15:00:24 #startmeeting neutron_l3 15:00:27 hi 15:00:29 Meeting started Thu May 28 15:00:24 2015 UTC and is due to finish in 60 minutes. The chair is carl_baldwin. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:31 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:33 The meeting name has been set to 'neutron_l3' 15:00:36 #topic Announcements 15:00:37 hi 15:00:43 #link https://wiki.openstack.org/wiki/Meetings/Neutron-L3-Subteam 15:00:57 hi 15:01:11 Great time at the summit last week. A lot was discussed and we’ve got a lot to do. 15:01:24 +1 15:01:33 Liberty-1 is coming right up. 15:01:40 #link https://wiki.openstack.org/wiki/Liberty_Release_Schedule 15:01:55 hi 15:02:02 * HenryG lurks 15:02:04 We’ve got a little over 3 weeks. 15:02:06 hello 15:02:30 Any other announcements? 15:02:53 #topic Bugs 15:03:20 Looks like we need to do some triage. I see 5 new bugs tagged. 15:04:02 Bug #1457900 has been hot lately. 15:04:02 bug 1457900 in neutron "dhcp_agents_per_network > 1 cause conflicts (NACKs) from dnsmasqs (break networks)" [Undecided,In progress] https://launchpad.net/bugs/1457900 - Assigned to Kevin Benton (kevinbenton) 15:04:54 https://review.openstack.org/#/c/185368/ , not a bug but maybe we can sort some of the issues (dont know if right now is the correct time) 15:04:54 This is one of those “I can’t believe no one noticed before” things. 15:05:47 gsagie: We’ll fit this in the agenda. 15:06:05 carl_baldwin: there was a response on the dnsmasq list about a udhcp change that fixed this, so might be unique to cirros 15:06:46 the client should only care about a NAK if from the same server that sent a lease 15:06:53 The bug report has a comment saying ubuntu / centos is also affected. Maybe it is just newer versions of each distro. 15:07:22 haleyb: I think I agree but apparently the dhcp client disagrees. :) 15:07:23 we all know how slow centos is at updating 15:08:03 Unfortunately, it is the dhcp client’s vote that counts, I think. :( 15:08:39 -1 on the client :) 15:08:54 From me too. 15:08:59 Any other bugs to discuss? 15:09:12 kevin's latest fix might be fine then, but we can take that to the review 15:09:58 Sorry I did not get around to fixing up the team page to update for this week. 15:10:08 #topic bgp-dynamic-routing 15:10:21 devvesa_: tidwellr: vikram: hi 15:10:21 hi 15:10:39 How is the rebase of the old patches going? 15:10:50 hi 15:11:08 devvesa_: did you have one more patch to share? 15:11:53 Also vikram mentioned neutronclient and horizon to me before the meeting. I wouldn’t expect that we have any Horizon work going on yet but I was wondering if we had a neutronclient patch. 15:12:09 I'm starting one 15:12:31 not up for review yet 15:12:31 As per the plan, once the rebasing for https://review.openstack.org/#/c/115554/ (introducing entities) is done I can provide other patches 15:12:59 'Yes we got write the code 15:13:24 Ryan is taking care of that 15:13:24 https://review.openstack.org/#/c/115554/ needs a little more TLC, and I think we need to be sure the spec is approved still 15:13:44 +1 15:14:19 Okay. Let’s be sure the spec is in good shape. I’ll take another look today. 15:14:31 Carl: So Horizon changes are not required? 15:15:24 So far, I haven’t seen much pressure to provide Horizon changes. But, I’m not sure if there is a requirement that is not being enforced. 15:16:01 whom we should confirm? 15:16:15 I’ll ask Kyle today when I talk to him. 15:16:30 thanks carl 15:16:53 All I know is that you don’t want me messing around with a GUI. 15:17:08 Okay, let’s move on... 15:17:13 #topic neutron-ipam 15:17:43 johnbelamaric: pavel_bondar: I want to check in on the patches. Are they once again ready for review? 15:18:10 carl_baldwin: yes 15:18:38 thanks all for reviews on the reference driver, that patch merged :) 15:18:52 johnbelamaric: Excellent! That was a lot of work. 15:18:55 driver loader patch is next, I think it's ready to go 15:19:15 johnbelamaric: do you happen to have a link for the current next patch up for review? 15:19:37 https://review.openstack.org/#/c/147479/ 15:19:43 #link https://review.openstack.org/#/c/147479/ 15:20:11 johnbelamaric: Thanks. That’ll help meeting attendees find it. I will review today and encourage others to review it too. 15:20:27 Also would like more feedback on the refactor patch 15:20:30 #link https://review.openstack.org/#/c/153236/ 15:20:52 which is next in line after the loader 15:21:48 johnbelamaric: ack, thanks for the link. 15:21:54 Anything else on ipam? 15:22:16 #topic dns 15:22:27 Nope, just need reviews at this point 15:22:33 mlavalle: Anything to discuss? 15:23:01 BTW, the dns stuff has been on the back burner for over a year now I think. It is good to see some interest in it. 15:23:12 just update the team that I got support from the nova PTL to move ahead with this and i am starting the code 15:23:44 also started participating in the designate weekly meetings yesterday. I'll keep doing it 15:23:45 mlavalle: Great! We should be able to get a fast start on this. 15:24:09 that's all so far 15:24:36 mlavalle: Thanks. I look forward to seeing the patches. 15:25:18 #topic address scopes 15:26:15 I’ve started the code for this but don’t have anything up for review yet. I’m starting with RPC changes to send the address scope to the L3 agent. Will have it up soon, I think. 