16:08:04 <rkukura> #startmeeting networking_ml2
16:08:05 <openstack> Meeting started Wed Nov 11 16:08:04 2015 UTC and is due to finish in 60 minutes.  The chair is rkukura. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:08:06 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:08:08 <openstack> The meeting name has been set to 'networking_ml2'
16:08:22 <rkukura> #topic: Agenda
16:08:34 <rkukura> #link: https://wiki.openstack.org/wiki/Meetings/ML2#Meeting_November_11.2C_2015
16:08:47 <rkukura> Anything to add to the agenda?
16:09:11 <rkukura> #topic Announcements
16:09:30 <rkukura> welcome back from the Summit/travel/week-off/etc.
16:10:01 <Sukhdev> back but, heavily jet-lagged :-)
16:10:05 <rkukura> Mitaka-1 is scheduled for 12/1
16:10:42 <rkukura> Would be good to try to at least get specs approved by then for anything we want to accomplish
16:10:58 <rkukura> Any other announcements?
16:11:30 <rkukura> #topic Mitaka Summit recap
16:11:53 <rkukura> #link https://wiki.openstack.org/wiki/Design_Summit/Mitaka/Etherpads#Neutron
16:12:06 <rkukura> That’s the link to the etherpads from the summit
16:13:28 <rkukura> I’ve included in the agenda the main ML2-related work items that were discussed at the summit: SG driver support, driver status, trunk ports, modular l2 agent, and topology
16:14:04 <rkukura> Any comments on the summit?
16:15:41 <Sukhdev> it will be nice if yamamoto can provide the writeup on the SG driver
16:16:11 <rkukura> Sukhdev: Your ironic presentation was nice!
16:16:23 <Sukhdev> rkukura : thanks
16:17:19 <rkukura> Any other ML2-related developments at the summit that I haven’t mentioned in the agenda?
16:17:49 <rkukura> #topic Driver API for SecurityGroup
16:17:57 <scheuran> sorry, got dropped for a moment
16:18:15 <rkukura> Doesn’t look like yamamoto is here to update on this
16:18:29 <rkukura> The spec is being actively reviewed:
16:18:37 <rkukura> #link https://review.openstack.org/#/c/240356/
16:19:16 <pc_m> rkukura: At the summit were some solutions discussed for handling SG?
16:19:36 * pc_m the BP seems to only mention the issue
16:19:56 <rkukura> pc_m: There was some discussion at the ML2 table on Friday mornign
16:20:22 <pc_m> rkukura: Ah. Will look forward to updates to the BP.
16:20:52 <rkukura> One issue from the spec review is whether the notification/callbacks are sufficient for drivers to track SG state updates, or if we need to add precommit/postcommit methods for SG resources.
16:21:31 <rkukura> It would be great for anytone interested in this to give this some thought and comment on the review
16:21:56 <Sukhdev> rkukura : yamamoto had made some comments about there being an issue with the callback mechanism
16:22:05 <rkukura> The other issue is whether we need ML2 port binding to guarantee that at least one MD takes responsibility for enforcing SGs when applicable.
16:22:21 <Sukhdev> hence, I asked him to document the issue so that it can be debated
16:23:13 <pc_m> rkukura: thanks for summarizing.
16:23:35 <rkukura> Sukhdev: One question is whether the same BP/spec should cover both aspects
16:23:53 <Sukhdev> I think it should cover both
16:24:10 <rkukura> I’d rather see one BP/spec provide a complete solution
16:24:25 <rkukura> OK, hopefully we’ll have some progress on this by next week
16:24:42 <Sukhdev> I will be happy to help and participate in this - if someone can take the lead on this
16:25:29 <rkukura> I think yamahata is leading this, and I’m willing to help out as well
16:25:31 <Sukhdev> I mean help in some of the implementation and testing
16:26:23 <rkukura> Sukhdev: Do you see switch-based SG enforcement being applicable with ironic/baremetal?
16:26:42 <Sukhdev> rkukura : Yes - very much so
16:27:04 <pc_m> I'm researching into what it takes for a MD to support SG under Ironic, so I'll be interested in all this and can try to help once I come up to speed.
