16:01:22 <Sukhdev> #startmeeting ironic_neutron
16:01:23 <openstack> Meeting started Mon May 23 16:01:22 2016 UTC and is due to finish in 60 minutes.  The chair is Sukhdev. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:01:24 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:01:26 <openstack> The meeting name has been set to 'ironic_neutron'
16:01:35 <mjturek1> o/
16:02:02 <jroll> \o
16:02:12 * jroll is in another meeting at the same time so sorry if I am slow
16:02:21 <Sukhdev> #topic: Agenda
16:02:22 <sambetts> o/
16:02:39 <Sukhdev> jroll : no worries - I was going bit slow for everybody to join in :-)
16:02:46 <Sukhdev> #link: https://wiki.openstack.org/wiki/Meetings/Ironic-neutron#Meeting_May_23.2C_2016
16:03:02 <Sukhdev> #topic: Announcements
16:03:21 <Sukhdev> N-1 is this week, right?
16:03:36 <jroll> next week
16:03:38 <jroll> http://releases.openstack.org/newton/schedule.html
16:04:12 <Sukhdev> right, 30th of the month
16:04:39 <Sukhdev> Any body has any other announcement
16:04:53 <jroll> next monday is a US holiday
16:04:58 <jroll> do we plan to have a meeting?
16:05:21 <Sukhdev> jroll : good thinking - lets celebrate holiday and skip the meeting
16:05:28 <jroll> cool :)
16:05:55 <Sukhdev> So, folks no meeting next week - I will post on the ML later
16:06:18 <sambetts> UK Holiday too
16:06:43 <Sukhdev> cool - even bigger reason to celebrate :-):-)
16:06:58 <Sukhdev> #topic: VLAN Aware Servers -
16:07:13 <Sukhdev> #link: https://review.openstack.org/#/c/277853
16:07:27 <Sukhdev> sambetts : thanks for updating the spec
16:07:38 <Sukhdev> I have not had a chance to review the updated version
16:08:16 <sambetts> \o/ hopefully I've covered 99% of what we discussed at the summit
16:08:39 <Sukhdev> even the earlier version looked good to me - so, I am sure this is in much better shape
16:08:45 <Sukhdev> others - please review this
16:09:00 <Sukhdev> and lets get this finalized
16:09:02 <sambetts> it wasn't a drastic change, just incorporating the neutron trunk stuff a little better
16:09:52 <Sukhdev> I will follow up with rosella to see how the neutron side implementation is coming along
16:10:07 <mjturek1> question, will this spec be targeted for Newton?
16:10:47 <jroll> mjturek1: it almost certainly won't happen in newton, imo
16:11:00 <mjturek1> jroll: got it, thanks!
16:11:34 <Sukhdev> we can get all the agreements and design done in N and then fully implement in the O
16:11:38 <jroll> mjturek1: just, lots of work and too much to do already :)
16:12:00 <mjturek1> jroll: Sukhdev: makes sense :)
16:12:26 <Sukhdev> #topic: Critical Patches for N
16:12:47 <Sukhdev> I added two additional to the list (nova patches)
16:12:56 <Sukhdev> so that we can track them on a weekly basis
16:13:29 <Sukhdev> So, now we have 6 patches
16:13:57 <Sukhdev> I could not follow up on the split of the patch -
16:14:08 <jroll> we need to pick those nova things back up, they're super old
16:14:38 <Sukhdev> jroll - one of them is really really old -
16:14:57 <sambetts> Yeah, I would also like to consider the new spec I've got up for the ironic/nova interaction
16:15:56 <Sukhdev> sambetts: I think jroll had some agreements with nova folks - we should align these patches with that - assuming that still holds
16:16:19 <sambetts> This -> https://review.openstack.org/#/c/313001/ may also affect us
16:16:42 <jroll> sambetts: which new spec?
16:16:51 <jroll> the plug_vifs thing?
