Wednesday, 2014-09-03

*** nkinder has joined #openstack-dns00:03
*** rmoe has quit IRC00:04
*** rmoe has joined #openstack-dns00:18
*** bandarji has joined #openstack-dns00:30
*** jmcbride has joined #openstack-dns01:06
*** jmcbride has quit IRC01:11
*** vinod1 has joined #openstack-dns01:12
*** mwagner_lap has joined #openstack-dns01:14
*** bandarji has quit IRC01:20
*** vinod1 has quit IRC01:28
*** harmw has quit IRC01:38
*** richm has quit IRC01:39
*** rossk has quit IRC01:41
*** harmw has joined #openstack-dns01:45
*** nosnos has joined #openstack-dns01:52
*** jmcbride has joined #openstack-dns01:59
*** msisk has joined #openstack-dns02:07
*** msisk_ has joined #openstack-dns02:09
*** msisk has quit IRC02:11
*** msisk_ has quit IRC02:15
*** nosnos has quit IRC02:16
*** msisk has joined #openstack-dns02:21
*** jmcbride has quit IRC02:22
*** msisk has quit IRC02:27
*** jmcbride has joined #openstack-dns02:31
*** msisk has joined #openstack-dns02:33
*** jmcbride has quit IRC02:35
*** amcrn has quit IRC02:38
*** EricGonczer_ has joined #openstack-dns02:41
*** EricGonczer_ has quit IRC02:43
*** msisk has quit IRC02:49
*** vinod1 has joined #openstack-dns02:53
*** EricGonczer_ has joined #openstack-dns02:56
*** EricGonczer_ has quit IRC03:00
*** EricGonczer_ has joined #openstack-dns03:01
*** EricGonczer_ has quit IRC03:17
*** vinod1 has quit IRC03:23
*** vinod1 has joined #openstack-dns03:25
*** vinod1 has quit IRC03:36
*** vinod1 has joined #openstack-dns03:43
*** vinod1 has quit IRC03:47
*** nosnos has joined #openstack-dns04:50
openstackgerritOpenStack Proposal Bot proposed a change to openstack/designate: Imported Translations from Transifex  https://review.openstack.org/11854406:07
*** k4n0 has joined #openstack-dns06:18
*** ttrumm has joined #openstack-dns08:05
*** HenryG is now known as HenryG_afk12:04
*** RaginBajin has joined #openstack-dns12:09
*** nosnos has quit IRC12:16
*** nosnos has joined #openstack-dns12:17
*** nosnos has quit IRC12:21
*** betsy has joined #openstack-dns12:23
*** RaginBaj_ has joined #openstack-dns12:23
betsykiall/mugsie: y’all around?12:24
*** RaginBajin has quit IRC12:27
*** RaginBaj_ has quit IRC12:27
*** RaginBajin has joined #openstack-dns12:27
Kiallbetsy: yep12:32
Kiallyou're up early!12:32
*** vinod1 has joined #openstack-dns12:35
Kiallbetsy: re https://review.openstack.org/#/c/118436/ - It looks like it covers off the issue, but some tests would be good :) If you're likely to be able to get some before our meet tonight, I'll hold off a +2 :)12:40
vinod1kiall: how about the cases when a server is added/deleted/updated - which changes the NS records - should that be handled later?12:48
KiallThat seems to be covered, since self.add_ns() etc is called, which in turn calls update_recordset12:48
KiallOOO12:49
Kiallit calls update record set with self.update_recordset(context, ns, increment_serial=False)12:49
vinod1precisely :-)12:49
vinod1my comments in the review have some more details on this12:50
betsykiall: Sure. I’ll add some tests12:50
betsyMy question was whether the change in server - add/delete/update should be making the call to increment update or just let the NS do it12:51
KiallI think it should let update_recordset do it, since it calls in there already12:52
betsyok.12:52
betsyI’ll change that and add some tests12:52
openstackgerritEndre Karlson proposed a change to openstack/designate: Support the same api versions format as keystone  https://review.openstack.org/11862512:57
ekarlsoany thoughts on ^12:58
Kiallother than the weirdness of `base` being referenced on L30, but declared on L35 ;)12:59
Kiall(Yes, I do actually think that will work, but it's still weird as hell!12:59
ekarlsooh13:01
ekarlsodon't know how that worked :113:01
*** richm has joined #openstack-dns13:02
*** vinod1 has quit IRC13:02
KiallI *think* it's a quirk of how inline functions are scoped13:02
KiallBut I've never actually tried it, since, it's weird as hell ;)13:02
ekarlsouploaded Kiall13:02
openstackgerritEndre Karlson proposed a change to openstack/designate: Support the same api versions format as keystone  https://review.openstack.org/11862513:02
