16:03:07 #startmeeting networking_ml2 16:03:09 Meeting started Wed Oct 19 16:03:07 2016 UTC and is due to finish in 60 minutes. The chair is rkukura. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:03:10 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:03:12 The meeting name has been set to 'networking_ml2' 16:03:18 #topic Agenda 16:03:28 #link https://wiki.openstack.org/wiki/Meetings/ML2#Meeting_October_19.2C_2016 16:03:47 Anything to add? 16:04:28 #topic Announcements 16:04:51 Summit is next week in Barcelona, so no ML2 IRC meeting 16:05:05 I will not be there :( 16:05:13 I'll be there 16:05:33 Sukhdev, did you finalize your plans? 16:05:46 yes - I will be there 16:05:49 great 16:06:00 any other announcements? 16:06:02 andreas_s : we should hook up there 16:06:03 nice! 16:06:15 Sukhdev, yes 16:06:36 #topic Ocata Planning 16:08:08 Given that the summit is next week, and Ocata development is underway, what should we try to accomplish with ML2? 16:08:41 certainly the live migration support is on the list, right? 16:09:05 yes, that one got more and more attention on neutron and nova side 16:09:26 it's also the main discussion point at the nova-neutron design session 16:09:51 I skimmed through last week’s ML2 IRC meeting minutes on that, but have not otherwise been tracking the discussion or proposals 16:10:36 Seems like the right points are being brought up (various use cases, compatability with nova and with ML2 drivers, etc.) 16:11:21 Are those of you going to Barcelona prepared to try to get agreement on how to move forward? 16:12:30 yes, I think so 16:13:01 so what else do we want to accomplish with ML2 in Ocata? 16:13:33 rkukura : can you summarize these on the wiki or etherpad 16:13:47 so, that others have access to the list 16:13:52 Sukhdev: What do you want me to summarize? 16:14:09 list of the potential topics 16:14:48 we’ve had a list from previous releases, which we could reconsider 16:15:20 rkukura : do you still have it somewhere 16:15:34 I'll be at summit and would like to discuss various VXLAN topics 16:16:07 about the only think I’m personally considering driving is to update the ML2 MD and ED APIs to conisistently cover all the resources that have been added to neutron 16:16:28 caboucha : can you eloborate on what specifically 16:17:31 With my driver, I'd like to configure VXLAN overlay with the switch. There are issues in regards to exceeding 4K vlans since there is only one vlan configured per network. 16:17:53 same vlan actually can be configured on various switches but different vni segments. 16:18:26 caboucha : you are not talking about HPB 16:18:30 Also has anyone attempted to implement VXLAN BGP-EVPN. 16:18:49 HPB - hierarchical port binding? 16:19:05 caboucha : yes 16:19:30 I don't believe so. Perhaps I can sync up with you at the summit. 16:19:40 caboucha : yes - Arista, Juniper 16:20:01 If it were HPB, I would have thought my long time openstack peers would have suggested it. 16:20:11 I'll look into it ahead of our discussion. 16:21:00 caboucha : i do not follow your use case of same VLAN on different VNIs - 16:21:17 caboucha : does the same VLAN represent the same neutron network? 16:22:22 caboucha: The main use case for HPB is to allow different switches (or switch ports) independently map the same VXLANs to different (or none at all) VLANs 16:22:25 caboucha : if your need is to support >4k VLANs - HPB is for you :-) 16:22:47 Yes rkukura that's what I'm trying to solve 16:23:09 THANK YOU ALL. I'll google the rest. 16:23:29 caboucha : come talk to me in Barcelona - we can help you with it 16:23:39 Much appreciated! 16:24:07 caboucha : there will be Arista folks who are experts on this subject, who just implemented it - I will introduce them to you 16:24:17 caboucha: There certainly could be followup work needed on HPB if it isn’t suitable for your use case 16:24:17 caboucha : and rkukura is the author of this 16:24:19 sounds good 16:24:45 caboucha : so, you are in the right company :-):-) 16:25:03 Sounds like it. 16:25:40 I should meet at the usual neutron gathering for ML2 folks? 