17:01:54 #startmeeting Designate 17:01:55 Meeting started Wed May 6 17:01:54 2015 UTC and is due to finish in 60 minutes. The chair is Kiall. Information about MeetBot at http://wiki.debian.org/MeetBot. 17:01:56 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 17:01:59 The meeting name has been set to 'designate' 17:02:06 Hey folks - Who's about? 17:02:09 o? 17:02:10 o/ 17:02:11 o/ 17:02:15 o/ 17:02:16 o? 17:02:27 hey 17:02:53 Cool :) So, the agenda looks somewhat out of date.. considerng we've already shipped Kilo! So.. I'm going to go ahead and ignore that one :) 17:03:08 #topic Design Summit Sessions 17:03:10 #link https://etherpad.openstack.org/p/designate-liberty-proposed-sessions 17:03:19 o/ 17:03:54 We have a bunch of possibles in there, with I'm sure more to come.. Has everyone had time to fill out any they wanted to discuss? 17:04:30 Yep 17:04:54 i think so 17:05:02 Okay, cool.. Should we have a look at those now and see which we think are useful etc? 17:05:25 ah.. lots more being filled out as we speak.. lets give it 2 or 3 mins for people to flesh out 17:05:31 o/ 17:05:53 I'll be there / interested to talk about horizon stuff with others for designate 17:06:20 ducttape_: cool. I have a few ideas I want to flesh out for it 17:06:36 Okay, looks like the editing has stopped.. 17:07:21 yup 17:07:41 So, there's 8 40 min "workrooms", 1 1 hr 40min "Fishbowl", and 1 longer "Meeup" 17:08:01 Some of these are back to back, and can be combined.. or not.. up to us 17:08:12 Input/feedback/questions from non-contributors -> meetup maybe? 17:08:21 but it is late on the friday 17:08:21 mugsie: ++ 17:08:23 mugsie: +1 17:08:33 Well, We can always use a workroom session for that too.. 17:08:43 friday is always a terrible summit attendance day 17:08:52 yeah, especailly the late sessions 17:08:59 as people tend to get flights etc 17:09:06 Maybe the early sessions on wednesday then? 17:09:24 Okay, Let's start at the top then, eh? 17:09:28 will there be a free form room where any team can grab a table too? that might help as well 17:09:35 ducttape_: there shouod be 17:09:39 ducttape_: I actually think those are gone this year? 17:09:44 *but* it does fill up 17:09:58 Kiall: oh, i thought the lounge was still there 17:10:09 we will need to find a designate sign, and place it at the closest bar then 17:10:16 lounge yes - but that's usually different / smaller to last time ;) 17:10:22 Okay - First up - Revisit DNSSEC.. 17:10:36 ++ 17:10:40 need it. 17:10:53 Seems like we have more of the pieces needed for that now. 17:10:57 I'd say No here, we should sync up with Barbican to get some commitment on integration.. But.. I'm not sure we have a session out of it.. We have a 90% plan together, and just need to act on it 17:11:17 (At least, that's my take on it based on last time :)) 17:12:07 true 17:12:12 timsim: Yea, we have lots of the pieces in place where we can go do it now, while we just didn't have it ready to go last time IMO 17:12:45 So - Others in the room? Let's give things "Priority".. "Low" for this one? 17:12:56 (Then we full out based on priority) 17:13:00 fill* 17:13:08 ++Low 17:13:21 Silence means acceptance 17:13:26 Next up.. Blueprint/Spec Reviews (IXFR, transaction support, granular access, more?) 17:13:53 Yep, I this this is a great one for day 1 .. Quick overview/rehash of what's been talked about etc 17:14:09 we should definitely discuss things we want to do in Liberty and so this would be high 17:14:15 Seems like those should be broken up into their own sessions. 17:14:33 yeah, more than 1 would be good 17:14:35 Those 40 minute sessions go by real quick. 