14:09:09 #startmeeting neutron_routed_networks 14:09:10 Meeting started Tue Aug 9 14:09:09 2016 UTC and is due to finish in 60 minutes. The chair is mlavalle. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:09:12 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:09:14 The meeting name has been set to 'neutron_routed_networks' 14:09:20 #chair jonh-d 14:09:21 Warning: Nick not in channel: jonh-d 14:09:22 Current chairs: jonh-d mlavalle 14:09:35 #chair john-davidge 14:09:37 Current chairs: john-davidge jonh-d mlavalle 14:09:50 #chair carl_baldwin 14:09:51 Current chairs: carl_baldwin john-davidge jonh-d mlavalle 14:10:13 in case carl_baldwin jojns later :-) 14:10:13 mlavalle: Thanks 14:10:24 Agenda is here: 14:10:27 #topic Announcements 14:10:31 #link https://etherpad.openstack.org/p/routed-provider-networks-notes 14:10:40 Sorry. /me totally late. 14:10:55 Mid-cycle is next week! 14:10:57 #link https://etherpad.openstack.org/p/newton-neutron-midcycle 14:10:58 carl_baldwin: I started the meeting already 14:11:00 Thanks for starting mlavalle 14:11:12 carl_baldwin: Hi Carl, it's all yours :) 14:11:15 I was thinking 9am 14:11:35 mlavalle, could you drive for a bit? 14:11:58 carl_baldwin: john-davidge is driving. this is the day I have an audio meeting at the same time 14:12:12 carl_baldwin: I'm happy to keep driving 14:12:12 john-davidge: carry on please 14:12:29 okie dokie, any other announcements? 14:12:46 #topic Reviews 14:12:55 #link https://review.openstack.org/#/q/status:open+topic:bp/routed-networks 14:13:06 #link https://review.openstack.org/#/q/status:open+topic:bp/service-subnets 14:13:21 #link https://review.openstack.org/#/q/status:open+topic:bp/neutron-routed-networks 14:14:15 Any reviews anybody would like to highlight? 14:14:40 Earlier this week I pushed https://review.openstack.org/#/c/350665/ 14:14:58 This patchset adds the segment_id to the ports dictionary 14:15:13 * john-davidge looks 14:15:27 it will be used by Nova when scheduling instances 14:15:36 o/ 14:16:29 mlavalle: Looks like it's had some review attention in the last couple days. Do you have everything you need to roll another patchset? 14:16:38 * carl_baldwin 's laptop not connecting to net 14:16:47 john-davidge: no. I just wanted the team to be aware of it 14:17:02 mlavalle: Ok, thanks 14:17:07 carl_baldwin: don't worry. we are managing well 14:17:28 The one I wish to highlight as a top priority is https://review.openstack.org/#/c/317358 14:17:50 There are a lot of patches in the review queue with jenkins failures and merge conflicts, so if you own any of those please make sure they're up to date 14:18:29 carl_baldwin: if that helps, I'll take a look later 14:18:36 today 14:19:08 I pushed a new patch for service subnets CRUD today - https://review.openstack.org/#/c/337851/ 14:19:25 It addresses armax's concerns 14:19:53 I'll have a new patchset for port creation soon - https://review.openstack.org/#/c/350613/ 14:20:12 john-davidge: Thanks for the quick attention on that. 14:20:36 carl_baldwin: Thanks for the plentiful reviews! 14:21:34 john-davidge: keep them coming 14:21:35 The python-neutronclient change is ready to merge as soon as the CRUD patch does https://review.openstack.org/#/c/342976/ 14:21:40 Thanks haleyb! 14:22:09 Any other reviews to highlight? 14:22:45 np, i never saw this meeting reminder, getting old or something 14:22:56 thank you for getting the other patches in shape 14:23:12 haleyb: np :) 14:23:16 haleyb: or maybe looking forward to vacation week :-) 14:23:26 #topic Docs 14:23:40 * mlavalle will miss his DEC buddy in Ireland 14:23:42 carl_baldwin: Any doc updates? 14:24:12 Not like I wanted to. I didn't get as much done on docs. 14:24:17 Still working... 14:24:36 Maybe something we want to tackle next week in Cork? 14:24:37 I might just have to acknowledge that docs will come after N-3 14:24:55 mlavalle: We could. Sometimes docs writing is a good on-plane activity. 14:25:04 ... unless I'm wiped 14:25:05 mlavalle carl_baldwin: Yeah, I'm sure we could find some time to sit down together on docs next week 14:25:33 carl_baldwin: I am willing to help with that, if you need help 14:25:39 I'll keep trying to get a rough draft, or at least an outline ready for then. 14:26:21 If not, we'll have to start from very little. 14:26:21 carl_baldwin: happy to take a look whenever you need reviews etc 14:26:51 #topic Testing 14:26:59 Hi 14:27:11 so we made good progress this past week 14:27:24 I'll let yb update the team 14:28:32 maybe yb left the meeting 14:28:39 Hi, I think I need write the test cases to test the instances in two different segmentations 14:28:53 we have scenario test up for review https://review.openstack.org/#/c/347188/ 14:29:10 mlavalle: I'll review it. 14:29:16 we are getting reviews 14:29:17 I have almost finish testing the pre-port creation and insufficient ip addresses scenarios 14:30:22 john-davidge: that's probably it on this subject this week 14:30:29 yb: mlavalle: Sounds good, thank you 14:30:36 #topic Integration with Nova Scheduler 14:30:38 yb: thanks for your hard work! 14:30:50 Thanks:) 14:30:52 Hi again 14:30:56 yb ++ 14:31:08 carl_baldwin::) 14:31:30 On this topic, as I mentioned earlier I worked on publishing the segment id on ports reesponses 14:32:04 This is important for the