15:26:34 carl_baldwin: I think there is still a lot of definition to do 15:26:58 carl_baldwin: at least, there are a number of open questions in my mind 15:28:16 johnbelamaric: Do you want to discuss them here? 15:28:53 * carl_baldwin notices he has uncommitted replies on the spec. Sorry. Will visit them soon. 15:29:17 carl_baldwin: I'll ad this topic to the team wiki and we can populate it as we learn / discuss more 15:29:28 carl_baldwin: they are on the review. The main question is how we are planning on using them in the future and their relationship to route advertisement as well as routers specifically as routes import/export 15:29:55 mlavalle: great, i am putting together a diagram to help keep it straight (helps me anyway) 15:30:20 johnbelamaric: Also, the code I’m starting is pretty basic stuff that I don’t think is in question. You might disagree but I thought the code could be started safely. 15:30:49 johnbelamaric: I’d like to see that diagram. When do you plan on having it? 15:30:58 carl_baldwin: Sure, no problem. Always good to get a head start. it can also help clarify thinking to have some code down 15:31:11 carl_baldwin: today I hope 15:31:53 Okay, feel free to send it my way. Maybe to the ML for others’ benefit too. 15:32:03 Ok 15:33:43 johnbelamaric: Let’s also start a discussion on the open questions you still have later. 15:34:09 Anything else to discuss now on this? 15:34:18 carl_baldwin: On the ML? Or on the review? I'll look forward to your resonses on the review and see where we land 15:35:15 johnbelamaric: Let’s start on the review and maybe in the irc channel to begin with. 15:35:21 ok 15:35:35 I’ll read over my responses after the meeting and we’ll go from there. 15:36:12 ok 15:36:16 #topic routing-networks 15:36:21 skipping this topic for today 15:36:24 #topic ipv6 15:36:54 I know of the PD work that is ongoing. Anything to discuss? john-davidge ? 15:37:38 The patch-split is finished now, dependency tree starts here: https://review.openstack.org/#/c/158697/ 15:38:00 Reviews appreciated! :) 15:38:33 carl_baldwin: I saw your follow-up patch, what did you think to my comment? 15:39:02 john-davidge: I have not seen it yet. 15:39:44 * carl_baldwin just read it 15:40:39 john-davidge: I noticed in the docstring for delete_conntrack_state that it said it should always be called soon after deleting the address. I didn’t see any need to call it alone. 15:41:39 I’ll respond on the review. 15:41:44 carl_baldwin: In this case, yes, it's always called after deleting the address. Can't think of a future use case that might need to do it otherwise off the top of my head but one may exist. 15:41:53 carl_baldwin: ok sure 15:42:06 john-davidge: Nothing prevents a future refactor if the need comes up. 15:43:02 sc68cal: HenryG: et al: Anything else around ipv6 that could use some meeting time to discuss? 15:43:09 carl_baldwin: True, I guess I tend to lean towards granularity where possible. The testing session last week has hammered that into me! 15:43:45 Not that I can think of for today 15:44:04 john-davidge: I could go either way. It might be better from a testing point of view (which I haven’t had time to deal with in that patch). 15:44:26 We can start beefing up the ipv6 gate tests now that cirros supports ipv6 better 15:45:43 +1 can't think of anything 15:46:02 HenryG: Anyone slated for that work? Or, are you just throwing it up? 15:46:37 I've got one bug assigned to me around that, I'll have to dig up the # 15:46:54 https://bugs.launchpad.net/tempest/+bug/1401726 15:46:54 Launchpad bug 1401726 in tempest "Tempest IPv6 scenario tests use IPv4 and floating IPs to connect and test" [High,Confirmed] - Assigned to Sean M. Collins (scollins) 15:46:54 carl_baldwin: I know of some people writing test cases, but have not synced up with them recently 15:49:05 Let’s keep this on the agenda. Maybe we can report back next week on progress of beefing up tests. 15:49:21 #topic Open Discussion 15:49:25 carl_baldwin: i'll this bug to our wiki page 15:49:31 gsagie: still around? 15:50:23 yes 15:50:34 #link https://review.openstack.org/#/c/185368/ 15:51:08 I responded to most comments, the only thing i still see for this patch is to change the names 15:51:51 You’re right DistributedDvrRouter doesn’t sound quite right. 15:52:11 I explained why Inheritance model is the most appropriate for this step (you can read it there) and the next step i would try to think how to change the model to a composition where we could attach functionalities to the router 15:52:21 In the initial design, the DVR team called that part the IR. I’m trying to remember what that stood for. 15:52:34 but that sounds a bigger change and i like to do this in smaller steps 15:53:00 gsagie: I’m okay if you want to take a smaller step. It should be marked with a #TODO comment. 15:54:09 ok great, so what i see for this patch is just better names picking (it will be easier with the composition task as these are becoming "Handlers", at least the way i see it, rather then router types) 15:54:45 #link https://docs.google.com/document/d/1jCmraZGirmXq5V1MtRqhjdZCbUfiwBhRkUjDXGt5QUQ/edit 15:55:03 They called it IR for InternalRouter. 15:55:48 Off the top of my head: DvrInternalRouter and DvrCentralRouter 15:56:19 Or DvrDistributedPart and DvrCentralPart 15:56:46 Hopefully these give you some ideas of your own. 15:57:04 yeah thanks i will submit a fix patch soon 15:57:27 and will send some small design once i have it for the composition task 15:57:39 gsagie: Thanks, great work. For the most part, you did exactly what I had in mind. 15:57:56 Anything else for our remaining 2 minutes? 15:58:09 not from me, thanks and got to go..good day everyone 15:58:45 Thanks everyone! 15:59:04 #endmeeting