17:00:51 #startmeeting Designate 17:00:52 Meeting started Wed Apr 26 17:00:51 2017 UTC and is due to finish in 60 minutes. The chair is timsim. Information about MeetBot at http://wiki.debian.org/MeetBot. 17:00:53 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 17:00:56 The meeting name has been set to 'designate' 17:00:59 #topic Roll Call 17:01:19 o/ 17:01:28 \o 17:01:33 ping mugsie kiall 17:01:52 o/ 17:02:16 well ok then bye kiall 17:02:26 #topic Bug Triage 17:02:28 https://bugs.launchpad.net/designate/+bug/1686060 17:02:29 Launchpad bug 1686060 in Designate "NSD4 commands fails with new worker model+notify" [Undecided,New] 17:03:01 If that backend does it's own notifies that probably needs to be fixed 17:03:08 yeah 17:03:26 Medium? 17:03:37 sounds right 17:04:35 mmm. I'm actually not sure there's anything broken, it doesn't do that. 17:04:47 It might just be that person didn't have host/port set in their target options 17:05:05 humm 17:05:13 I'll assign it to me to look before I confrim 17:05:52 We don't have anything to backport... 17:05:52 ah - no the host and port are set to the nsd vaules 17:06:03 and worker model was sending notifies to that by default 17:06:04 ohhhhh 17:06:07 yeah 17:06:15 its a worker model probelm 17:06:39 Yeah, or that backend should have different things that host/port for that control stuff, like rndc 17:07:01 ok 17:07:06 yeah - we need to audit the options across backends and make them standard 17:07:11 #topic Open Discussion 17:07:14 s/we/someone/ 17:07:35 hi 17:07:56 the patch for fixing the gate is on the floor https://review.openstack.org/#/c/451569/ 17:08:16 please review it 17:08:48 done 17:08:53 I pin the eventlet version to 0.19.0 and bypass the openstack-requirements gate 17:09:13 OK - then we can work on eventlet 0.21 17:09:20 no 17:09:39 testing 0.21.0 still make UT fail 17:09:39 we can work on fixing designate for 0.21 17:09:47 yes 17:10:03 hieulq_: is it still the nsd test failing with 0.21 ? 17:10:14 mugsie yes 17:10:26 OK. I can take that this evening 17:10:36 that should be small enough 17:10:36 sometime it passes, sometime it fails 17:10:49 * mugsie may remove it - its a terrible unit test 17:11:02 Yeah that's fine by me. 17:11:07 it runs an external command and compares it to a static string 17:11:10 mugsie great :( 17:11:13 We obviously weren't testing that backend well enough to prevent bugs anyway :P 17:11:17 yeah 17:11:49 hieulq_: thank you very much for doing this work 17:11:51 could we mock the nsd4 backend instead of creatong the eventlet fake nsd4 backend? 17:12:07 Yes, we greatly appreciate it. 17:12:12 I think we could. I need to look at the test to be sure 17:12:27 nsd has never been well tested by us :( 17:12:42 we should have, as it is an open source server 17:12:42 if so, I can rewrite the UT for nsd4 to make designate work with eventlet 0.21.0 17:12:54 cool - please feel free to do so 17:13:01 ok, I will try :) 17:13:27 nsd4 is very noisy to test 17:14:20 Alright, anything else to talk about? 17:14:26 there are no question from me 17:15:02 my peer are still working on o.vo for his rolling upgrade work in designate 17:15:13 but look like he can not attend today 17:15:33 OK, we can still itterate in the specs review 17:15:43 thanks :) 17:17:01 o/ 17:17:20 Hi trungnv , we were just wrapping up, anything you'd like to discuss? 17:17:28 yes 17:17:49 We are implementing OVO on Designate project then test these codes with local workstation. then we will implement the items for Rolling upgrade. 17:18:15 Everything are working fine at the moment with our code-base. my colleague who named hieuld will F2F with you at submit about this feature for Designate. 17:18:36 trungnv: do you have any code you can show so we can see what you are trying? 17:18:43 I just talk about our status with OVO at the moment 17:18:49 yes 17:19:16 I am just test with small case with my code in https://github.com/trungnvfet/designate_ovo 17:19:52 that is tests_filters 17:20:40 I am trying to convert whole object into OVO then test them with rolling upgrade 17:20:56 OK 17:21:29 I know that some class in OVO is not suitable with Designate code 17:21:48 I am still working to get solution for some issues at the moment 17:21:52 OK 17:22:15 let us know when you have something to show 17:23:02 yes. every week I will push new code to my github and will discuss with you via IRC 17:23:18 trungnv: is there any of the new code in that repo now? 17:23:45 not yet. I am just test with my repo. 17:24:01 I will push whole code when I complete them. 17:24:26 pushing into gerrit. 17:24:42 OK, if you push small bits to https://github.com/trungnvfet/designate_ovo we can look at it in advance and give advice 17:25:04 https://github.com/trungnvfet/designate_ovo/blob/master/designate/objects/zone.py 17:25:06 yes, thanks. 17:25:31 trungnv the zone object is in o.vo now? 17:25:51 yes. 17:26:39 OK - that will not work (at all) 17:26:46 yes 17:26:51 have you tried to run zone.validate() 17:27:09 not yet 17:27:39 as I said before ovo will not work with our objects - especially if you try to make them OVO objects 17:28:10 you need to separate the versioning from ovo and just use the versioning code 17:28:26 not try and use OVO as is 17:28:59 I will try both. 17:29:24 thanks so much. 17:29:37 I will try and ask o.vo team. 17:30:34 Alright, anything else? 17:30:51 yes 17:31:01 Go ahead 17:31:15 timsim mugsie have you tried devstack desinate with master branch? 17:31:23 heh - no 17:31:32 does it explode ? 17:31:34 I don't use devstack. 17:31:53 ok, another boom .. 17:32:20 I guess new worker model cause all services always receive rpc timeout message 17:32:25 yes. master branch a have big problem now. 17:32:38 If you file a bug with the details it would help 17:32:44 but - the gate does use devstack 17:32:49 so it is passing :/ 17:33:13 yes frankly I dont know why 17:33:28 that's why I dont file the bug 17:33:49 will take a look deeper tomorrow and file a bug if need 17:34:01 thanks, that's all from me 17:34:57 me too. thanks 17:35:12 I am good as well 17:36:13 Just as a general note, re: devstack. Designate only really depends on Keystone, and that's easy enough to run alongside Designate without devstack. You can also turn on the noauth auth provider, and you don't need keystone, just for basic testing of the API. 17:36:25 But for things like the gate, grenade, etc, it's necessary. 17:36:37 Just throwing that out there in case it helps you all. 17:36:52 Ok, see you all in #openstack-dns 17:36:55 #endmeeting