*** nkinder has quit IRC13:11
betsykiall: I’ve got tests for create soa, update soa, create ns, update ns and I added some code to make sure that the serial number is incremented. Any other tests I should add?13:12
KiallStandard Create/Update/Delete recordset calls too?13:12
*** vinod1 has joined #openstack-dns13:13
betsyWhat do you mean?13:14
betsyWhen do soa and ns recordsets get deleted? When the zone is deleted?13:15
betsyDo we currently delete records when zones are deleted?13:17
betsyand recordsets^13:17
vinod1I am assuming Kiall meant that when Standard Create/Update/Delete recordset calls are called - is the serial # updated?13:18
betsyOh, duh.13:18
betsyI guess I’m still half asleep13:18
KiallYep - Sorry, back and forth from RDP sessions -> Desktop -> RDP sessions -> Desktop ;) Hard to notice chat messages13:18
betsyno problem13:19
betsykiall was right. It is too early for me. :D13:19
*** eankutse has joined #openstack-dns13:20
*** EricGonczer_ has joined #openstack-dns13:44
*** vinod1 has quit IRC13:52
openstackgerritRon Rickard proposed a change to openstack/designate-specs: Server Pools - Storage  https://review.openstack.org/11344713:54
*** ttrumm_ has joined #openstack-dns13:54
*** ttrumm has quit IRC13:58
*** nkinder has joined #openstack-dns13:58
*** ttrumm_ has quit IRC14:05
*** jmcbride has joined #openstack-dns14:18
openstackgerritBetsy Luzader proposed a change to openstack/designate:  Creating a Recordset doesn't increment the serial number  https://review.openstack.org/11843614:26
Kiallekarlso: some comments on https://review.openstack.org/#/c/118625/14:27
*** paul_glass has joined #openstack-dns14:27
*** paul_glass has quit IRC14:28
*** paul_glass has joined #openstack-dns14:28
*** HenryG_afk is now known as HenryG14:36
*** vinod1 has joined #openstack-dns14:43
*** betsy has quit IRC14:48
*** timsim has joined #openstack-dns14:52
*** msisk has joined #openstack-dns14:56
*** betsy has joined #openstack-dns14:58
*** vinod1 has quit IRC15:02
*** jmcbride has quit IRC15:02
*** vinod1 has joined #openstack-dns15:03
*** paul_glass1 has joined #openstack-dns15:04
*** paul_glass has quit IRC15:04
*** jmcbride has joined #openstack-dns15:07
ekarlsols15:09
*** jmcbride has quit IRC15:11
openstackgerritEndre Karlson proposed a change to openstack/designate: Support the same api versions format as keystone  https://review.openstack.org/11862515:12
*** msisk has left #openstack-dns15:20
openstackgerritBetsy Luzader proposed a change to openstack/designate:  Creating a Recordset doesn't increment the serial number  https://review.openstack.org/11843615:48
*** rmoe has quit IRC15:48
*** rmoe has joined #openstack-dns16:05
*** paul_glass1 has quit IRC16:05
*** simonmcc____ has joined #openstack-dns16:06
*** CaptTofu__ has quit IRC16:08
*** k4n0 has quit IRC16:08
*** simonmcc___ has quit IRC16:08
*** k4n0_ has joined #openstack-dns16:09
*** CaptTofu__ has joined #openstack-dns16:09
*** amcrn has joined #openstack-dns16:10
*** simonmcc____ is now known as simonmcc___16:12
*** mwagner_lap has quit IRC16:22
*** diga has joined #openstack-dns16:53
*** vinod1 has quit IRC16:54
*** rjrjr_ has joined #openstack-dns17:00
*** diga has quit IRC17:01
*** vinod1 has joined #openstack-dns17:01
*** amcrn_ has joined #openstack-dns17:02
*** paul_glass has joined #openstack-dns17:03
*** rossk has joined #openstack-dns17:05
*** amcrn has quit IRC17:06
*** amcrn_ is now known as amcrn17:06
*** openstack583 has joined #openstack-dns17:08
*** jmcbride has joined #openstack-dns17:15
*** dtx00ff has joined #openstack-dns17:15
*** dtx00ff is now known as darshan10417:15
*** mwagner_lap has joined #openstack-dns17:16
*** openstack583 has quit IRC17:23
*** diga has joined #openstack-dns17:27
*** darshan104 has quit IRC17:45
rjrjr_timsim?18:04
vinod1rjrjr_: timsim stepped out for lunch18:05
vinod1But we were wondering if need help with the pool manager stuff18:06
rjrjr_i'm planning on finishing the spec work today.  it has a dependency on the minidns spec.  i'd like to start coding too, but was waiting to hear how we would commit our code.  WIPs will work for now.18:09
rjrjr_which piece of pool manager do you want?18:09
rjrjr_also wanted to discuss the performance issues timsim brought up a few weeks back.  i didn't make last weeks meeting, so i don't know how that turned out.18:11
