16:04:47 #startmeeting networking_ml2 16:04:48 Meeting started Wed Oct 1 16:04:47 2014 UTC and is due to finish in 60 minutes. The chair is rkukura. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:04:49 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:04:52 The meeting name has been set to 'networking_ml2' 16:05:23 #link agenda: https://wiki.openstack.org/wiki/Meetings/ML2#Agenda 16:05:36 #topic Announcements 16:05:58 Kilo is now open for new specs 16:06:27 There is a new template as well, so some updates will be needed moving Juno specs to Kilo 16:06:41 Spec template has changed a bit - 16:07:01 Sukhdev: anything significant? 16:07:18 rkukura: not really 16:07:59 Kilo Summit topics for Neutron: https://etherpad.openstack.org/p/kilo-neutron-summit-topics 16:08:02 One question, I would like to propose a new CLI in python neutron-client , do I need to propose a specs in kilo ? 16:08:53 romilg: Is this part of a spec for a new feature, or purely a client change? 16:10:18 The summit topics etherpad above lists quite a few ideas related to ML2. I added an agenda item for today to discuss ML2 themes for kilo that may help us determine what summit sessions would be most useful. 16:10:56 Any other announcements? 16:11:26 #topic Action Items 16:12:07 We had: banix to work with mlavalle and the team to figure out best testing strategy for bulk operations 16:12:25 but I don’t see either banix or mlavalle online 16:12:51 We also had an AI for shivharis to followup with mestery regarding RC1 bugs 16:13:01 But shivharis is traveling 16:13:16 Any other AI updates? 16:13:41 #topic Bugs 16:14:35 When I last checked, there were 3 bugs blocking RC1, but none seemed to involve ML2 16:14:57 There was one ML2 bug regarding a deadlock that was resolved yesterday 16:15:32 One of the 3 bugs is the removal of the old openvswitch and linuxbridge plugins :) 16:15:52 Does anyone have any bugs that need discussion today? 16:16:26 rkukura: looks like our bug czar is missing today 16:16:32 https://bugs.launchpad.net/neutron/+bug/1179223 16:16:34 Launchpad bug 1179223 in neutron "Retired GRE and VXLAN tunnels persists in neutron db" [High,In progress] 16:16:37 Sukhdev: right, I mentioned that 16:16:54 it would be great if we could set the milestone for the same 16:17:29 Another 16:17:29 https://bugs.launchpad.net/neutron/+bug/1224978 16:17:31 Launchpad bug 1224978 in neutron "port binding on multi segment networks could lead to agent misconfiguration" [Medium,In progress] 16:17:31 I think it is already in good shape. 16:17:32 romilg_: That one does seem important, but is not currently targeted as a Juno RC. I think there is a patch, right? 16:17:56 yeah 16:18:02 romilg_: Shiv had action item to follow up on that bug with mestery to include in RC1 16:18:14 ok 16:18:24 The 2nd sounds like one for kilo 16:18:27 The first one is https://review.openstack.org/#/c/121000/ 16:18:52 re: 2nd one, +1 for kilo. 16:19:22 amotoki: do you feel the 1st one should block RC1? 16:19:35 rkukura: no. it is not a blocking issue. 16:20:20 amotoki: It does involve schema changes, so would not easily be backported 16:20:22 rkukura: it contains db change, so we cannot backport it after the release. this is the only reason I am reviewing it. 16:21:09 amotoki: Seems we should talk to mestery about it. 16:21:22 Any other bugs to discuss today? 16:22:00 #topic Specs 16:22:36 We’ll want to update https://wiki.openstack.org/wiki/Tracking_ML2_Subgroup_Reviews#Under_Review as we post specs for Kilo 16:22:48 Anything else on specs for today? 16:23:18 #topic Kilo Themes 16:23:34 rkukura, will you be updating, reposting the hierarchical spec? 16:23:44 rcurran: yes 16:24:01 My goal in adding this to the agenda was to get us started prioritizing work for ML2 during Kilo 16:24:56 And then to use that to determine how we can make best use of the design summit 16:25:34 So I added a number of items, mostly things that didn’t get completed in Juno, and/or that have been discussed in the past 16:26:03 rkukura: are you talking about https://etherpad.openstack.org/p/kilo-neutron-summit-topics ? 16:26:21 amotoki: No, the list in this meeting’s agenda 16:26:36 rkukura: thanks 16:26:40 #link https://wiki.openstack.org/wiki/Meetings/ML2#Meeting_October_1.2C_2014 16:27:55 What do others feel are the top priorities for ML2 during Kilo (other than vendor-speciic stuff)? 16:28:43 rkukura: I think your list is good 16:29:06 Anything missing? 16:29:25 Anything anyone feels is not worth at least considering? 16:29:49 I think it is good coverage for ML2 items. 16:30:49 Lets all think about this during the next week, and try to prioritize a bit next week, and then we can start talking about which items need time at the summit 16:30:53 Possible items related to ml2 are external attachments, port security, vlan aware VMs.... and so on, but it is not specific to ML2. 16:31:06 rkukura: perhaps the item on moving drivers out of tree can be removed and let it be at Neutron level 16:31:30 Sukhdev: perhaps, but maybe we’d need to stabilize the driver API to enable that 16:32:09 rkukura: Yes, that will be a general issue across the whole Neutron APIs 16:32:17 Sukhdev: good point 16:33:05 Let’s revisist this next week then 16:33:13 #topic Open Discussion 16:33:44 I’m happy to wrap up early an have some lunch before my next meeting ;) 16:34:12 Anyone want to discuss anything ML2 related? 16:34:56 how about splitting drivers dir into type and mech in Kilo? 16:35:16 we see many drivers in the same dir now. 16:35:18 amotoki: what about extension drivers? 16:35:36 rkukura: ah.. just forgot it. 16:36:26 amotoki: I think we’ll be seeing more cases where vendors have their own type drivers for their fabrics, etc., so these might be more coupled with the mechanism drivers. Similarly with vendor-specific extensions 16:36:52 I am thinking similar. one point i am not sure is how to place files if a driver is related to multiple drivers (e.g, ext and mech ) 16:37:04 Maybe we should organize into vendor subdirectories, along with a directory for the standard type drivers? 16:37:42 The nice thing about using entrypoints for drivers is that we can reorganize the code without breaking configs 16:37:47 it might be better. 16:38:22 Any other discussion topics? 16:38:55 Thanks everyone! 16:39:25 #endmeeting