16:16:58 <sambetts> jroll: https://review.openstack.org/#/c/317636/
16:16:59 <sambetts> yeah
16:17:21 <jroll> yeah
16:17:37 * jroll feels like there's almost too much happening here sometimes
16:18:14 <Sukhdev> sambetts : I do not think routed networks related work will impact us in any way
16:18:56 <jroll> Sukhdev: it will, it splits the port-create into a create and a later update
16:19:08 <sambetts> what jroll said
16:19:40 <jroll> I *think* it's more like, we need to work with john to make sure it just works
16:19:46 <jroll> but there's discussions to be had, at a minimum
16:19:55 <sambetts> ++
16:20:21 <Sukhdev> we already work with that assumption - i.e. create port first and then update
16:20:33 <Sukhdev> this is like early binding vs. late binding
16:20:52 <Sukhdev> this shifts allocation of ID in the late binding -
16:21:21 <Sukhdev> our integration work (sort of) already works with late binding model
16:21:50 <jroll> I mean, this changes the entire port allocation process, we do weird things, we'll need to be involved
16:21:59 <jroll> even if it's just keeping an eye on it to make sure it doesn't break us
16:22:24 <Sukhdev> While I agree that we should align our work, but, we should not derail ourselves because of this
16:22:38 <Sukhdev> Keeping an eye on it and driving correctly makes sense
16:23:06 <jroll> sure
16:23:19 <jroll> hence the "this may affect us" :)
16:24:27 <Sukhdev> right - agree, but, I am worried - rather then being driven by these, I rather proactively drive these
16:25:19 * sambetts crys because cross project is hard
16:25:21 <Sukhdev> I will speak with carl_baldwin and see if there is anything which could impact our work
16:25:58 <Sukhdev> sambetts : yes, part of the issue is that we delayed our integration a bit -
16:26:32 <Sukhdev> and the world is moving on - hence, it impacts us :-):-)
16:26:55 <sambetts> yup :) networking based schedualing will be good for us in the end
16:28:29 <Sukhdev> routed networks is like HPB - i.e. allocation of the segmentation IDs and IP allocation is pushed to the later stage via update port - as oppose to create port
16:29:43 <sambetts> sounds like a good model for us
16:29:44 <Sukhdev> I have an RFE out for manila-neutron integration - it impacts that integration more than Ironic-neutron integration
16:30:06 <Sukhdev> but, yes, things are moving too fast from all sides, hence, we need to stay atop of this work
16:31:12 <Sukhdev> i will follow up with carl_baldwin and armax to understand the changes and see what we have to do (if anything at all) to align with this
16:31:53 <Sukhdev> I less plugged into nova side - I rely on jroll for that
16:31:56 <jroll> well, the nova internals are the parts I'm concerned about
16:32:02 <jroll> the neutron stuff shouldn't change much
16:33:14 <Sukhdev> jroll : we should see how our nova patches are impacted - we need to update them anyways
16:33:37 <jroll> I'm concerned about the entire system, not just those patches
16:33:38 <jroll> but, agree
16:33:48 <sambetts> ++, hopefully my new spec should help us (ironic) interate faster on the ironic specific networking stuff
16:34:29 <sambetts> by moving alot of the logic out of the nova driver
16:36:54 <Sukhdev> What is the best way to proceed from here? We have couple of specs - perhaps review these and see how to align them?
