15:02:02 #startmeeting Routed Networks 15:02:03 Meeting started Tue May 24 15:02:02 2016 UTC and is due to finish in 60 minutes. The chair is mlavalle. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:02:04 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:02:07 The meeting name has been set to 'routed_networks' 15:02:37 So we will still have the meeting, right? 15:02:54 M sorry my nick isn't changing 15:03:01 Carl had an unexpected meeting 15:03:35 Let's have the sync up 15:03:41 xiaohhui, blogan: carl_baldwin has an unexpected meeting. He won't be able to make it to this one 15:04:03 mlavalle: alrighty 15:04:19 xiaohhui, blogan: we can definetely have the meeting today ourselves 15:04:40 ok 15:04:52 Count me in as well mlavalle (it's reedip here) 15:05:19 \o/ 15:05:22 xiaohhui, blogan: Carl is going to send a message to the ML, proposing to do an email update 15:05:45 HoloIRCUser3: interesting name reedip 15:06:01 Its from an App on the fone 15:06:08 reedip: are you still working on this patch https://review.openstack.org/304647 15:06:10 xiaohhui, blogan, HoloIRCUser3, jschwarz: but we can go on here.... 15:06:17 Stupid app doesn't let me change the NICK 15:06:20 #topic Announcements 15:06:24 mlavalle: ok lets do it 15:06:44 HoloIRCUser3: sounds like a limited app :) 15:06:57 Keep in mind that Newton-1 is a little more than a week ago 15:07:29 xiaohhui : I will start from tomorrow... I shifted from Japan to India in between the Austin Summit so my whole work was haywired 15:07:44 Any other announcements from the team? 15:08:22 If not, let's move on... 15:08:46 #topic Progress Reports 15:09:40 As I said earlier, carl_baldwin is going to send his progress report to the ML. So, let's start with DHCP 15:09:46 blogan: you are up 15:10:17 so I've per the ML thread I reworked the scheduling a bit back to where I initially had it 15:10:52 https://review.openstack.org/#/c/311931/ 15:11:17 it's still a wip and its not pretty and it still needs tests but i ddidn't want to write tests until the direction is solidified 15:12:44 the not pretty part is because we need to bind a dchp agent to a segment ONLY when segment's exist, otherwise we need to do it by the network. however, the base scheduler and base resource filter's interface don't expect that 15:13:06 so that interface doesn't work in this case, and if i change that interface it'll probably break out of tree schedulers like bgp 15:13:36 its a very hairy problem 15:14:08 blogan: do you need any help? 15:14:15 in the meantime ive been working on the agent side of things, to make it segment aware, and that is definitely shaping up to be a large effort 15:14:45 mlavalle: just help in the sense of needing some ideas on how to solve that particular issue 15:15:01 blogan: post ideas in the reviw? 15:15:07 review^^^ 15:15:12 mlavalle: yeah, that'd be good 15:16:11 blogan: I will take a look later today 15:16:21 ill comment where the ugliness is 15:16:25 on the review 15:16:51 will also take a look at it tomorrow 15:18:20 anything else blogan? 15:19:07 mlavalle: nope that's it for now, just want to reiterate the agent side of it will be more work than I expected nad probably any of us expected :) 15:19:30 blogan: will you need help with that piece? 15:21:08 mlavalle: I can probably break the tasks up into chunks so that other's can take them if they want, otherwise i can chip away at it 15:21:32 ok 15:21:38 let's move on 15:21:43 thanks mlavalle 15:21:50 #topic Host Segments mappping 15:22:08 ON this subject we made great progress last wekk 15:22:13 last week^^^ 15:22:51 great! 15:23:23 https://review.openstack.org/#/c/285548/ got +2ed by carl_baldwin and +1ed by xiaohhui . Thanks for all the help with reviews 15:24:18 after xiaohhui rechecked it last night, some test cases in neutron.tests.unit.agent.test_securitygroups_rpc started failing 15:24:34 I think it has been fixed 15:24:41 just a rebase can resolve it 15:25:17 xiaohhui: great, thanks for saving me some research time. I was so frustrated a few minutes ago :-) 15:25:21 hopefully there aren't conflicts 15:25:49 so moving on... 15:26:17 #topic Enable Create / Delete of Segments on Existing Network in ML2 15:26:25 xiaohhui: this is you, right? 