rjrjr_(i think it was timsim that brought them up.)18:11
vinod1For the perf issues, we were waiting for the SOA and AXFR patches to land, to run the tests18:12
vinod1We will be running the tests and coming up with the numbers and then plan to revisit the discussion18:12
rjrjr_i don't mind sharing the server pool work.  then i can work on some other specs i've been wanting to work on.18:13
rjrjr_just wondering what you are interested in.18:13
vinod1I have no preference - but just wanted to get the serverpools work done soon :-)18:14
vinod1Initially I was planning to work on the minidns serverpools code18:14
vinod1The first thing though needed though might be the pool manager service18:15
rjrjr_storage.18:15
rjrjr_did the API spec get done?18:15
vinod1betsy is working on that18:15
vinod1in what order were you planning to work on the server pool items?18:16
rjrjr_so, storage is a foundation for the other work.  that is how i wrote the specs.18:16
vinod1we could pick up something that is lower on your list18:17
rjrjr_okay.  i'll finish the spec today and you should talk and let me know what you want to work on.  i'm not picky.18:18
vinod1okay we will chat again with you in a day or 2 once the specs are approved18:19
*** paul_glass has quit IRC18:22
openstackgerritGraham Hayes proposed a change to openstack/designate: warn against sorting requirements  https://review.openstack.org/11872318:25
*** darshan104 has joined #openstack-dns18:30
vinod1mugsie: is your change for j3?18:30
mugsievinod1: doesnt have to be18:32
*** darshan104 has quit IRC18:32
mugsietbh, no18:32
ekarlsomugsie: where's kiall hiding18:34
ekarlsowaiting for the client release18:34
*** diga has quit IRC18:34
*** darshan104 has joined #openstack-dns18:41
*** darshan104 has quit IRC18:42
*** darshan104 has joined #openstack-dns18:42
*** richm has quit IRC18:49
*** amcrn_ has joined #openstack-dns18:51
*** amcrn has quit IRC18:54
*** paul_glass has joined #openstack-dns19:02
*** richm has joined #openstack-dns19:03
*** vinod1 has quit IRC19:11
*** RaginBajin has quit IRC19:28
openstackgerritRon Rickard proposed a change to openstack/designate-specs: Server Pools - Manager  https://review.openstack.org/11346219:43
openstackgerritRon Rickard proposed a change to openstack/designate-specs: Server Pools - Storage  https://review.openstack.org/11344719:45
rjrjr_not done with the Pool Manager specs yet.  just was testing something.  should have it shortly.19:46
betsymugsie: Yeah. Looks like Remove the priority field for v2 (https://review.openstack.org/#/c/108797/) needs to go in j-3. It wasn’t in j-219:50
*** vinod1 has joined #openstack-dns19:50
mugsiebetsy: was there a bp filed for it?19:51
betsyyes.19:51
betsyhttps://blueprints.launchpad.net/designate/+spec/remove-priority-field19:51
betsyLooks like the bp never got  marked as approved19:52
mugsiecool - moved it19:52
betsyThx!19:52
*** jmcbride has quit IRC20:00
openstackgerritA change was merged to openstack/designate: Imported Translations from Transifex  https://review.openstack.org/11854420:02
Kiallekarlso: client v1.1.0 tagged20:06
KiallIt'll be on pypi whenever openstack infra get's around to building it20:07
*** jmcbride has joined #openstack-dns20:09
Kialljmcbride: ... back in town? ;)20:09
Kiallekarlso: done.. https://pypi.python.org/pypi/python-designateclient20:10
*** tmc3inphilly has joined #openstack-dns20:11
openstackgerritA change was merged to openstack/designate:  Creating a Recordset doesn't increment the serial number  https://review.openstack.org/11843620:16
*** eankutse has quit IRC20:19
openstackgerritA change was merged to openstack/designate: Support the same api versions format as keystone  https://review.openstack.org/11862520:31
*** mwagner_lap has quit IRC20:34
rjrjr_kiall/mugsie: i think most of the comments I'm seeing in the two specs (storage and pool manager) are a product of having 2 specs.20:35
rjrjr_understandably, having 2 specs helps us break out the work, but the storage spec is heavy on concept and pre-work implementation, but light on using that pre-work.20:37
rjrjr_the pool manager spec piggy backs on that storage spec immensely.20:37