16:27:09 <Sukhdev> hence, I am willing to step in and help with the implementation and testing of this
16:27:48 <rkukura> I think we’ve got sufficient interest to really try to get this into Mitaka
16:28:01 <rkukura> Anyhing else on it before we move on?
16:28:07 <Sukhdev> pc_m : we can be good friends once you figure out the details - as I am looking into this as well
16:28:23 <pc_m> Sukhdev: sure!
16:28:29 <rkukura> #topic Driver status coalescing/visibility
16:29:06 <rkukura> This discussion was followup to the discussions at the pre-cycle meeting
16:29:47 <rkukura> We had some volunteers to put together a proposal: John Joyce, yamahata, and asomya
16:30:00 <rkukura> I don’t think any of them are here right now
16:30:48 <rkukura> If we are going to do this in Mitaka, I think we need to get a concrete proposal soon, so we’ll continue to track this
16:31:02 <rkukura> I’m sure all driver maintainers have an interest
16:31:15 <rkukura> Next...
16:31:33 <rkukura> #topic Trunk port binding for VLAN-awere VMs
16:31:45 <rkukura> s/awere/aware/
16:31:57 <rkukura> Is Bence here?
16:32:17 <rkukura> Anyway, this was presented and discussed at the summit
16:32:32 <rkukura> #link Trunk port binding for VLAN-awere VMs
16:32:42 <rkukura> #link https://blueprints.launchpad.net/neutron/+spec/vlan-aware-vms
16:33:06 * rkukura should read pasted text before hitting enter
16:33:43 <rkukura> I volunteered to help figure out how ML2 port binding will work with trunk ports
16:34:02 <rkukura> Sukhdev: This also seems relevant to ironic
16:34:24 <Sukhdev> rkukura : yes, I was typing and you asked
16:34:40 <Sukhdev> this is very relevant to ironic - hence, I can help with this as well
16:34:52 <rkukura> I think the focus so far has been on OVS agent assembling the networks into a trunk, but we should consider switches doing this as well
16:35:50 <rkukura> Any other comments on this?
16:35:52 <Sukhdev> rkukura : this was presented in vancouver summit and thing really happened - do you think Bence is serious in pusuing this one?
16:36:18 <rkukura> Sukhdev: I think so
16:36:44 <Sukhdev> Good - I will keep my eye out for this
16:36:54 <rkukura> OK, next…
16:37:03 <rkukura> #topic Modular L2 agent
16:37:14 <rkukura> scheuran: Do you have an update on this?
16:37:21 <scheuran> to make it short, I don't have any updates
16:37:40 <scheuran> will try to bring a new patchset up later this week
16:37:45 <scheuran> and then come back next week
16:38:05 <rkukura> I think I just noticed an openstack-dev thread related to this
16:38:32 <scheuran> can you share? Maybe I've overseen it...
16:38:58 <rkukura> #link http://lists.openstack.org/pipermail/openstack-dev/2015-November/078263.html
16:39:27 <rkukura> I haven’t yet read the thread
16:39:28 <scheuran> thanks!
16:40:07 <rkukura> I get the feeling there are several different people/groups working on making L2 agents more modular, but efforts aren’t really coordinated yet
16:40:57 <scheuran> not sure. At least this thread is about agent extensions like QoS
16:41:13 <scheuran> but I need to dig into the details
16:41:37 <rkukura> scheuran: Right, which is not exactly the same as what you are working on, but overall both add to modularity
16:42:07 <rkukura> OK, anything else on this?
16:42:15 <scheuran> not from my side
16:42:29 <rkukura> #topic Topology
16:42:44 <rkukura> This was presented at the summit
16:42:55 <rkukura> I don’t see shivharis or asomya here today
16:43:16 <rkukura> There seemed to be significant interest
16:43:22 <rkukura> any comments?
16:43:58 <rkukura> moving along then…
16:44:03 <rkukura> #topic Open Discussion
16:44:14 <rkukura> Anything to discuss today?
16:45:29 <rkukura> Thanks Sukhdev, pc_m and scheuran!
16:45:45 <scheuran> welcome
16:45:45 <pc_m> rkukura: Thanks for the info!
16:45:50 <rkukura> last call…
16:45:52 <Sukhdev> thanks for chairing today
16:45:55 <rkukura> #endmeeting