15:00:33 #startmeeting neutron_l3 15:00:34 Meeting started Thu Jun 30 15:00:33 2016 UTC and is due to finish in 60 minutes. The chair is mlavalle. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:35 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:38 The meeting name has been set to 'neutron_l3' 15:00:39 o/ 15:00:41 hi 15:00:45 hi 15:00:48 hi 15:00:49 hi 15:00:54 #chair tidwellr carl_baldwin 15:00:54 Current chairs: carl_baldwin mlavalle tidwellr 15:01:02 hi 15:01:09 * njohnston has a fence repair guy coming; mfranc213 will represent FWaaS and the L3 agent extension work. 15:01:17 Hi everybody 15:01:31 njohnston: noted. thanks 15:01:44 Hi 15:01:52 #topic Announcements 15:01:57 hello 15:02:01 Agenda is here: 15:02:16 #link https://etherpad.openstack.org/p/neutron-l3-subteam 15:02:53 Newton-2 is just around the corner. You can find the official Newton delivery schedule here: 15:03:11 #link http://releases.openstack.org/newton/schedule.html 15:03:23 So it is July 11 - 15 15:04:10 Also a reminder to those who plan to submit a talk proposal for the Barcelona summit. The submission deadline is July 13th 15:05:10 Finally a reminder of the Neutron mid-cycle meeting in Cork Ireland, August 17th - 19th. The etherpad is here: 15:05:22 #link https://etherpad.openstack.org/p/newton-neutron-midcycle 15:05:31 I'm all booked. 15:05:54 Any other annoucements from the team? 15:06:03 carl_baldwin: Great! 15:06:41 ok, moving on 15:06:47 #Topic Bugs 15:07:24 First up is https://bugs.launchpad.net/neutron/+bug/1564335 15:07:24 Launchpad bug 1564335 in neutron " [Pluggable IPAM] delete subnet in ml2 plugin does not comply with pluggable ipam (deletes ip allocations directly from db)" [High,In progress] - Assigned to Pavel Bondar (pasha117) 15:07:59 I don't think Pavel is online today 15:08:29 I was looking at it a little last night. I don't think we need to expose this through the API. 15:08:34 I'll finish my feedback soon. 15:08:53 perfect, thanks! 15:09:03 any other comments about this one? 15:09:44 ok, moving on... 15:09:59 Next one is https://bugs.launchpad.net/neutron/+bug/1596075 15:09:59 Launchpad bug 1596075 in neutron "Neutron confused about overlapping subnet creation" [High,Incomplete] 15:10:36 Last night I tried to replicate this bug. I wrote a small script, that I posted in the bug itself 15:11:04 Yeah, something strange is going on. 15:11:15 Ran the script many many times, couldn't replicate 15:11:31 I looked through the code and can't think of anything but duplicate API requests that would cause it. 15:11:34 Could it be the orchestration stuff? 15:12:09 The reason of using a script was to remove that element and sure enough the bug didn't show up 15:12:14 I don't know. Could a load balancer duplicate requests? 15:12:47 I don't think so 15:13:43 I'll look today in the logs the reporter posted there to see if I find something. I'll comment my findings 15:13:51 Well, I think we need some more input on this one. 15:14:18 Agree. Let's see if we can get more data 15:14:41 ok 15:15:04 Those are all the high priority bugs we have 15:15:11 Any other bugs from the team? 15:15:39 There's a new one. 15:15:53 #link https://bugs.launchpad.net/neutron/+bug/1597561 15:15:53 Launchpad bug 1597561 in neutron "L3 agent allows multiple gateway ports in fip namespace" [High,In progress] - Assigned to Carl Baldwin (carl-baldwin) 15:15:59 I filed it and I'm working on it. 15:16:19 It is High because of the detrimental effect it has on the network. It isn't very common though. 15:16:55 ok, you filed it after I updated the etherpad yesterday. I'll add it 15:17:18 I think I'm going to start by adding code to make sure there is only one fg port after adding one. 15:18:03 That's all on this one. 15:18:17 Thanks! 15:19:37 ok let's move on 15:19:47 #topic Routed Networks 15:20:22 I think we're in pretty good shape. We need to stay on DHCP to see it through but I think we're close. 15:21:11 I'm getting good feedback from matt on my Nova patch. Once that is agreed on, I'll wrap up the Neutron side: the new extension to mark deferred IP ports. 15:21:20 I feel pretty good about those parts. 15:21:37 There is the service subnets code. How is that john-davidge ? 15:21:41 haleyb: ^ 15:22:13 carl_baldwin: Ball is currently in haleyb's court on that 15:22:14 carl_baldwin: i've been working on re-basing to master, need to finish that up 15:22:43 haleyb: ping me anytime you need a review. 15:23:24 will do 15:24:37 The big risk is Nova scheduling. 15:24:47 We're stuck behind a long pole there. 15:28:08 mlavalle: Anything to report? 15:29:01 yeah, the work that johnthetubaguy is doing to prep allocate for instance to it is not priority 15:29:13 and it freezes today 15:29:21 yeah, thats as we expected, sadly 15:29:37 there is always the exception process, but its a long shot 15:31:05 johnthetubaguy: so from your point of view, the g-r-p and scheduler work will be done in this ccyle? 15:31:30 johnthetubaguy: in other words, will the scheduler will be using g-r-p to shedule instances? 15:31:31 hmm, I hope will make enough progress to unblock you this cycle 15:31:36 johnthetubaguy: The consequence is that we pay a much higher price when we race for that last IP on a segment, right? 