rjrjr_instead of breaking down the work in the manner i did, maybe i can divide the work between the specs in a better manner.20:38
rjrjr_what is the goal of the 'storage' spec exactly?  i understand the need for the pool manager spec, but the storage spec is very nebulous.20:39
rjrjr_i'm open to suggestions from anyone on this, not just kiall or mugsie.20:39
timsimrjrjr_: I think the goal is that each patchset has a spec associated that tells what the code is supposed to accomplish. Since this is a rather complicated thing, each spec (and as it follows, patchset) will do a small piece. If it's broken down well, everyone can review the tiny parts of the spec to make the coding bit easier.20:42
rjrjr_timsim, i understand.  so, how would you break the work down?  i feel i have most of the work encapsulated in the two specs and am adding the missing pieces and tweaking.  i'm concerned that i'm working on comments that would not exist if the two specs were taken as a whole.20:44
rjrjr_for example, in the storage spec, i define the server configuration which will be used to create the server objects.  but i have comments in the pool manager spec asking about this.  it's there, but in the storage spec.20:45
rjrjr_i think the problem here is breaking this down as 'storage' and 'pool manager'.  to me, it all is 'pool manager'.20:45
rjrjr_so, i'm reading these comments as nitpicking, when they probably aren't.  it is just the breakdown is unclear.20:46
timsimTo me storage should just be the actual database code. So what new tables are you making, and why?20:46
rjrjr_not the objects that go with the database code?20:47
timsimSure, why not.20:47
rjrjr_so, by extension, i should be created the server object in which spec?  i did it in the storage spec.20:47
timsimThat's fine, imo.20:48
rjrjr_remember, a server is based on the configuration file, not a database table per our seattle discussion.20:48
timsimRight, but the actual creation of those objects happens from the Pool Manager.20:48
rjrjr_okay, so the breakdown i have is correct.  i'll just add comments to have the storage spec referenced for more information then.20:49
timsimThe biggest part is probably these complicated processes.20:51
rjrjr_timsim: quick question, you had a performance concern with Designate and listed some solutions.  how did that conversation end?20:51
rjrjr_i was concerned about having a databases that spanned regions/geographies and wondered if we couldn't solve the performance issue by distributing the work in Designate.20:52
timsimWe probably need to define some sort of framework for our specs that answer key questions like: "How does a Server object get it's information" instead of trying to fit it into "Central Changes" or something like that.20:52
rjrjr_these are the templates i was given. 8^(20:53
timsimBasically we're concerned with MiniDNS' ability in many different regions to work with the Designate database in a scalable fashion.20:53
timsimMight have to write some new templates.20:53
rjrjr_it is hard to have a conversation in 1 hour about these things, so let me try and explain what i was thinking.  rather than have the database for all records in all places, couldn't we have Designate communicate with other Designate instances to avoid that?20:54
rjrjr_i know there are things you can do in BIND for example, to have BIND send dynamic update requests to other servers.20:54
timsimSo you'd have the request for information go over RabbitMQ, rather than a database connection?20:55
rjrjr_so, you can have a slave BIND server get a dynamic update request and the slave turns around and send the request to the authoritative BIND master.20:55
rjrjr_or through minidns.20:56
rjrjr_or a REST API call.20:56
rjrjr_i don't have the details worked out, but i was thinking "distribute" the work like DNS distributes the work.20:56
timsimI think the issue there is that the database is the source of truth. So you have to go there if you want to get the right information.20:57
*** tmc3inphilly has quit IRC20:57
rjrjr_correct.   a designate server gets a request for a zone it doesn't have.  it sens the request to the designate server that does handle it.20:57
