16:02:07 #startmeeting ironic_neutron 16:02:07 Meeting started Mon Aug 10 16:02:07 2015 UTC and is due to finish in 60 minutes. The chair is Sukhdev. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:02:08 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:02:11 The meeting name has been set to 'ironic_neutron' 16:02:42 Hello folks - we are ready for Ironic-Neutron integration meeitng 16:02:49 Sukhdev: sorry for the delay ;-) 16:02:55 hihi 16:02:59 apuimedo: no worries 16:03:00 Hi 16:03:04 hi 16:03:10 hi 16:03:31 jroll: are you here? 16:03:43 #topic: Agenda 16:03:50 #link: https://wiki.openstack.org/wiki/Meetings/Ironic-neutron 16:04:07 Anybody like to add anything to agenda? 16:04:28 #topic: Announcements 16:04:39 morning Sukhdev 16:04:50 Ironic mid-cycle is this week - Wednesday to Friday 16:05:05 jroll, mitch, and I will be attending 16:05:23 jroll: good morning 16:05:37 Anybody like to make any announcement? 16:05:57 Well, lets dive into the agenda 16:06:06 #topic: Patch reviews 16:06:40 yhvh_: thanks for pushing updated patches 16:06:45 np 16:06:50 more to come 16:07:22 If you do not mind, I would like to start with - https://review.openstack.org/#/c/206163/ 16:07:44 did you folks have an opportunity to look at my review comments? 16:08:12 * Sukhdev waiting 16:08:18 yes 16:08:29 not did we will next 16:08:56 yhvh_: did the comments make sense? 16:09:26 I felt some deviation from the agreed upon design - unless I misunderstood the code snippets 16:09:51 ok they seem simple enough, we will address them asap 16:09:56 so... a more general concern I have. it seems we're going to have much more nova change than I initially thought, including some config options. and we're past non-priority feature freeze for nova. I have a feeling this stuff won't be landing in Liberty. 16:10:56 jroll: are you serious? 16:11:05 jroll: can you elaborate, please? 16:12:00 Sukhdev: https://wiki.openstack.org/wiki/Nova/Liberty_Release_Schedule#Dates_overview 16:12:08 we're past the freeze for non-priority features 16:12:36 personally I had no idea we also had this patch for port groups,e tc 16:13:03 and nova really doesn't like config options past freeze dates, even if we can get the rest in 16:13:45 we can put up a blueprint for M and get things landed early in the cycle 16:14:04 folks can apply these patches if they wish to deploy it after liberty 16:14:15 and we can work on getting the rest of this really solid in the meantime 16:14:41 jroll: few points - 16:14:45 it's disappointing, but it's too late now :/ 16:15:16 1) I do not see any nova config changes - what config changes are you referring to? 16:15:40 1) https://review.openstack.org/#/c/206163/1/nova/virt/ironic/driver.py and we'll probably need one for the feature as a whole as well 16:16:48 Oh - I was thinking this is ironic driver - hence, we should be OK 16:17:42 jroll: I thought you were in touch with nova guys and they would let this go - 16:18:26 jroll: what if we split this into two - take out port-group for the first cut 16:18:28 Sukhdev: no, that was when there was one simple patch, and the sentiment was it didn't need a spec, not that it could land whenever 16:20:06 jroll: there are two ways to go with this in that case - 16:20:32 A) as you mentioned, have all the patches ready and let the customers use them and then land them in M release 16:21:10 B) make a slight adjustment to the plan and push the port-group stuff in the M release, but, push the non-portgroup stuff still in L release 16:21:36 this way we can have basic framework in L and then we can continue to work on it in M release 16:21:52 the thing is 16:22:02 both pieces depend on things being done in ironic or neutron 16:22:05 that aren't done yet 16:22:14 so we'd have to go even later in the cycle 16:22:42 I don't think that's going to fly, so I vote we go with A 16:23:18 neutron side is good to go - and, you mentioned we can continue to go with ironic side late in cycle (as there is no FF concerns) 16:23:35 as long as we manage the nova side, we should be good, no? 16:23:52 neutron side is merged? 16:24:03 and we have drivers that support the dont-bind-yet flag or whatever? 16:24:14 jroll: we have until first week of Sept to merge neutron side 16:24:22 ok, right 16:24:38 yeah the rest should be fine then, assuming we continue to make good progress on the ironic code 16:25:18 Think ironic side stuff is making good progress now 16:25:19 because this is a lot of ironic code :) 16:26:32 we've addressed most comments, should have the rest addressed by mid-cycle 16:26:37 jroll: It would be worthwhile to consider splitting the port group and non-portgroup code and see if we can make it to the L cycle 16:27:01 Sukhdev: we're already past non-priority feature freeze, it won't happen 16:27:14 sorry :( 16:27:26 I lost track of dates, this is partially my fault 16:27:32 partially/mostly/all 16:28:36 jroll: I am bit lost - if we pull out the port groups, then there is no change in nova config - am I missing something? 