16:02:17 <Sukhdev> #startmeeting networking_ml2
16:02:18 <openstack> Meeting started Wed Aug 26 16:02:17 2015 UTC and is due to finish in 60 minutes.  The chair is Sukhdev. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:02:19 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:02:22 <openstack> The meeting name has been set to 'networking_ml2'
16:02:44 <Sukhdev> #topic: Agenda
16:02:50 <Sukhdev> #link: https://wiki.openstack.org/wiki/Meetings/ML2
16:03:11 <Sukhdev> Anybody would like to add anything to the agenda?
16:03:57 <Sukhdev> #topic: Announcements
16:04:17 <Sukhdev> rkukura is off today - he will not be joining us
16:04:46 <Sukhdev> Neutron FF is next week - so, basically we have this week to push any code for Liberty
16:05:00 <Sukhdev> Anybody has any announcement?
16:05:45 <Sukhdev> #topic: Action items
16:06:13 <Sukhdev> rkukura was going to finalize the host for the late-cycle sprint
16:06:34 <Sukhdev> we have three offers for hosting - Cisco, Yahoo, and Brocade
16:07:09 <Sukhdev> Since rkukura is out today, I sent out the email to all the (potential) hosts to respond
16:07:18 <Sukhdev> Hopefully, we will finalize this soon
16:08:00 <Sukhdev> Here is the link to the etherpad for the late-cycle sprint - https://etherpad.openstack.org/p/Neutron_ML2_Mid-Cycle_Sprint
16:08:20 <Sukhdev> moving right along
16:08:47 <Sukhdev> #topic: Driver API for security group
16:09:12 <Sukhdev> We discussed this last week
16:09:37 <Sukhdev> yamahata: you were going to investigate this
16:09:55 <yamahata> yamamoto?
16:09:55 <yamamoto> i guess it wa me
16:10:14 <yamamoto> i had no chance to work on it this week.
16:10:26 <yamamoto> and i don't think i have time in the next two weeks.
16:10:39 <Sukhdev> my bad - sorry auto-fill feature matched yamahata insetad of yamamoto - sorry about that :-)
16:11:02 <yamamoto> no problem, it's very common mistake :-)
16:11:13 <yamahata> yeah, very common.
16:11:45 <Sukhdev> yamamoto: I understand because of the release deadlines, everybody is very busy - including me
16:12:11 <Sukhdev> yamamoto: since this will be for next release cycle, I guess we can delay it a bit -
16:12:17 <yamamoto> sure
16:12:38 <Sukhdev> yamamoto: however, we should plan in such a way that we meet the deadline to submit for Tokyo Design Summit
16:13:18 <Sukhdev> yamamoto: the deadline is not announced yet, but, but my guess it will be within next 2 weeks or so.
16:13:45 <Sukhdev> So, we have time - no worries
16:14:08 <Sukhdev> If you want to share anything, you can alway email me at sukhdev@arista.com
16:14:32 <yamamoto> noted, thank you
16:14:48 <Sukhdev> moving on to the next topic
16:15:06 <Sukhdev> #topic: Physical Topology Discussion
16:15:36 <Sukhdev> I do not see asoumya or shivharis here -
16:16:00 <Sukhdev> I will ping them off-line and see if they have made any progress
16:16:08 <yamahata> In the context of physical topology, I'd like to introduce common layer for provider network support.
16:16:50 <yamahata> Hopefully it can be discussed at Tokyo summit...
16:17:08 <Sukhdev> yamahata: it is my understanding that this deals with the physical connectivity of the network elements -
16:17:27 <Sukhdev> yamahata: networks will overlay on top of the physical topology
16:17:43 <yamamoto> yamahata: common among what?
16:17:55 <yamahata> I mean https://blueprints.launchpad.net/neutron/+spec/dynamic-network-segment-configuration
16:19:13 <Sukhdev> yamahata: I do not believe this BP is related to the physical connectivity
16:19:13 <yamahata> Right now physical bridge mapping is handled by each agent. But make it common and usable to mechanism driver.
16:19:35 <yamahata> Right. I thought it's a bit related, though.
16:21:07 <Sukhdev> drivers (especially type drivers) can use the topology information to allocate/assign segment-id based upon the connectivity of the devices - rather than globally across the data centers
16:22:00 <Sukhdev> this allows the drivers to dynamically allocate segments/IDs, etc.
16:22:19 <Sukhdev> It also helps Nova in terms of placing the VMs -
16:23:06 <Sukhdev> for instance, one could define affinity among the VMs so that the placement can take advantage of the physical connectivity information of the hosts
16:24:06 <Sukhdev> e.g. place all VMs related to a given application on specific host which is closest to the DB host, etc..
16:24:21 <yamahata> I see. Thanks and sorry for hijaking this topic.
16:24:47 <Sukhdev> yamahata: no worries - this meeting is for such discussions :-)
16:25:29 <Sukhdev> any other question on this topic, before we move along?
16:26:01 <yamahata> no from me
16:26:09 <yamamoto> nothing from me
16:26:20 <Sukhdev> cool - thanks
16:26:46 <Sukhdev> #topic: modular l2 agent and macvtap
16:27:34 <Sukhdev> scheuran is out on PTO and he did not update the status - so, we do not have any update other than that this is begin worked on
16:27:49 <Sukhdev> he was planning on trying to get into L3 -
16:28:25 <Sukhdev> I am sure sean is making progress on this
16:28:51 <Sukhdev> with this we are on the last topic in the agenda
16:28:59 <Sukhdev> #topic: Open Discussion
16:29:52 <Sukhdev> yamahata: I will make sure when rkukura has finalized the host for late-cycle sprint, he touches base with you
16:30:09 <yamahata> Sukhdev: sure. thanks for arrangement.
16:30:26 <Sukhdev> yamahata: I am hoping that you are still planning on joining us for the sprint
16:30:51 <yamahata> I'll attend. I'm not sure about full week attendance though
16:31:29 <Sukhdev> yamahata: I will pull in out expert on the sync issues from Arista - he has lots of data to share on the sync related issues that he has been testing
16:31:54 <yamahata> Sukhdev: That's very great.
16:31:57 <Sukhdev> yamahata: many of the issues that you pointed out in your slide deck few weeks back
16:32:55 <Sukhdev> if there is nothing else to discuss - lets cut this short this week
16:33:44 <Sukhdev> thanks for attending today's meeting
16:33:52 <yamamoto> thank you
16:33:56 <yamahata> thanks
16:33:58 <Sukhdev> #endmeeting