17:01:38 <Kiall> #startmeeting Designate 17:01:38 <openstack> Meeting started Wed Dec 3 17:01:38 2014 UTC and is due to finish in 60 minutes. The chair is Kiall. Information about MeetBot at http://wiki.debian.org/MeetBot. 17:01:39 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 17:01:41 <openstack> The meeting name has been set to 'designate' 17:01:49 <Kiall> Heya - Who's about? 17:01:51 <rjrjr_> o/ 17:01:56 <timsim> o/ 17:02:01 <mugsie> o/ 17:02:23 <Kiall> vinod1 / betsy about today? 17:02:39 <Kiall> Guess not :) 17:03:05 <timsim> They'll be along shortly I bet 17:03:06 <Kiall> Quick off-agenda item - Kilo-1 status page .. Anything we're missing from it, please add https://launchpad.net/designate/+milestone/kilo-1 17:03:09 <betsy> yes 17:03:10 <betsy> o/ 17:03:21 <Kiall> #topic Actions Items from last week 17:03:53 <Kiall> I'll skip a few of these that are duplicates ;) 17:03:53 <Kiall> timsim file bug for PUT behavior in the API - Done 17:03:53 <Kiall> Kiall give timsim mockup code for totalcount - Done 17:04:00 <Kiall> mugsie push horzion code upstream - Done - Needs a +A 17:04:10 <rjrjr_> transition powerdns backend to MDNS. - is that needed? 17:04:13 <Kiall> people review ekarlso- 's secondary zones / v2 API bindings - Any takers on that one? 17:04:22 <Kiall> rjrjr_: I think it needs to be renamed to pools :) 17:04:28 <vinod1> I am looking at the secondary zones 17:04:32 <timsim> I reviewed secondary zones, I think vinod did/is in progress 17:04:34 <mugsie> ^ is just an initial push of internal code, so unless there is glaring errors, can we get it merged? 17:04:39 <Kiall> Anyway - We can discuss those in detail next week, I'll be putting Kilo release status on 17:04:48 <Kiall> the agenda from then on 17:05:03 <vinod1> i can do a quick scan of the horizon code today 17:05:06 <Kiall> Just want people to fill out any missing stuff before next week :) 17:05:23 <Kiall> vinod1: excellent :) 17:05:40 <Kiall> Okay - Moving on! 17:05:41 <Kiall> #topic Pools - Where are we? (kiall) 17:05:53 <Kiall> So - We've got lots merged, where do we go from here? :) 17:06:15 <betsy> I’ve still got my work 17:06:17 <mugsie> i think testing multiple pools, (even with manual switching) and seeing what breaks 17:06:24 <betsy> Almost done just getting the tests working 17:06:35 <mugsie> betsy: cool 17:06:42 <rjrjr_> i'm in the middle of a integrating pool manager with central without going through the backend proxy. 17:06:50 <Kiall> mugsie: ++ 17:06:53 <betsy> Afraid there’s going to be crazy merge issues 17:07:03 <mugsie> betsy: there always is 17:07:13 <Kiall> Beyond that - We'll need to port over the backends (PowerDNS, Fake, NSD4 etc. We'll discuss IPA style in the next topic) 17:07:17 <betsy> :) 17:07:33 <Kiall> I've started PowerDNS - But hit a blocker and it was late ;) 17:07:47 <rjrjr_> we also need to decide how to display via CLI/API the status and action. 17:07:48 <Kiall> Do we have any takers for the other easily portable backends? 17:08:15 <mugsie> i will have a look at nsd 17:08:23 <rjrjr_> i handled fake already. 8^) 17:08:38 <timsim> rjrjr_: You're doing bind9 too or no? 17:08:47 <Kiall> Easily portable: Fake (done), NSD4 (mugsie), PowerDNS (me), DynECT 17:09:05 <rjrjr_> bind9 is done. 17:09:08 <Kiall> I'm guessing myself / ekarlso- / mugsie will need to do DynECT due to having an account with them 17:09:13 <timsim> Cool. 17:09:20 <Kiall> I'll ask ekarlso- to, as he wrote it the first time ;) 17:09:46 <Kiall> Okay .. During porting - I'd suggest we replace the old backend, rather than add a new _pool version. Thoughts? 