15:31:43 the external interface is a little tricky 15:31:47 yeah 15:32:00 well, its worse, probably no scheduling to start with 15:32:44 johnthetubaguy: We're not giving up on that yet. 15:34:01 johnthetubaguy: with a little of your guidance, could we continue working on the moving of allocate for instance to the conductor? 15:34:43 At least one of our early adopters has their own solution for scheduling. Maybe the rest can just monitor IP usage and keep on top of it until we wrap it up. 15:36:31 mlavalle: I think that's about all, right? 15:36:42 yeah, let's move on... 15:36:51 cores please look for reviews on the etherpad. 15:37:06 #topic BGP Dynamic Routing 15:37:16 hi 15:37:30 I put up https://review.openstack.org/#/c/335310/ 15:38:00 I think this is what it takes to get the dsvm jobs to pass, it still needs a lot of work 15:38:23 I could you some pointers on how to get these jobs running 15:39:04 we're getting late in the cycle and have code starting to be proposed against neutron-dynamic-routing, and still are missing important check jobs 15:40:09 tidwellr: Still needs testing work? 15:40:50 tidwellr: I can look at it once those are worked out. 15:40:52 carl_baldwin: no, we just need the check jobs running 15:41:20 carl_baldwin: I can run the tests locally, they just don't run in the gate yet 15:41:41 tidwellr: I mean the failing unit tests. 15:42:30 carl_baldwin: somehow the formula of adding tempest to test-requirements.txt and creating the tempest plugin makes the unit tests fail 15:43:00 tidwellr: Interesting. 15:43:07 carl_baldwin: a very suprising turn of events, the changes I made in that patch shouldn't affect the unit tests at all 15:43:31 carl_baldwin: it does something to the imports that makes things go haywire 15:43:52 I don't understand it all yet and I'm just fumbling around at the moment :) 15:44:17 if someone is looking to pitch in, this is a great place to start 15:45:10 that's all I had 15:45:35 Thanks for the update tidwellr! 15:46:00 #topic FWaaS 15:46:03 Regarding BGP EVPN, the one think I was not clear on is neutron client versus OSC 15:46:15 What is the status of OSC for neutron dynamic routing? 15:46:18 #undo 15:46:19 Removing item from minutes: 15:47:03 tidwellr: ^^^ 15:47:34 mickeys: neutron client vs OSC is still unclear to me too 15:47:57 tidwellr: Do we know if OSC will be addressed in this cycle? 15:49:06 mickeys: I thought we were moving in that direction, but it seems like it's moving slowly 15:49:21 mickeys: I have to confess I'm not following it too closely 15:49:24 tidwellr: Anyone else I should check with? 15:50:10 mickeys, tidwellr: I would start with rtheis 15:50:21 mlavalle: OK, thanks. That is all I had. 15:50:36 He may not be the person for this specific topic, but I am sure he can point us in the right direction 15:50:55 #topic FWaaS 15:51:02 hello everyone. njohnston and i recently refactored the l2 agent extension code to generalize it for both l3 and l2 extensions. i'll be removing the WIP with the next PS; and at this time we hope to get feedback on this change (understanding that we'll move parts of it to neutron-lib at some point). https://review.openstack.org/#/c/329701 15:51:05 mfranc213: the floor is yours 15:51:11 hi :) 15:51:30 so in addition to hoping to get review on that, 15:51:46 nate and i had two questions, on the above change and as we move forward in implementing the fwaas l3 agent. (if people have feedback maybe they can post to #openstack-fwaas as these dicussions are there from earlier today.) 15:51:55 first: what do people think of the idea of collapsing the [L2/L3]AgentCoreResourceExtension classes into one? 15:51:59 and ... 15:52:07 second: does/should the l3 agent use the same RPC callbacks library that the l2 agent uses? (we couldn't see that it did--maybe our newness to neutron.agent.rpc versus neutron.api.rpc lies at the root of this question.) 15:52:30 sorry for pasting so much... 15:54:07 mfranc213: can these questions be answered in the review to the patchset? 15:54:15 mfranc213: Is it ready for reviews? 15:54:37 Oh, I missed a line up there. Hold on... 15:54:38 yes, they can, and yes, it is. i need to remove the WIP. 15:55:20 mfranc213: I would ask those questions as commenbts in the relevant parts of the patchset 15:55:36 okay, i will. thank you! 15:55:42 to draw attention from reviewers 15:55:56 yes, makes perfect sense. 15:56:00 I can take a look at the patch set soon. 15:56:08 great. thank you. 15:56:17 mfranc213: anything else? 15:56:33 nope, that's it. 15:56:48 mfranc213: thanks for the update! 15:56:54 my pleasure 15:57:04 #topic Conversion to pluggable IPAM 15:57:37 I am on the hook for the IPAM docs - just want to let all know I won't be able to get to them until the week after next. 15:58:00 johnbelamaric: that'll be fine. 15:58:15 carl_baldwin: ok, good 15:58:40 johnbelamaric: Thanks for taking that on. 15:58:45 anything else on this topic? 15:59:28 ok.... 15:59:40 move to osc is slower than we would like but plan is in devref within neutronclient and amotoki is working on OSC plugin which I think will be direction for dynamic routing cli 15:59:51 mlavalle: not now - pavel_bondar is not here and I haven't had a chance to sync with him. i think we are out of meeting time now 15:59:58 Time is over 16:00:04 Bye. 16:00:06 Thanks for attending! 16:00:11 bye 16:00:13 #endmeeting