ;-) 21:10:13 naturally 21:10:23 any questions from anyone about this release? 21:10:58 ok, let's move on 21:11:03 #topic PTG next week 21:11:06 #link https://etherpad.openstack.org/p/swift-ptg-pike 21:11:11 there's the scheduling etherpad 21:11:22 just for the record, there will be no weekly IRC meeting next week 21:11:29 in lieu of the ptg 21:12:21 after I get the release notes for 2.13 pushed, I'll be organizing the etherpad into blocks of topics so that there's as little overlap of interested people as possible 21:12:49 but remember that the three days we've got in our room will be lots of conversations at once 21:13:12 so that being said, come prepared to talk and to listen :-) 21:13:16 so use your inside voice? I don't have one of those :P 21:13:55 mattoliverau: I'm looking forward to your sharding topic. ready? 21:14:15 sure am.. ready as I'll ever be :) 21:14:26 there is some rough documentation up. 21:15:06 /going down the etherpad... 21:15:19 there's two topics related to golang object servers and another about the TC not-python rules 21:15:36 I'll be organizing some info on the TC rules so we can discuss that 21:16:02 I'm hoping that zaitcev, clayg_, and kota_ can help organize the golang stuff 21:16:04 was that today? 21:16:18 those names chosen because they're the 3 common names on those topics :-) 21:16:32 i hate the reconstructor; i hate all rebalances 21:17:18 kota_: looks like good notes on the global ec topic. anything you need from the rest of us on that topic? 21:17:18 * cschwede_ nods in clayg's direction 21:17:33 global ec includes composite rings discussion 21:18:13 notmyname: i think i added all i thought if i remember all things :P 21:18:20 heh, ok :-) 21:18:34 kota_: will kazuhiro be there? 21:18:41 notmyname: sure 21:18:45 great 21:19:24 between kazuhiro, kota_, tdasilva, jrichli, we'll cover policy migration, auto tiering, and symlinks. anything you need from the rest of us before we all get togehter? 21:20:29 ...I'm going to assume that's a "no, you're all good" ;-) 21:20:40 sorry, yeah, i'm all good 21:20:40 notmyname: or a ring zipper! 21:20:45 +1 21:20:50 clayg: write it down! :-) 21:20:54 hoping to have new version of symlink until then 21:20:57 what are we doing? 21:21:19 tdasilva: high-latency middleware? do you know more about that or need anything? (you put your name down on it) 21:21:34 there's a meeting tonight 21:21:37 at 4am EST 21:21:38 :D 21:21:41 tdasilva: sorry, media. not middleware 21:21:45 yikes! 21:21:54 lol 21:21:55 high latency middleware would be bad 21:22:01 lol 21:22:03 lol 21:22:24 what i know is that they have been meeting constantly and I'm guessing slavisa and others will be there at the PTG?? 21:22:31 cool 21:22:39 i'd like to follow up on how they are doing 21:22:40 on the sync to elasticsearch topic, I'll bug timur about it to make sure he's got what he needs 21:22:42 I think I have seen some high latency middleware before. 21:22:54 jungleboyj: I think I've written some! ;-) 21:23:01 :-) 21:23:25 will the people driving that topic be there? 21:23:27 fah. let's put it in the server itsefl! https://github.com/openstack/swift/blob/master/swift/obj/server.py#L132 21:23:33 cschwede_: are you good on part power increase topic? I'm assuming that the only "discussion" by next week is who buys the beers 21:23:35 there is the slowdown middleware thats great for testing, but hopefully noone is using in prod :P 21:23:42 clayg: HLM? i believe slavisa will be there 21:24:07 notmyname: well, that would be a great topic actually ;) i'm fine with partpower increasing 21:24:09 mattoliverau: oh yeah i forgot about that guy! 21:25:00 ok, I've talked to romain about the small files storage topic. they've done some pretty cool investigation at OVH and are interested in working on that further 21:25:23 current OVH info on that is at https://wiki.openstack.org/wiki/Swift/ideas/small_files/experimentations 21:25:24 cool 21:25:31 so read up on it so we can jump right in 21:25:40 nice 21:25:53 jrichli: notifications? good to go there? need anything from the rest of us before next week? 21:26:03 good to go 21:26:12 jrichli: and keymaster v2? 21:26:21 mathiasb would be the one to ask 21:26:37 but if he isn't here, i can ask him. 21:26:41 I'll be there for keymaster v2 21:26:42 mathiasb: question to you. need anything for us before next week? 21:26:48 for that topic? 21:26:58 no, I think we're good 21:27:00 kk 21:27:06 jrichli: and container sync? 21:27:28 good to go. not sure if we will get to that, anyway. so many topics! 