17:09:57 <vinod1> +1 17:10:09 <rjrjr_> i think we should wait until we have pool manager integrated completely. 17:10:34 <Kiall> well - there's a catch-22 there of can't kill the old code without breaking all the old-style backends 17:10:39 <rjrjr_> never mind. you are correct. go for it. 17:11:06 <vinod1> rjrjr_: With your current patchset - would the pool manger integration be complete? 17:11:19 <rjrjr_> when it is completed, yes. 17:11:22 <Kiall> okay - so, beyond those ports (and IPA style backends, which we'll come to in the next topic). API chanegs to expose the status etc was suggested as a next step? 17:11:42 <rjrjr_> i have questions about unit testing, but i'll wait until the open discussion. 17:12:15 <Kiall> kk.. I guess that leads into the next topic then.. Any more general pools status / progress / questions? 17:12:35 <vinod1> Kiall - your code for recordsets seems to be propagating to pool attributes and now domain attributes 17:12:49 <vinod1> so we need to get that cleaned up 17:13:01 <Kiall> vinod1: ah yes, timsim filed a bug for that.. 17:13:17 <Kiall> Do we have any takers on it? 17:13:26 <mugsie> whats the bug? 17:13:31 <Kiall> (I can't remember if tim assigned himself, or just filed it) 17:13:41 <timsim> Nah I didn't assign it to myself. 17:13:43 <Kiall> bug 1396720 17:13:47 <uvirtbot> Launchpad bug 1396720 in designate "PUT on a Recordset with multiple Records results in delete/recreation of all Records" [High,Triaged] https://launchpad.net/bugs/1396720 17:14:10 <mugsie> I can look at it 17:14:19 <mugsie> havent done much upstream workn this month ;) 17:14:26 <Kiall> Thanks mugsie .. 17:14:36 <rjrjr_> i hope we are not making any big changes to central until i get this patch in. just a wish really. 8^) 17:14:45 <Kiall> rjrjr_: lol... 17:15:01 <mugsie> rjrjr_: I am going to change everyhthing 17:15:05 <Kiall> It should be an API layer fix ;) 17:15:08 <mugsie> just for the fun of it ;) 17:15:11 <Kiall> Okay.. Moving on :) 17:15:18 <Kiall> #topic Pools - Removing the old backend code, migration path for IPA style backends? (kiall) 17:15:53 <Kiall> rjrjr_: patchset has started this, but myself and mugsie started talking about it last night.. 17:16:31 <Kiall> We're thinking we really need to provide the migration path for IPA style backends (also applies to the InfoBlox backend which should arrive soon) 17:16:33 <rjrjr_> based on the conversation yesterday, it seems like we will drop support for IPA in Kilo and add it back in the next cycle. 8^) 17:17:00 <mugsie> I think that we can avoid that - hopefully 17:17:03 <timsim> What does that entail? 17:17:03 <Kiall> mugsie managed to change my mind on that, so wanted to discuss with the group and come to a decision. 17:17:23 <Kiall> We figure it kinda fits into timsim's new-agent proposal .. 17:17:35 <mugsie> I would think that we can re purpose the "new agent", by extending it slightly 17:18:07 <Kiall> If we wanted old-style backends to work as is.. Agent listens for NOTIFY, does AXFR (i.e copy the secondary zone code), then splits that into a backend - for bind - that would just re-render the zonefile and rndc reload.. 17:18:23 <timsim> What extra work is there for that backend for IPA style things? 17:18:29 <Kiall> For IPA, it might do an AXFR against IPA, compare, and call the old-style backend methods to write our the changes 17:18:50 <timsim> Seems doable. 