Nova scheduler to select hosts where an instance will be schduled 14:32:31 I have also been playing with the Generic Resource Pools API in Nova 14:33:01 That is the API that we will use to inform Nova scheduler of our routed networks / segment topology 14:33:28 mlavalle: Anything you'd like to get some eyes on? 14:33:52 The key idea I want to share with the team is that I want to use is the association of subnets with a segment 14:34:17 #link https://github.com/openstack/neutron/blob/master/neutron/plugins/ml2/plugin.py#L972 14:34:53 In my opinion we want only to send segments data to Nova when it is indeed a segment associated with a routed network 14:35:27 mlavalle: ++ 14:35:39 Therefore I want to use the notification of subnet create (when segments are associated to subnets) to trigger the update of Nova 14:36:12 And that got me thinking also about segment host mapping 14:36:43 Maybe we want to simplify it and only add segments to the host mapping when they are associated to subnets 14:36:56 that may be a follow up port N-3 14:37:11 we can discuss in Cork 14:37:37 that's it on this topic 14:37:51 mlavalle: Sounds like good progress, thanks! 14:37:58 Let's talk about it next week. 14:38:04 #topic Service Subnets 14:38:22 We touched on this already, but I'll highlight the review topic again: 14:38:36 #link https://review.openstack.org/#/q/status:open+topic:bp/service-subnets 14:39:12 Progress has been very good I think. Everything will start falling into place once https://review.openstack.org/#/c/337851/ merges 14:39:18 I'll have another look today. 14:39:23 haleyb: Anything to add? 14:40:54 john-davidge: no, i think it's in good shape. There will probably be some l3-agent work once we get the ipam changes in, hopefully not too bad 14:41:46 haleyb: what needs to change in the agent? 14:41:58 haleyb: I hadn't anticipated any, what did you have in mind? 14:42:52 carl_baldwin: i'm thinking there might be a small tweak when the router IP is in a different subnet than the floating IP, might be wrong but it gets to the email sent to the list from Hui (?) 14:43:39 * carl_baldwin behind on ML again 14:44:31 haleyb: Yeah I remember the email, will see if I can dig it up again. Thanks 14:44:48 haleyb: Let's follow up next week 14:45:03 it was from last week, but relates to having an existing subnet route when adding the floating IP 14:45:25 Anyone have a link? 14:45:54 carl_baldwin: i could be completely wrong too, looking for the link 14:46:38 http://lists.openstack.org/pipermail/openstack-dev/2016-July/100494.html 14:47:22 Right, I do remember that one (even responded at one point). 14:48:09 I might have to reread to see if there is still any issue. Also, we'll need to do some full testing in a multi-node system. 14:48:24 So, there might yet be some work to do. 14:48:34 carl_baldwin: ++ 14:48:34 Thanks for pointing that out, haleyb 14:48:42 right, it's more of a concern until we get some testing in, didn't want us to think it was clear sailing from here 14:48:55 haleyb: ++ 14:49:39 haleyb: Thanks for raising that, let's move on to the next topic 14:49:42 #topic Client 14:50:00 rtheis: Hi 14:50:04 hi 14:50:23 rtheis: Any client work you'd like to highlight? 14:50:42 no, just waiting for create and delete segment to merge 14:51:07 https://review.openstack.org/#/c/317358 14:51:50 rtheis: Thanks 14:51:58 yw 14:51:59 * john-davidge reminds himself to go and review that patch 14:52:15 #topic Open Discussion 14:52:22 That's it, anything we've missed? 14:52:26 rtheis: thanks 14:52:58 Now that we have a good start on service subnets, I'm starting to think about how to do floating ips on a routed network. 14:53:32 We need to be able to associate a fip with a port on the routed network instead of through a neutron router. 14:53:43 And, it will require routing. 14:54:02 is this something we want to do in Newton? 14:54:09 No. 14:54:14 ok 14:54:23 Just thinking about it. 14:54:40 Something to keep my mind busy while running, biking, swimming, etc. 14:55:00 or doing yoga 14:55:03 :-) 14:55:11 mlavalle: I should do that too. 14:55:53 I can't usually think about anything other than the TRON Legacy soundtrack while doing those things 14:56:09 john-davidge: lol 14:56:10 BTW, I was running blissfully this morning at 8am thinking that our meeting was an hour later. Apologies again for that. 14:56:11 neutron thinking comes just as im trying to fall asleep :) 14:56:35 john-davidge: you mean it puts you to sleep. 14:56:54 carl_baldwin: Quite the opposite! 14:56:58 john-davidge: that happens to me as well some times. and no, it wakes me up 14:57:05 carl_baldwin: ^^^ 14:57:11 * john-davidge hasn't slept since learning what a Neutron was 14:57:25 john-davidge: I see, neutron induced insomnia . I get that too. 14:57:37 nii 14:57:41 #action carl_baldwin to take his morning run an hour earlier next week 14:58:13 And speaking of next week, I imagine many of us will be travelling for the midcycle on tuesday 14:58:19 will we cancel the meeting? 14:58:47 john-davidge: I will be in HPE's Galway office. So, I could go either way. 14:59:13 I'll be on a plane I believe 14:59:27 I will be already in Cork, so I can attend if needed 14:59:50 Ok, we're out of time so lets discuss on the ML if neccessary 14:59:54 thanks everyone! 15:00:06 #endmeeting