21:27:52 lol, hackaton was hard to cover everything in 5 days, now we only have 3 21:28:03 actually it was 4 days 21:28:07 cschwede_: tdasilva: kota_: ring management topic. need anything from the group before next week? anything we can be reading over? 21:28:25 tdasilva: Cinder team has the same worry. Will be interesting to see how this PTG thing works out. 21:28:37 jungleboyj: yeah, agree 21:28:45 technically the full event is 5 days... 21:28:48 notmyname: not yet, i might write something down during my flight in 21:28:53 cschwede_: ok, thanks 21:29:22 acoles: kota_: the topic of ec issues. good to go on that? anything the rest of us can do before next week? 21:29:57 clarkb: just talking about the stuff we need to cover that's swift-specific that we've got a room for. not knocking the other two days :-) 21:30:20 that is a list we're currently tracking, if anyone else notice another issue, please feel free to add your issue over there. 21:30:26 ok 21:30:46 kota_: tdasilva: pyeclib/libec work? consolidation, etc? 21:31:04 notmyname: yes, ok for the ec issues 21:31:11 nothing from me atm, hoping we can discuss there what we would like it to be 21:31:16 ok 21:31:17 and maybe we can even chat with infra guys 21:31:21 tdasilva: true 21:31:23 is onovy going to be there? 21:31:29 kota_: timburke: swift3 work? do we need a topic? 21:32:07 notmyname: I don't have special for swift community but I think we still have some conversation while PTG 21:32:13 kota_: ok 21:32:19 so adding for notifiycation we will have that 21:32:41 and if anyone is interested in, welcome for joining that. 21:32:56 for swift3 21:33:08 tdasilva: I don't know if onovy will be there. 21:33:15 notmyname: ack 21:33:44 there's also a note from eranrom that there's some storlet topics and they've got a room too 21:33:45 notmyname: i'll ping him tomorrow to check and try to get some input from him in case he is not making it 21:34:05 tdasilva: great. (would also be nice to check with him in light of zigo's email this morning re packaging) 21:34:19 ok, that's the topics we have on the etherpad 21:34:24 yeah, i saw that...ok, i will ask him 21:34:52 since we've only got 3 days instead of the 4 we'd have at a hackathon, we WILL NOT be spending time at the front building a schedule and doing dots or anything 21:35:06 notmyname: WATH!? 21:35:07 so if there's somethign you want to talk about, PUT IT ON THE ETHERPAD :-) 21:35:12 how do you even hackathon without dots!? 21:35:17 i'm so confused 21:35:19 lol 21:35:28 * notmyname gives clayg a hug 21:35:37 * clayg deep breaths - i'm sure notmyname has got this 21:35:55 there will be a small amount of logistics just to see that we're getting stuff covered. and you know me. I like to talk ;-) 21:36:13 that's true 21:36:19 and show graphs ;) 21:36:20 so first question: what's left off of the etherpad right now? 21:36:31 notmyname: i do have a question. Do we know who is going to be there on Monday/Tuesday and assuming we might be able to find a corner somewhere are there any topics that could or could not be discussed in those two days? 21:36:36 mattoliverau: graphs are great. talking about graphs? just wonderful 21:36:50 tdasilva: I want to come back to mon/tues in just a bit 21:37:01 ok 21:37:06 first, anything left off for wed-fri that we know of right now? 21:37:36 basically this is last call for https://etherpad.openstack.org/p/swift-ptg-pike right? 21:37:54 sortof 21:38:03 we can add stuff as we go, of course 21:38:09 not that I can think of in my under caffenated state. 21:38:10 we've had that link in the last three meetings and everytime I look it it has more good stuff - but my stuff is already on there - so ... if you haven't got your stuff on there? WTF are you doing? 21:38:14 :P 21:38:16 but we won't be "together" like this any more 21:38:32 ...before atlanta 21:38:43 sorry. kinda awkward return placement there 21:39:09 sounded like we were all dying or something. EVERYTHIGN IS FINE 21:39:18 phew! 21:39:24 ok, let's talk mon/tues then 21:39:49 from what I've heard, i think most people will be there all week, but there's probablty about 1/3 of us who will only be there wed-fri 21:40:07 #link https://wiki.openstack.org/wiki/PTG/Pike/Etherpads 21:40:21 that's got the links for all the etherpads 21:40:31 AFAIK there is not an overall schedule 21:40:58 so I think we will need to look at each of those mon/tues ones and figure out where we need to be and when we need to be there 21:41:08 why don't you add the list of names who's be in which day in the etherpad? 21:41:16 there is for the shared "projector/discussion/Iforgetwhatitscalled" rooms but otherwise I think the idea was to use #openstack-ptg to coordinate things 21:41:16 kota_: that's a great idea 21:41:40 clarkb: where's the schedule for that shared room? 21:42:02 I guess i missed the #openstack-ptg channel info 21:42:09 https://ethercalc.openstack.org/Pike-PTG-Discussion-Rooms (its on the wiki page you linked) 21:42:12 /join #openstack-ptg 21:42:21 right at the top above "only monday" 21:42:47 to be fair, it's very tiny words ;-) 21:43:31 does anyone think we'll need a projector for a particular topic? 