17:18:56 <mugsie> yeah, i think so 17:19:10 <Kiall> Yea, I think it sounds doable at 1000 ft alright :) 17:19:24 <mugsie> i just don't want to dump some of the backends, with out an attempt to help them work 17:19:28 <betsy> kiall: It always does :D 17:19:38 <rjrjr_> so, what does that mean for my current work? 17:19:51 <rjrjr_> once it is accepted, IPA is broken. 8^( 17:20:10 <rjrjr_> the timing here is going to be tricky. 17:20:14 <mugsie> i think we can get ^ done fairly quickly 17:20:17 <Kiall> Anyway, If we want to ensure we keep that support - then getting that work (work == bones of new-agent) done becomes a priority... thoughts? 17:20:44 <timsim> I'd agree. 17:20:44 <Kiall> rjrjr_: yea, that's where I'm hoping we come to a decision and get the bones written up really quickly so as to not block you for too long 17:21:00 <rjrjr_> i'm concerned, but that's because i see what still needs to happen with pool manager. 17:21:00 <mugsie> I would like to make it a prioroty 17:21:12 <timsim> (The Agent is a rather large RAX priority, so it looks like I was going to be working on it anyway) 17:21:36 <Kiall> rjrjr_: Yea, finding the balance is the fun one ;) 17:21:46 <Kiall> #link https://wiki.openstack.org/wiki/Kilo_Release_Schedule 17:21:47 <mugsie> i think that we can have 2 groups working in tandem, and then we can do some git foo to have the switch done right 17:22:00 <timsim> mugsie: We can do a topic branch! oh wait... 17:22:07 <Kiall> k1 is Dec 18th, and we've been hoping for Pools to be effectively feature complete by then.. 17:22:31 <mugsie> timsim: funnily, neutron has topic branches 17:22:33 <mugsie> :D 17:22:59 <Kiall> which is 431 commits behind master ;) 17:23:03 <timsim> Kiall: So would you want to wait on merging any "agent" stuff until after k1? 17:23:05 <Kiall> Gonna be a fun rebase! 17:23:06 <mugsie> I think we can look a tthe spec for the agent, propose changes today / tomorrow, and get work started quickly 17:23:27 <rjrjr_> but, pool manager implies broken IPA without agent. 17:23:29 <mugsie> for K1 the agent doesnt have to pretty or quick 17:23:35 <Kiall> timsim: well, I dunno.. Do we break IPA style for K1, and fix it in K2? or not have 100% pools in K1? 17:23:38 <mugsie> have to be* 17:23:57 <rjrjr_> i think pools should be the priority. but i'm biased. 8^) 17:24:12 <Kiall> I think I'm OK with breaking IPA style in K1, assuming we commit to fixing it in K2 17:24:12 <timsim> Kiall: Might be a stretch to get the agent+IPA business done by K1 17:24:23 <mugsie> i think we can postpone that call till next week. if next week there is not much progress, we break IPA in K1 17:24:31 <timsim> I think that seems reasonable. 17:25:09 <timsim> So action people to review the agent spec then? 17:25:09 <rjrjr_> i'm okay with that. 17:25:15 <mugsie> yup 17:25:18 <Kiall> Any other opinions or shall we call it break IPA-style in K1,commit to fixing in K2, unless we can agree next week that Pools+Agent is doable by K1? 17:25:34 <Kiall> #action all review new-agent spec 17:25:42 <mugsie> aim to keep it, and decide next week 17:25:49 <timsim> mugsie: +1 17:25:52 <Kiall> mugsie: yea, aim to keep, see how reality based that is 17:26:05 <timsim> #link https://review.openstack.org/#/c/131530/2 Agent spec 17:26:18 <rjrjr_> this is going to be fun... 17:26:24 <mugsie> rjrjr_: :) 17:26:27 <Kiall> rjrjr_: yea, sorry :( 17:26:43 <Kiall> On the plus side, you'll learn more about git rebase :D 17:26:48 * Kiall ducks for cover 17:26:53 <timsim> You're going to be a rebase master rjrjr_ :P 17:26:54 <rjrjr_> yeah, there is that. 17:27:01 <Kiall> Okay - Anything else on this before Open Discussion 17:27:05 <Kiall> ?