17:15:07 timsim: True, I think some of those deserve a full session.. But I suspect we should use 1 slot to recap on things that don't get a slot? 17:15:23 Sounds good 17:15:42 (e.g. IXFR, We're closer on that .. I believe I saw an updated spec come in from rjrjr, but digging into that without him and the summit seems counter productive) 17:16:07 yes, it is being updated. 17:16:26 I'd say "High" here - with the specific set of things being open, and being those that don't get a slot of their own. 17:16:39 I won't be at the summit, but if the hook point api is discussed, I'm happy to hop on a chat / hangout 17:17:05 elarson: I'm sure that can be arranged :) 17:17:19 ++High 17:17:27 Again, silence means agreement ;) But I'd prefer to see Yay or Nay's ;) 17:17:34 ++ 17:17:37 ++high 17:17:51 Thanks :) 17:17:56 I think we need to dig into the hook point stuff, and in person would be better 17:18:26 Next up, Input/feedback/questions from non-contributors - I'd go with Medium here? 17:18:44 med/high 17:18:52 it is a good place to get people 17:19:01 high 17:19:02 Yep 17:19:15 Yeah remember how many randos we hadn't met showed up at our things last year. High 17:19:23 Yay !!! 17:19:26 Okay.. Sounds like it's leaning towards High.. Who's bringing the weed for everyone, that's legal in Vancouver, right? 17:19:40 "informally" legal 17:19:44 A+ 17:19:46 * Kiall watches the tumbleweek pass ;) 17:20:16 Kiall: its only legal about 50 miles south ;) 17:20:46 And IRC seems dead 17:20:51 Next up Hackathon 17:20:51 ? - I reckon that fits into the last day Meetup session, or other gaps 17:20:51 Oh great. IRC seems to have died 17:20:52 * ducttape_ is glad he joined this room today 17:21:02 Ah.. there we go! 17:21:11 haha :) 17:21:12 Kiall: use a real IRC client :p 17:21:16 Psh 17:21:21 yeah, like in emacs 17:21:27 Next up Hackathon.. 17:21:36 yeah i'd like to see a Friday Hackathon 17:21:48 I think that's a Low, in that we can fit that in around sessions / in the last day meetup 17:21:49 elarson: irssi 4 lyf 17:21:57 Kiall: ++ 17:22:03 definitely low 17:22:14 As long as it gets in there somewhere 17:22:14 "You have been disconnected." from etherpad -_- 17:22:31 and.. connected again 17:22:35 seems my ISP hates me 17:22:41 Next up - Database Interactions / Rework 17:22:49 I added this one, so my vote is clearly for High ;) 17:22:49 high 17:23:00 our DB is .... in an interesting state 17:23:08 very high 17:23:22 ++High 17:23:27 organic growth is the word.. What used to be nice a simple, has turned into an annoying crutch we have to workaround 17:23:40 High it is 17:23:46 Metering & Billing / Ceilomemter Integration 17:23:55 med 17:24:12 Low/Med 17:24:26 I am not sure there is a huge amount of stuff we need to do, and most of it needs to be talking to celiometer 17:24:28 I'd actually say High - Not because of M&B, but because the underlying infrastructure we need to support it at scale will be new - and will underpin other future work too 17:25:07 oh, true 17:25:16 e.g. every 60 mins or so, we have to emit a "Yes, this domain still exists" event.. With 5mil domains, that becomes a interesting challenge to manage and ensure we only send 1 out etc, and that it can scale out horizontally 17:25:29 (timsim will remember my mention of "tooz" a week or two back ;)) 17:25:45 Alright, sounds like fun. 17:25:50 It also plays into active/active pool managers, and not double double work on the sync's etc 17:25:57 (i.e. it's foundational) 17:26:08 not doing double work* etc 17:26:13 if it brings in the sharding stuff, high 17:26:18 So - My vote is High, others? 