15:59:51 #startmeeting neutron lbaas 15:59:51 Meeting started Tue Mar 3 15:59:51 2015 UTC and is due to finish in 60 minutes. The chair is dougwig. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:59:53 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:59:56 o/ 15:59:56 The meeting name has been set to 'neutron_lbaas' 16:00:01 o/ 16:00:02 #topic roll call 16:00:05 o/ 16:00:06 o/ 16:00:06 o/ 16:00:07 o/ 16:00:08 0/ 16:00:14 present 16:00:14 O/ 16:00:33 o/ 16:01:04 o/ 16:01:30 i'm behind, so i didn't type up an agenda, but it's similar to last week. 16:01:33 o/ 16:01:35 #topic Announcements 16:01:40 copied from the neutron meeting: 16:01:43 21:01:47 #info Feature Proposal Freeze: March 5 16:01:43 21:01:55 #info Feature, String, and Dependency Freeze: March 19 16:01:43 21:01:58 #info RCs: April 9-23 16:01:43 21:02:02 #info Kilo release: April 30 16:01:53 i think one has been created 16:02:02 to put this another way, anything not submitted by thursday *will not be in kilo*. 16:02:25 what if it is a 1 line placeholder for an eventual 1000 line patch? 16:02:30 oh, nice. agenda: 16:02:30 https://wiki.openstack.org/wiki/Network/LBaaS 16:02:58 blogan: i'm sure we'll sneak that in. but if you have a new driver, or feature, thursday is it. and gate merges start getting iffy from here on out. 16:03:17 yep 16:03:29 o/ 16:03:46 #topic v2 update, Docs and Tests 16:03:59 o/ 16:04:02 the agent driver got approved yesterday, and we'll be reviewing the tests and TLS hard today. 16:04:07 anyone blocked or need help on anything? 16:04:27 I am trying to get the agent driver running, not having 100% success. 16:04:31 I need help with testing 16:04:40 Hi, I have a question about CI 16:04:47 Not sure if fnaval is here, but I think we got the neutron lbaas v1 scenario test running for v2 (configuration issues out the booty) so we might be able to have that ready for review soon :D 16:05:01 And dougwig I have few questions on tox. 16:05:04 synced with fnaval, but could not get the enough time to sync u 16:05:07 up* 16:05:20 ok, let's do these one at a time. ajmiller, what's your issue? 16:06:01 I set things up in devstack, and things look OK but when I try to create a load balancer it says there aren't any appropriate agents 16:06:14 do you have the lbaasv2 agent running? 16:06:45 It may be better to save this for after the meeting and discuss with smaller group. I'm running the https://review.openstack.org/#/c/158189/ patch 16:07:06 with the devstack mods I posted last night. 16:07:07 ok, let's move that one in channel afterwards. 16:07:09 okay ptoohill or I can help you out with it 16:07:15 madhu: what's up? 16:07:17 Yes, sounds good. 16:07:55 i ran the lbaas v agent with this patch https://review.openstack.org/#/c/158189/, and the tests are not hitting well, but its working for fnaval. Looks strange though 16:08:37 madhu the errors that you pasted looks like you're not pointing to the right neutron_lbaas 16:08:45 it looks like it's using the old neutron_lbaas 1.0 16:09:01 yea, there may be just some config issues here 16:09:12 once the devstack plugin gets setup with the agent, this will become much easier 16:09:14 maybe the new devstack mods help here? 16:09:25 ok, yea, thats not ready then :) 16:09:39 well thats what ajmiller needs help with 16:10:07 okay fnaval and ptoohill, how about having sometime to discuss this oftnoon 16:10:18 So yday, dougwig said that tox -e functional, tox -e api, tox -e tempest, in neutron, are examples of them putting tempest style tests in-tree, and setting up the right scripting/environment for it to work. Neutron support is required 16:10:32 evgenyf: what's up? 16:10:42 Tempest does not test v2 currently so CI should just keep testing v1? 16:10:55 madhu: try to be in the same discussion we'll be having with ajmiller, i bet that'll solve your issues too 16:11:22 sure blogan, looking forward to it 16:12:04 But when i ran tox -e api against neutron, it throws error saying that Neutron support is required.. 16:12:31 dougwig: is it working in your env..? 16:12:32 evgenyf: since we're late on that, a vendor CI that hits v1, or just loads and tests the api tests with the noop driver, will be enough to continue with reviews. 16:13:14 well, this is a neutron problem so we should take it up with them 16:13:25 Aish: those tests require linux and installed devstack, unless you're doing them via devstack-gate. they don't pre-setup their entire environment like the units. 