16:26:49 caboucha: Hopefully Sukhdev will coordinate something for the “unplugged” session, assuming that works similarly to previous summits. I will not be in Barcelona. 16:27:22 So from last summit, some of the ideas for ML2 are in https://etherpad.openstack.org/p/newton-neutron-summit-ideas 16:27:44 oh wow, what a list 16:27:52 rkukura : I have been busy with change of employer - did not spend much time going through the etherpad 16:28:11 I’m not even sure what process is being used this round 16:28:57 rkukura : that is older list - did you mean to post the one for Ocata? 16:29:26 I haven’t seen one for Ocata, but I’m sure it exists 16:29:56 #link https://etherpad.openstack.org/p/ocata-neutron-server-next 16:30:00 ^Neutron server 16:30:10 #link https://etherpad.openstack.org/p/ocata-neutron-agents 16:30:14 ^ Neutron agents 16:30:37 thanks andreas_s! 16:32:22 I’m not seeing anything really specific to the ML2 plugin in those, except maybe the bit about dropping the l2pop MD 16:32:41 One topic to think about might be the impact of OVO on ML2 16:34:39 rkukura : I just posted a comment on the ether pad regarding dropping of L2Pop 16:34:46 I see 16:35:28 rkukura : OVO - what is the full form (pardon my ignorance) 16:35:30 Any thoughts on OVOifying ML2? 16:35:39 oslo versioned objects 16:35:46 thanks 16:36:34 I guess they already started the turn the port object into an ovo 16:37:10 https://review.openstack.org/#/c/351368/ 16:37:11 Since much of my focus is on GBP, which is still based on mitaka, I’m not all that familiar with what has been done with OVO in Newton 16:38:09 andreas_s: It seems those ehterpads are not the complete list of neutron topics - those are just two specific topics 16:38:35 Sukhdev, yes 16:38:53 Sukhdev, this is the whole list: https://wiki.openstack.org/wiki/Design_Summit/Ocata/Etherpads#Neutron 16:39:21 andreas_s : ah - thanks - that is what I was looking for 16:39:53 #link: https://wiki.openstack.org/wiki/Design_Summit/Ocata/Etherpads#Neutron 16:40:28 did trunk ports get completed during Newton? 16:44:40 I don't know, haven't followed that up 16:45:45 Sukhdev, andreas_s: Please report back if there are any discussions regarding incorporating the versioned objects into the ML2 APIs. One idea would be to pass these as part of the Context objects, allowing drivers to use use the versioned objects in place of the attribute dictionaries. 16:46:34 rkukura : will do 16:46:39 thanks 16:46:51 anything else on the “Ocata Planning” topic for today? 16:47:18 no 16:47:36 Sukhdev: anything from you? 16:47:51 no really 16:47:55 rkukura, I am currently working on incorporating versioned objects here: https://review.openstack.org/#/c/382037/ . Please have a look when free. 16:48:26 #topic Bugs/RFEs 16:49:03 sshank: thanks! 16:50:19 anything new to discuss on Bugs/RFEs? 16:50:49 I expect progress will be made on live migration next week, so we can have an update here the following week 16:51:00 no, just FYI, I continued the work on the nova-neutron live migration devref: https://review.openstack.org/#/c/274097/ 16:51:06 yes, I think so, too 16:51:43 I'll probably not be available the week after the summit, but the week after that again 16:51:46 andreas_s : I saw the etherpads for migration and posted my comments on them - 16:51:59 Sukhdev, excellent, thanks! 16:52:03 #topic Open Discussion 16:52:10 andreas_s : have a look and let me know 16:52:21 Sukhdev, I'll do 16:52:44 Given andreas_s’s comment above, should we skip the week after the summit due to travel/recovery? 16:53:24 rkukura : +1 16:53:29 ok 16:53:35 any other items to discuss? 16:54:28 if not, lets wrap up 16:54:43 have a great summit, and enjoy Barcelona!!! 16:55:00 thanks! 16:55:01 rkukura, thank you! 16:55:05 bye 16:55:09 #endmeeting