16:37:33 <sambetts> Sounds like the logical way forward
16:38:53 <Sukhdev> Lets all review these specs and put appropriate comments on these -
16:40:33 <Sukhdev> hshiina points out that nova feature freeze is June 30
16:41:02 <sambetts> non-priority FF specifically
16:41:21 <Sukhdev> lets keep that in mind, in the past we got delayed because we missed the FF deadline :-)
16:41:37 <sambetts> Nova non-priotry spec freeze is next week
16:42:05 <sambetts> so if we need any specs landed we need to push in the next week
16:43:06 <Sukhdev> sambetts : oh - i thought it was June 30 -
16:43:11 <sambetts> spec freeze
16:43:16 <sambetts> not feature
16:43:24 <Sukhdev> oh my bad - got it
16:44:16 <sambetts> Our nova network spec was approved for mitaka, is it still valid or do we need to resummit? I'm not sure of the nova process
16:44:37 <jroll> sambetts: it was re-accepted already iirc
16:44:50 <jroll> yep http://specs.openstack.org/openstack/nova-specs/specs/newton/approved/ironic-networks-support.html
16:45:20 <sambetts> jroll: ah right, I was looking at the Blueprint, and it said Accepted for Mitaka
16:45:54 <jroll> could ask mriedem to fix that
16:46:58 <Sukhdev> So, we are good from the spec point of view - we need to check the impact of John's spec on our work - right?
16:47:23 <Sukhdev> and see how all this aligns with sambetts's proposed spec
16:47:52 <sambetts> Sounds like action items to me
16:48:28 <Sukhdev> I just wanted to be clear so that I can focus my work for the week
16:48:49 <Sukhdev> I will review these patches - others please do so
16:48:58 <Sukhdev> as well
16:50:26 <Sukhdev> sambetts : there are couple of items you added to the agenda last week - we covered them, but, I left it on the agenda - just in case
16:50:48 <sambetts> Sukhdev: yup for the ironic interface attach API one is the spec I was talking about above ^
16:50:57 <sambetts> Sukhdev: so we've covered that today already
16:51:25 <Sukhdev> I thought so, but, wanted to make sure you are good with that
16:51:37 <sambetts> And did you have a chance to do some digging into if neutron is looking to support bonding for virt?
16:52:08 <Sukhdev> no - I could not - the week was very crazy for me -
16:52:23 <Sukhdev> I will try to do it this week
16:53:09 <Sukhdev> hshiina has an item on agenda under Open Discussion
16:53:19 <sambetts> cool, that I think is really important to know because if we merge the portgroups API stuff then have to deprecate it in favor of neutrons model that'll hurt a lot
16:53:21 <Sukhdev> #topic Open Discussion
16:53:39 <Sukhdev> sambetts : understood -
16:53:48 <Sukhdev> hshiina : are you around?
16:54:04 <hshiina> yes. I'm afraid i was late to join.
16:54:21 <hshiina> we've tried portgroup in our environment.
16:54:58 <hshiina> then, we think we need one more fix to pass bonding information to an instance.
16:55:21 <hshiina> using configdrive. is it correct?
16:55:25 <sambetts> right, so that part isnt covered by either of our current nova patches
16:56:34 <Sukhdev> #link: http://specs.openstack.org/openstack/nova-specs/specs/liberty/implemented/metadata-service-network-info.html#rest-api-impact
16:57:05 <sambetts> its something that does need doing and I have a POC of generating config drive network info here https://review.openstack.org/#/c/289412/1/nova/virt/ironic/driver.py
16:58:06 <sambetts> its based around the VLAN aware VMs stuff, but could be quite easily changed to fit into the current model
16:58:23 * Sukhdev time check 2 min
16:58:58 <hshiina> sambetts, your patch would be helpful. i think.
16:58:59 <sambetts> btw this does all the logic inside of the ironic virt driver, in the end we might need to hook into novas actual metadata generator but it works like this for now
16:59:21 <Sukhdev> #link: https://review.openstack.org/#/c/289412/1
16:59:23 <sambetts> also the file that is generated by this, currently will only be fully processed by glean
17:00:21 * Sukhdev time check - we are out of time folks
17:00:23 <sambetts> cloud-init doesn't have full support yet, but my patches to ensure bonding and multivlan worked for simple-init/glean were merged
17:01:02 <Sukhdev> lets take this to ironic channel -
17:01:21 <Sukhdev> Thanks for attending folks - remember no meeting next week
17:01:22 <Sukhdev> bye
17:01:30 <Sukhdev> #endmeeting