16:13:29 dougwig: Will keep v1 CI listening to neutron-lbaas 16:14:22 dougwig - we had both running but there still was a configuration problem but I don't think w eneed to solve that problem to get ahead with neutron-lbaas 16:14:29 Aish: let's take this offline, can you pastebin me your output/logs ? 16:14:36 xgerman: ok 16:14:44 dougwig: ok.. I will do that 16:14:57 dougwig: Should CI be registered somehow bc it's moving from neutron to neutron-lbaas? 16:15:12 #action dougwig submit non-voting job for v2 tempest tests, to provide a simpler environment for Aish's testing 16:15:39 evgenyf: your vendor CI? if you use zuul, it's just a layout edit. you don't need to do anything official. 16:16:01 dougwig: OK 16:16:03 did i miss anyone's questions/issues, or have we talked about and/or punted all until after? 16:16:17 I have another issue 16:16:31 fyi, grenade is busted because of an lbaas change. all grenade, so rechecks won't help right now. blogan is working on it. 16:16:34 ajmiller: shoot 16:16:37 >< 16:16:48 dougwig that was my issue. 16:17:06 ajmiller: my red bull is working today, if i can answer before you ask. 16:18:04 dougwig: thanks for the grenade update; was wondering why it wasn't passing 16:18:07 #topic "Make subnet_id for lbaas v2 member create optional"? 16:18:15 who's item is this? you have the floor. 16:18:29 mine 16:18:55 we had this patch/bug and never decided if this is a good idea + if it should be done for Kilo 16:19:29 We would need to update the client, api docs, and client docs for this change. 16:19:34 in order to get it in we need changes in docs + CLI so I am just wondering if we should punt it o liberty or decide it;s important enough to move it in 16:19:35 xgerman: what is the reason behind? 16:19:53 well, it's not my patch I am just opposed to have it in Kilo 16:20:16 (for schedulingreasons) 16:20:44 i put the bug in, i thought it was something we had agreed on, plus i think its just better on the end user to not have the subnet_id required on the member 16:21:10 but if we keep it required now and make it optional later, that doesn't break backwards compatibility 16:21:25 The patch xgerman is referring to is https://review.openstack.org/#/c/160460/ 16:21:36 blogan, so subnet will be taken from lb? 16:21:59 yeah, as I said my main worry is that we won't get evryhting done for kilo 16:22:13 isn't subnet something you can just lookup from the member, if it's a nova vm? 16:22:23 evgenyf: essentially it just means the user expects that member ip doesn't need anything extra set up 16:22:35 if the member is a nova vm, but thats not always the case 16:23:39 anyone feel strongly, or can we punt this? 16:24:46 the sound of deafening silence sounds like punting to me. 16:24:49 punt it 16:24:52 +1 16:24:52 #topic Open discussion 16:25:10 Any ETA on the grenade fix? 16:25:10 feel free to discuss the test questions here, or we can close this meeting and switch to channel. 16:25:16 Is there any wiki/notes for CI setup for devstack,neutron-lbaas 16:26:39 Santosh_NS: if you want to use vanilla devstack-gate/jenkins/zuul, my entire config is here: https://github.com/a10networks/neutron-thirdparty-ci 16:27:39 Santosh_NS: this should work also: https://wiki.openstack.org/wiki/Neutron/LBaaS/DeployWithDevstack 16:27:59 however, you'll still need to configure it for 2.0 16:28:57 on a different note, the vpnaas team meeting is merging into the neutron on-demand meeting. that's one input for our "which way to merge" discussion. 16:29:18 I am concerned for neutron-lbaas v2 api tests 16:29:42 dougwig: Q about https://review.openstack.org/153079 16:29:52 should we move the docs into our tree? 16:30:07 dougwig: Yeah, my email to ML didn't go out, so I'm resending, to let team decide, but pretty sure we'll go to on-demand. 16:30:26 dougwig: Only 1-2 people at meetings 16:30:39 xgerman: i'm a fan of anything that can go in-tree, does. 16:30:48 me, too 16:30:59 just checking if you heard about that somewhere else 16:31:28 if not, let's push the envelope. :) 16:31:42 ok, I will advise Aish to do that ;-) 16:32:15 ok, let's close this meeting and move the test related questions to channel. expect blogan to be scarce until the grenade issue is sorted. 16:32:45 #endmeeting