17:26:21 but that could probably be separated out 17:26:25 high 17:26:50 I'm fine with high 17:26:55 Settled - Next.. Horizon / OpenStack Client Integrations 17:26:58 high 17:27:35 +1 17:27:38 I'm open on this one.. It's somewhat vague.. do we have specific ideas in mind? 17:27:38 high for me, I can probably tackle a few things others might like 17:27:57 e.g. getting into a room with a topic of "Horizon" seems a bit wide open :P 17:28:08 its split out below it ;) 17:28:20 I'd think something like additional features for cloud admin type users might be of interest? 17:28:22 That said, I know we can fill the hour )Oh.. True.. I write the first bullet there but forgot 17:28:23 lol 17:28:31 * mugsie says nothing 17:28:32 :D 17:28:34 timsim / jmcbride etc? :) 17:28:43 Also.. Horizon / OpenStack Client Integrations 17:28:46 that seems like 2 things to me 17:28:56 ++ 17:28:57 Yeah those do seem separate 17:29:14 Horizon isn't a priority for us, so much. 17:29:16 the concepts should be similar, but they can be split np 17:29:20 client integrations first, then those can support GUI 17:29:24 timsim: I figured that would be your answer :D 17:29:42 it is a priority for us however. 17:29:53 I'd say it's a Medium, but majority is leaning towards High 17:30:15 +1 seems more than worth of a short session or two 17:30:24 Heat Resources for Designate? .. I threw it down, and consider it a Low.. Esp as someone from Heat seems to be working on it. 17:30:41 Agreed, Low 17:30:57 +1 17:31:02 +1 17:31:08 Next - OpenStack Client Integrations .. 17:31:47 I kinda think it's a Low? We have tonnes of things we know can do better.. names rather than UUIDs etc, getting to that point, then revisiting for extra makes sense 17:32:14 ok, low 17:32:19 Others? 17:32:26 Low 17:32:42 Moving on quickly! 17:32:47 Nova/Neutron Integration 17:32:52 high, super high 17:32:59 Everybody gonna want that 17:33:04 High from me, and I'd like to grab the Nova/Neutron folks into the session too ;) 17:33:08 ++++ 17:33:32 Seems an obious High then ;) 17:33:34 Non MiniDNS Backends as 1st class citizens 17:33:36 ++high 17:33:43 high 17:33:46 Yea, we need to fix that story once and for all 17:33:47 High 17:33:51 oops, ++high for nova/neutron 17:33:51 High 17:34:04 (We're running out of slots to even handle all the "High" .. lol) 17:34:26 bharath: oh, too late ;) I recorded your vote :P 17:34:38 :) 17:34:57 Designate Sink - Multiple handlers with different topics 17:35:04 I can guess who says High here :P 17:35:26 high 17:35:28 I had put this up based on our experience here but we can discuss this offline too 17:35:50 not a session worthy subject though. 17:35:55 Yep, I'd call it a Medium.. Not because it's something we shouldn't do, but because it's something we can do outside a session 17:36:01 rjrjr++ 17:36:21 yeah 17:36:33 med then 17:36:51 Okay.. Seems stuff is still coming in :D 17:36:56 root domain redirect.. 17:37:24 I see this as being aligned with geo/weighted round robin/etc 17:37:24 ALIAS type 17:37:31 mugsie: yea 17:37:52 I think it is distinct from them, but may share some simlar implementations 17:38:05 I'm not sure alias/geo/wrr/etc is something we have time this cycle to tacle.. 17:38:20 many sites need it, but not as part of dns load balencing 17:38:50 Yea, Implementation wise, I think they all end up very very similar.. "Do stuff with DNS your DNS server doesn't do out of the box" kinda sums the group up 17:39:02 yeah 17:39:19 Discussing ALIAS without Geo/WRR etc would miss a large chunk of 99% identical requirements etc 17:39:54 So - I think I'd say Low for this summit, High for next kinda thing 17:39:58 Others? 