15:26:30 yeah 15:26:38 The code is ready for review 15:26:45 https://review.openstack.org/#/c/317358 15:27:06 I hope you guys can reiview it 15:27:26 yeah will take a look today also 15:27:50 thanks for pushing this forward.... anything else on this topic? 15:28:09 I think there is nothing else 15:28:25 since xiaohhui has the mike already... 15:28:36 #topic OVN plugin 15:29:17 The patch from richard has been merged. https://review.openstack.org/#/c/312705/ 15:29:30 (Convert core plugin to ML2 mechanism driver) 15:30:18 But the ovn part is still no progress, I think it would be better to move on, once the ml2 code has settled. 15:30:35 xiaohhui: ok 15:30:52 Maybe we can update the host / segment mapping for ovn, once your patch has been merged. 15:31:51 There is a wip patch for it, I will re-evaluate it this week. https://review.openstack.org/#/c/302623/ 15:31:57 xiaohhui: yeah, I expect that to merge any time soon.... especially after your big help of a few minutes ago ;-) 15:32:08 :) 15:32:51 moving on then ... 15:32:52 Nothing else for OVN from me. 15:33:12 #topic Integration with Nova Scheduler 15:33:35 ON this front, I had a conversation with cdent earlier today 15:34:10 He is implementing the code for generic resource pools in Nova 15:34:43 There are several patches for this. The top of the stack is here: https://review.openstack.org/#/c/293104/ 15:35:03 I intend to follow this work closely and help with reviews where I can 15:35:45 Also johnthetubaguy has pushed revisions to https://review.openstack.org/#/c/313001 15:35:56 Maybe he cares to comment about it? 15:36:33 This is the refactoring of 'allocate_for_instance' in Nova compute that allocates network resources for instances 15:37:15 it is being refactored, some pieces to the conductor, pre-scheduling, so segments data con be used in the scheduling decision 15:37:36 In any case I intend to review that spec and help where I can 15:38:47 Finally, I attended yesterday's Nova Scheduler meeting and reminded people to post reviews for the Routed Networks spec for Nova: https://review.openstack.org/#/c/263898/ 15:38:53 ill review too, more for my own learning though :) 15:39:32 Several of them said they are going to review it: johnthetubaguy, bauzas, edleafe 15:39:52 that's it for this week on this topic 15:40:20 #topic L2 Adjacency Extension 15:40:52 HoloIRCUser3: are you still woking on https://review.openstack.org/#/c/304647/? 15:41:56 Mlavalle:yeah, will restart it tomorrow 15:42:07 Sorry for the delay 15:42:13 to begin with, it needs a rebase :-) 15:42:17 I shifted from Japan to India 15:42:31 So things have been a bit messy for now 15:42:42 ahhh. How do you like being back in the Mother Land? 15:43:24 Not different... having stayed almost a year in Japan makes me feel both are pretty good places to live and love... 15:43:35 :-) 15:43:38 :) thanks for asking though... 15:43:46 any thing else on this topic? 15:44:34 if not, let's move on... 15:44:37 Not from my end for now... wll be back in the meeting tomorrow... 15:44:58 #topic Client 15:45:21 rtheis: anything to report on client for Routed Networks? 15:45:27 richard mentioned to me yesterday that he may be not able to attend this meeting. 15:45:36 back from the dentist just in time 15:45:52 we have SDK patch merged for network segments 15:45:57 rtheis: lol.... you are not in pain right? 15:46:09 mlavalle: no pain, just a cleaning today 15:46:13 rtheis: lucky u... I connected via the fone while driving( now home) 15:46:37 I am going to rebase the CLI patch set and then it should be ready 15:47:13 * mlavalle stalled enough, so rtheis could make his report on time 15:47:50 nothing else from me on client 15:48:01 rtheis: thanks for the status report 15:48:05 yw 15:48:10 moving on... 15:48:32 #topic Open Agenda 15:48:39 did I miss anyone? 15:49:42 any other topics to bring to the team? 15:50:25 In that case, we will return 10 minutes to everybody's agenda 15:50:38 Thanks for attending and for your reports ! 15:50:43 #endmeeting