16:01:43 #startmeeting ironic_neutron 16:01:44 Meeting started Mon Jul 27 16:01:43 2015 UTC and is due to finish in 60 minutes. The chair is Sukhdev. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:01:46 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:01:48 The meeting name has been set to 'ironic_neutron' 16:01:58 #topic: Agenda 16:02:03 #link: https://wiki.openstack.org/wiki/Meetings/Ironic-neutron 16:02:37 Good morning everybody 16:02:40 Anybody like to add/change anything on agenda? 16:03:00 #topic: Announcements 16:03:29 I moved our approved on the top of the wiki - you may have noticed 16:04:03 reminder about the Ironic mid-cycle coming up shortly 16:04:38 jroll: want to update us on your nova mid-cycle? 16:05:08 Sukhdev: so, we spoke briefly about my nova patches, the nova team still seems to think it isn't a problem, no real surprises there 16:05:13 I didn't however have time to hack on it there 16:06:02 jroll: good - as long as we have higher level understanding - cleaning up/fixing code should not be that big of deal - do you agree? 16:06:09 yes 16:06:17 cool - thanks 16:06:28 anybody has any announcement? 16:06:55 OK - lets move on then 16:07:03 #topic: Specs under review 16:07:26 I did not have anything to add here - does anybody have anything to share? 16:08:02 * Sukhdev waiting 16:08:09 thanks everyone for reviewing the patch last week to get the ironic-neutron spec merged in 16:08:55 lauramoore: you got 4 +2 - that is really cool :-):-) 16:09:46 anything on the patches? 16:10:00 #topic: Testing 16:10:12 We did not get to talk much about testing last week 16:10:43 I was hoping this morning we can do some brain storming about the test strategy here 16:11:14 looks like amtoki is not here - he had some ideas about it - but, we ran out of time 16:11:24 so basically, I think what we need is an OVS ML2 thing, that behaves like a bare metal ML2 thing (e.g. requires switchport info etc) 16:12:35 jroll: I looked around - we do not have anything. We can create a simple ML2 driver for this 16:12:44 Sukhdev: yeah, that's what I figured 16:13:07 jroll: so, there are really two parts - 16:13:38 1) to verify that the correct information gets passed/consumed between Ironic/Neutron 16:14:28 2) to verify that the instance lands on correct network during the pre-boot and post-boot cycles (i.e. network flip) 16:15:18 right, so I think 1 will be covered by making the OVS thing require the right data 16:15:47 correct - 16:15:55 2 can be covered by some ping checks, etc, maybe even make sure provisioning network is completely firewalled from tenant net 16:17:15 I am not sure if I understood correctly the firewall part - how does that help with the network flip logic 16:18:28 essentially, we want to make sure that server is attached to prov. network in the early stage and then latter flipped to tenant n/w and is able to ping 16:18:42 so 16:18:59 if the two networks are firewalled, and the control plane is in prov network 16:19:20 if the instance is deployed successfully, we'll know it started in prov network 16:19:35 if it pings from the tenant network after deployment, we'll know if flipped properly 16:20:03 Ah I see now - 16:20:23 pretend like it's the real world :) 16:20:23 I think that works - 16:20:37 So, this is really a true functional test 16:20:46 yes 16:21:16 I do not believe we have anything in place that we could use to build this upon - unless somebody know 16:21:25 s/know/knows 16:22:27 I mean, we could build it on the current ironic functional jobs 16:22:38 configure that to use this new OVS thing, add the firewall 16:22:43 and I think it would be mostly done 16:23:43 I see 16:24:03 this would be a perfect topic for mid-cycle 16:24:58 anybody else wants to share their wisdom/ideas on this subject? 16:25:24 it is just jroll and I going on this - it would be cool to hear others too :-):-0 16:25:45 * Sukhdev waitinv 16:25:53 * Sukhdev waiting 16:26:35 jroll: I think this sounds like a good direction - 16:26:59 mitchjameson and I will look into the ML2 part 16:27:21 and then we can look at hooking up into the ironic functional job framwork 16:27:32 cool 16:27:39 we can chat more about it at the midcycle too 16:28:16 correct - that is why I made the comment that it will perfect topic for mid-cycle 16:28:53 shall we move on? 16:29:17 #topic: Documentation 16:29:48 probably not much to chat about here yet :) 16:29:51 amotoki is not here - he and I discussed the strategy about documenting it form the neutron side 16:29:58 he is not here to provide any update - 16:30:06 so, I will punt this to next week - 16:30:21 Overall, I think we are in good shape on this part 16:30:46 We will have to look into documenting it in the Ironic side - but, we can do that later 16:31:06 Anything on the documentation? 16:31:25 #topic: Open Discussion 16:31:47 I wanted to ask everybody - if we are missing anything for Liberty delivery? 16:32:29 I think we're just missing hours in the day, mostly 16:33:21 ha ha - did you not know that Obama declared that an extra hour is added to every body's day :-):0 16:33:28 hah 16:33:32 I need more than an extra hour :) 16:33:41 as FYI we are uploading some initial patches of the implementation today 16:33:53 they are still work in progress :) 16:34:00 anyway, I think we're in good shape, I'd like to get the nova code done ASAP because feature freezes etc 16:34:02 lauramoore: cool - thanks 16:34:07 lauramoore: link? 16:34:36 the ironic one: https://review.openstack.org/#/c/206132/ 16:34:47 the nova one: https://review.openstack.org/#/c/206135/ 16:34:52 jroll: correct - please keep feature freeze in mind - Neutron is very strict about it and I am sure Nova will be too 16:34:54 the CLI one: in process 16:35:18 still in testing but welcome to any early feedback since this is our first contributions 16:35:38 lauramoore: possible we could split the ironic patch up a bit? that's pretty large to review in one shot 16:35:48 yeah i thought you'd say that! 16:35:54 lauramoore: thinking a chain with DB, then objects, then RPC, then API 16:35:56 :) 16:36:09 will work on this over next couple of days and re-send any links 16:36:15 lots of tests, though, I like :D 16:36:35 well they are also WIP :) 16:36:54 lauramoore: Thanks for sharing this - looks like a great start 16:36:58 :jroll, thanks for this advice, will aim for that 16:37:29 lauramoore: and on the nova patch... we don't want nova to send the link info, right? 16:37:42 we want it to send bind_requested: False, and let ironic update the link info later 16:37:57 because we don't want to put the node on the tenant network until after the deployment 16:39:18 ah ok, i'll fix this and update 16:39:40 python-ironicclient commit for review: https://review.openstack.org/#/c/206144/ 16:40:49 yhvh_: nice! 16:40:52 yhvh_: Thanks and welcome 16:42:15 yhvh_ and lauramoore: Thanks for sharing the links - we'll get on with the reviews and provide the feedback 16:42:45 Anything else? 16:43:15 Looks like we are done 16:43:29 awesome 16:43:33 jroll: here I give you 17 minutes :-):-) 16:43:41 a wild armax appears 16:43:52 armax: we were just finishing, anything you want to add? 16:43:59 or chat about 16:44:32 armax: we are wrapping up Ironic-Neutron integration meeting 16:45:04 * Sukhdev waiting 16:45:21 I guess we are done - 16:45:38 Thanks folks joining - it was a great discussion. 16:45:42 bye 16:45:49 #endmeeting