17:40:10 high 17:40:22 dooh.. stalemate ;) 17:40:23 actually, medium 17:40:32 med 17:40:49 Okay, sounds like leaning towads mediym 17:40:51 medium* 17:41:13 ptr records in multi? I'm not quite sure what this one is 17:41:22 Who's purple? ;) 17:41:23 my bad 17:41:34 jmcbride is finishing that off, one sec 17:41:43 should be: in multi-tenant untrusted environments (e.g. public cloud) 17:42:03 jmcbride: specifically for Public IPs, or private overlapping Neutron IP space? 17:42:07 that exists does it not? 17:42:20 Public IPs. 17:42:21 I'm thinking public ips first. 17:42:39 Right.. that should in theory work todetay, but I don't think anyone has taken it to production y 17:42:41 et 17:42:47 the v2 API has a reverse endpoint 17:42:58 * ducttape_ just found out there will be open tables at the conference to borrow 17:43:09 use case: I don't want anyone to claim a ptr record for an ip they don't potentially use 17:43:56 mugsie: really? 17:43:59 Oh .. I see what's happening here.. RAX's deploy model is different to "normal" :) 17:44:07 :P 17:44:26 RAX does the older nova-network model - every instances gets a "fixed" public IP, while others use FLoatingIPs you have to ask for etc 17:44:39 I'm not sure if the current stuff works for that model or not.. it certainly does for Floating IPs 17:45:12 (BTW - I prefer the RAX model when you have unlimited IPs ;)) 17:45:31 Let's park this one, and revisit once timsim or someone can poke at the existing stuff? ;) 17:45:39 Agreed 17:45:48 Can you point us to the v2 endpoint? 17:45:57 https://github.com/openstack/designate/blob/master/designate/api/v2/controllers/floatingips.py 17:46:05 (I had it open ;)) 17:46:07 Ok, that's what I thought 17:46:58 There actually is a missing piece now that I think about it, it's not a required piece, as cleanup will happen eventually, but a sink handler listing to Neutron to activy clear out stale PTRs 17:46:59 ok, I'm good with tabling this for now. 17:47:11 actively* 17:47:15 wow - I can't type at all today. 17:47:55 7 high, 2 medium, 4 low, 1 parked 17:48:32 I'll take a stab at scheduling those after this meet, keep an eye on that etherpad. 17:48:43 ok 17:49:01 kk 17:49:01 (I'll need to try find slots without talk conflicts, other teams we want to kidnap etc etc) 17:49:26 #topic Open Discussion 17:49:30 10 mins left. 17:49:37 Anyone have any other items? 17:50:16 i am good 17:50:20 mugsie already commented on https://review.openstack.org/#/c/180203/ but I'd love any other feedback. 17:50:40 fyi - I will not be able to make it to the summit this time because of some other scheduling conflicts . 17:51:07 vinod1: :( 17:51:10 I'm working on vinod1's replacement… but it just might be me! 17:51:55 Oh, we'll want a leader for each session.. Someone with a vested interest in each topic ;) 17:52:16 * elarson would like to make it but has a new baby to make silly faces at 17:52:31 BP's to present at the work sessions IMO are Optional.. 17:53:08 elarson: Ah, just bring the new baby :) 17:53:13 elarson: thats a terrible exucse :) 17:53:41 kids can learn python at like 2 weeks old, right? 17:53:52 Probably better than any of us by age 4 ;) 17:53:59 <-- Soon to be unemployed 17:54:00 * elarson is focusing on elisp first ;) 17:54:06 Child abuse 17:54:24 Yea, that (elisp) really is.. 17:54:26 Okay, If nobody else has anything else, we can wrap up.. 17:54:28 Thanks all :) 17:54:33 o/ 17:54:34 o/ 17:54:36 o/ 17:54:37 keep an eye on the etherpad 17:54:44 will fill out over the next hour or 2 17:54:48 #endmeeting