*** jmcbride has quit IRC20:58
rjrjr_we are already going to have designate pools that divide the zones among DNS servers.  have the accompany pool/minidns be responsible for the zones.20:58
rjrjr_(responsible = queries, updates, etc.)20:58
rjrjr_i'm hoping i'm being clear here.  again, not heavy on details, but wondering if the concept makes sense and would help with the issue.20:59
timsimIf you can split up the zones in different pools, you could have endpoints that only know about certain zones, then that might be possible. But if you need all of your zones in a single pool, or at least a lot of them, I don't really see a way around that.20:59
timsimShort of having the same zones in different pools, which would be tricky.21:00
rjrjr_i'm very concerned about a database that contains all the records for every zone and that database is distributed everywhere you need Designate.  especially if some of those zones are not used in a particular region/geographic location.21:00
rjrjr_which i believe is the issue you are addressing.21:01
timsimPretty much, except all of the zones would be in every region/geo location.21:02
rjrjr_maybe we have "slave" Designate servers that work on NOTIFY requests to keep local databases in sync.21:02
rjrjr_that is how DNS does it.21:02
timsimThat kind of sounds like our "agent" idea plus a cache.21:03
*** jmcbride has joined #openstack-dns21:03
rjrjr_i'm concerned about the feasibility of using DB technology for the problem.  my experience with those technologies has not been good.21:03
rjrjr_small databases, great.  geographically close databases, good.  geographically far databases, not so good.21:04
rjrjr_i'll have to relook at your proposals.  it's been a few weeks.21:05
timsimYeah, that's pretty much what our Ops folks said. Thus, the issue. They don't think we can scale the database like that.21:05
rjrjr_i agree.  i like the idea of not having all the records in a single database and having Designate instances communicate with each other.21:06
rjrjr_i know your problem is on a larger scale because of the service your organization provides.21:07
timsimBut surely that would be worse than trying to replicate the database.21:07
*** jmcbride has quit IRC21:07
timsimBecause eventually the information has to get to the same place, so things would just be waiting for even longer than if they were trying to access a database across the ocean.21:08
rjrjr_for us, i'd place zones close to where they will be used.  for queries, that is why we have slaves.  for updates, i can see a problem, but we don't have the need to have every zone everywhere.21:09
rjrjr_and the few "remote" updates that are needed would be an exception, not the norm.21:09
rjrjr_interesting problem.  i'll read what you wrote and am interested in hearing more about how to solve it.21:09
timsimCool.21:10
rjrjr_back to the specs...21:10
rjrjr_i appreciate your thoughts on the specs.21:11
openstackgerritVinod Mangalpally proposed a change to openstack/designate-specs: Server Pools - MiniDNS support  https://review.openstack.org/11268821:19
*** EricGonczer_ has quit IRC21:20
*** vinod1 has quit IRC21:22
openstackgerritRon Rickard proposed a change to openstack/designate-specs: Server Pools - Storage  https://review.openstack.org/11344721:27
*** jmcbride has joined #openstack-dns21:29
*** jmcbride has quit IRC21:33
*** paul_glass has quit IRC21:49
*** timsim has quit IRC22:01
*** nkinder has quit IRC22:10
*** openstack has joined #openstack-dns22:10
*** ChanServ sets mode: +v openstack22:10
*** msisk has joined #openstack-dns22:17
*** nkinder has joined #openstack-dns22:26
*** msisk has quit IRC22:32
*** nkinder has quit IRC22:53
*** openstackstatus has quit IRC23:19
*** openstackstatus has joined #openstack-dns23:20
*** ChanServ sets mode: +v openstackstatus23:20
*** jmcbride has joined #openstack-dns23:27
*** mwagner_lap has joined #openstack-dns23:45
*** nkinder has joined #openstack-dns23:51
*** jmcbride has quit IRC23:55
*** vinod1 has joined #openstack-dns23:58
openstackgerritA change was merged to openstack/designate: warn against sorting requirements  https://review.openstack.org/11872323:58

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!