21:43:35 clarkb: I would like my monday and maybe tuesday to be sitting in a room of people who know enough stuff about devstack that when I say "I hate out devstack's swift support does xyz and I changed like this" they can say - "sure that's fine" - or "no you're stupid" 21:44:15 if so, we'll need to reserve the projector room and move over there. I'd suggest trying not to need a projector 21:44:16 i may just wander around yelling devstack until someone stops me 21:44:27 clayg: you want to find the QA room 21:44:34 I think that would be a good place for doing such things 21:46:37 what are other cross-project stuff we need to be yelling about on mon-tues? 21:46:42 tdasilva: what'd you have in mind? 21:47:00 tdasilva: or are you just wanting to find other swift people on mon/tues? ;-) 21:47:36 notmyname: i was just wondering if it does happen that a group of swift folks end up spending a couple hours together mon or tuesday if there was any topic that could be discussed 21:47:45 as to make time for the wed-friday discussiosn 21:47:49 tdasilva: there always is :-) 21:47:57 assuming there will probably not be enough time to discuss everything 21:48:16 as clayg said once, "I dare you to put two swift devs in teh same city any try not to do code/design work" ;-) 21:48:44 not enough time is probably a good assumption, unfortunately 21:49:00 I'll only really be there on Tuesday, but might need to go sit in Openstack Ansible so I can figure out if there is anything swift I can fix up for them. But that probably wont be the whole day. I just need to figure out if and when they talk about the os_swift role 21:50:00 does anyone else have anything else to bring up this week? 21:50:29 mattoliverau: what? you're only there on Tuesday? Single day? 21:50:47 cschwede_: for the mon/tues part :) 21:51:19 I actually arrive monday arvo.. but I don't think anyone would want to see me then :P 21:51:27 mattoliverau: ok, got it. it's late... 21:51:30 notmyname: maybe we need to solicit some more feedback to for cross project work? 21:51:41 clayg: in what way? 21:51:48 I've heard on numerous occasions that swift in devstack (and in particular logging) is not great - so that's what I want to work on 21:52:03 clayg: yeah, great 21:52:33 we've been lambasted for not making cross-project work a prioirty - so ... i guess our self prioritziation is wrong? 21:53:00 for folks that will be there Mon-Tues - we should probably be looking to get extrernal prioritization of how to spend that time? maybe? idk? 21:53:29 didn't someone write something something TC golang - shouldn't we talk to someone about that myabe? no one on our eatherpad even signed up as *interested* 21:54:13 ah, so a qustion come up in my mind 21:54:40 for the cross-project, where we can catch up the topics and schedule? 21:54:59 kota_: the best I know of is to look at the individual etherpads 21:55:06 ok 21:55:17 it would be scheduled already 21:55:19 clayg: all I can come up with to answer is "yes" but I don't feel that's a very helpful answer 21:55:42 kota_: sortof. there's no global scheduling. different teams are each given a room, and they can do whatever they want in that room 21:55:43 cross project or inter-projects? really both I suspect 21:56:07 mattoliverau: the inter-project stuff would be wtih the other wed-fri rooms 21:56:20 (if my definitions are current) 21:56:29 notmyname: that would make sense to me 21:56:34 the cross-project stuff is mon/tues 21:56:47 So going to help with say QA, infra or releases would be cross. I always though inter would be working with an other team on say swift stuff. 21:56:56 mattoliverau: right 21:57:24 i have to admit, i am not sure what to do monday. i think i am just gonna find one of you and follow along. 21:57:54 best I can see for cross project stuff is to show up in one of those rooms and listen? 21:58:00 I'll admit that's not a great plan 21:58:30 one thing that will be good eg is the tc not-python rules 21:58:34 there's been a bunch of chatter on the ML coming from osc's swift support - anyone going to be in https://etherpad.openstack.org/p/osc-ptg-pike 21:58:37 They have already gotten feedback that the cross-project needs to be better thought out. 21:58:48 i feel like an idiot: but i dont see anything on the room schedule for monday excpet for oslo at 11. what am i missing? 21:58:50 Needs to be more focus on inter-project. 21:59:04 clarkb: was QA - question/answer or quality assurance? 21:59:31 jrichli: are you just looking at the projector rooms sched? 21:59:34 https://wiki.openstack.org/wiki/PTG/Pike/Etherpads <- jrichli maybe this one is better? 22:00:17 clayg: Quality assurance (devstack, granade, etc) 22:00:22 storyboard? do we need to know about storyboard? 22:00:27 I'll likely try to go by the docs room to talk about consolidating our docs trees. i'll likely try to go by the infra one to talk about golang stuff 22:00:27 mattoliverau: I DID NOT KNOW THAT!? 22:00:36 looks like we're running out of time 22:00:45 kota_: ah, you're right 22:01:01 I'm looking forward to seeing everyone next week 22:01:02 notmyname: the WSGI thing? 22:01:17 clayg: yeah, the release goals are important for us to be present in 22:01:38 thanks everyone for coming. thanks for working on swift 22:01:42 #endmeeting