* 17:27:08 <mugsie> I am good. 17:27:13 <timsim> I'm good. 17:27:42 <Kiall> #topic Open Discussion 17:27:46 <rjrjr_> mid-cycle summit - will rackspace be able to attend in San Jose? 17:27:50 <vinod1> rjrjr: Just wanted to check if you had any inputs on the horizon code 17:27:52 <Kiall> Any other topics? rjrjr_, you had one? 17:28:11 <vinod1> since ebay was interested in horizon 17:28:24 <rjrjr_> i handed the code to PK. he is in the process of getting it up and running. i'll check with him today and get back to you about it. 17:28:37 <rjrjr_> mid-cycle summit? 17:28:41 <Kiall> rjrjr_: if he's stick, ping us :) 17:28:50 <Kiall> stuck* 17:28:55 <mugsie> yeah, I have been talkig to him 17:28:58 <rjrjr_> sure. stick or stuck. 17:28:59 <Kiall> kk 17:29:39 <vinod1> for the mid-cycle summit - i still haven't heard back anything from the higher ups 17:29:51 <rjrjr_> okay. 17:30:02 <Kiall> If I remember right - Joe had said it should be fine, but official approval etc etc 17:30:08 <vinod1> correct 17:30:11 <timsim> Yeah, that's what it looks like. 17:30:18 <rjrjr_> can we get a go/no go by 12/24? 17:30:35 <timsim> We can try 17:30:46 <Kiall> It's a "go" from HP anyway... 17:30:52 <rjrjr_> cool! 17:30:57 <timsim> We'll sick Joe on it :P 17:31:06 <rjrjr_> okay, now a question about unit testing... 17:31:24 <rjrjr_> so, if i'm understanding what i'm seeing, i need to have central, pool manager, and mdns running. 17:31:33 <rjrjr_> pool manager will use the fake backend i wrote. 17:31:46 <rjrjr_> mdns needs fake stuff too, correct? 17:31:48 <Kiall> Ideally, we fix Pool Mgr's shiney new tests to, well, not need all that 17:32:24 <rjrjr_> the problem is central is failing miserably on tests. 17:33:03 <rjrjr_> ideally, mdns would have fake calls for NOTIFY, AXFR, etc. 17:33:15 <rjrjr_> since we won't really have a backend. 17:33:28 <Kiall> If I remember right, all central calls -> pools are cast - i.e. fire and forget - those can be mocked easily, and we can just validate the right stuff has been passed into the call 17:33:54 <rjrjr_> so, a mock pool manager then? do we have an example of a mock service? 17:34:08 <Kiall> Well, not a full mock... you can do something like this.. 17:34:14 <rjrjr_> i'm new to Python unit testing, so any guidance you can provide is appreciated. 17:34:30 <Kiall> with mock.patched(pool_manager_api, 'some_method') as m: 17:34:56 <Kiall> then when central calls "some_method" it'll be just record the args and return nothing (like the real call would) 17:35:17 <Kiall> and m.calls[0].args or similar will let you validate the correct things are passed in 17:36:26 <rjrjr_> okay. this is a learning curve for me. this might take me a couple of days. 17:36:51 <Kiall> https://github.com/openstack/designate/blob/master/designate/tests/test_mdns/test_notify.py#L37 17:37:00 <Kiall> ^ Similar example from mDNS code 17:37:28 <Kiall> In that case, patching the dns.query.udp() method 17:37:35 <Kiall> We can sync up after the meet if you like? 17:37:49 <rjrjr_> okay. 17:38:26 <Kiall> Cool - Any other topics? 17:38:37 <timsim> I'm good. 17:38:47 <Kiall> mugsie / vinod1 / betsy ? 17:38:51 <mugsie> nopoe 17:38:52 <vinod1> Nothing from me 17:39:21 <betsy> I’m good 17:39:24 <Kiall> Cool, will call it a day then :) 17:39:47 <Kiall> rjrjr_: give me 10 mins and I'll ping you if your free then... 17:39:47 <Kiall> Thanks all :) 17:39:47 <Kiall> #endmeeting