16:29:13 jroll: then it strips down to stuffing the binding:profile information and the ironic API 16:29:15 Sukhdev: I think we still want a config there whether to send bind_requested, for those not using this feature 16:29:25 and even without the config it's a hard sell 16:29:36 I can run it by john to be sure, but I really doubt it 16:29:56 jroll: is John going to be in Seattle this week? 16:30:06 no 16:30:20 oh 16:30:40 shall we have a discussion with him - with the stripped down plan and see if this will work? 16:31:25 I'll chat with him 16:31:26 if we can get the basic patches merged in L and then still continue to work on the advanced patches and possibly push them in early M - that would be a huge win 16:31:46 but, this is a non-priority feature and we're long past the non-priority feature freeze 16:31:54 they've already blocked some pretty critical stuff 16:32:07 and besides, like I said, this depends on work in other projects that is not done 16:32:23 jroll: my understanding has all along been that they will allow this (at least the stripped part) 16:32:44 all that was said to me is that the ironic spec was probably enough, and we won't need a nova spec for this. 16:32:50 that doesn't mean none of the process applies. 16:33:24 I see 16:33:29 I totally dropped the ball on getting this stuff in, I lost track of dates and got distracted with other work. I'm sorry for that 16:34:47 I think it would be worthwhile to have a chat with John and others to see if can land some of it in L - otherwise, we have no choice but to go with plan A 16:35:45 sure 16:36:02 jroll: we should discuss this with devananda as well - he may have some thoughts as well 16:36:30 when we meet in Seattle on Wednesday, lets discuss it further 16:36:44 ... sure, I don't see why devananda would be able to circumvent nova processes, but ok 16:37:31 regardless, we should not slow down the progress and keep marching forward 16:37:46 with that in mind - lets get back to the patches 16:38:38 jroll: the last two patches in the list - I posted comments - do not know if you had a chance to review or not 16:39:13 no, I haven't yet 16:39:42 jroll: we should come up with one patch which addresses the sprit of both of these patches, and add the binding: profile stuff to it 16:40:04 yeah, need to review what exactly should be there and update the patch 16:40:45 jroll: also, we need an additional patch to cover the network flip logic 16:41:12 that's in ironic 16:41:31 so, we need two additional patches - 16:41:33 and should look much like https://review.openstack.org/#/c/139687/ 16:41:41 but updated to not be 6 months old :) 16:42:53 Oh - I have not seen this beforeā€¦. 16:43:12 this was before the updates for port groups etc 16:43:18 so don't read too much into it :) 16:43:30 but https://review.openstack.org/#/c/139687/5/ironic/networks/neutron_plugin.py 16:43:36 should be pretty close 16:43:47 correct - the updated version of something like this and combination of the two of your other patches (along with stuffing of the binding profile) 16:44:06 right, my goal is to get all of this done this week 16:45:03 jroll: cool - so, lets get this done and yhvh_ is making good progress on the other three patches 16:45:31 +1 16:46:20 with all of this in place, having a discussion with John, devananda and other Nova stake holders, lets see what can be landed in L and what needs to be pushed out to M 16:47:31 yhvh_: what do you think? 16:48:35 sounds good 16:48:54 yhvh_: Are you good with the feedback on the patches? 16:49:52 yeah I'll maybe continue one or two comment threads but that is the better place to discuss 16:50:26 yhvh_: just curious - as we were discussing, if we were to split the port group stuff - wondering how simple the patch becomes 16:51:48 * Sukhdev time check - 9 min. left 16:51:52 it already is split, best I can tell 16:53:13 jroll: i mean taking out the port group related config out of this - https://review.openstack.org/#/c/206163/1/nova/virt/ironic/driver.py 16:54:03 I think that patch needs the config 16:54:07 or maybe it doesn't need it at all 16:54:13 but there's nothing to split IMO 16:55:16 jroll: I wonder if port group was not in the play what modifications are needed in that driver - which seems to be a blocker from the nova point of view :-) 16:55:40 I don't understand 16:55:49 the blocker is that it's a feature after the feature freeze. 16:56:02 not that it's a config. not that it's a driver patch or not driver patch. 16:56:41 now I am lost :-):-) 16:57:14 we'll talk wednesday 16:57:47 sounds good 16:58:09 down to 3 mins - anybody wants to discuss anything else? 16:58:16 I'll try be around irc while you are meeting for any fixups this end 16:59:04 sounds good 16:59:19 Thanks folks 16:59:25 thanks 16:59:30 tyty 16:59:30 bye 16:59:34 #endmeeting