15:00:57 #startmeeting neutron_l3 15:00:57 Meeting started Thu Jan 8 15:00:57 2015 UTC and is due to finish in 60 minutes. The chair is carl_baldwin. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:58 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:01:00 The meeting name has been set to 'neutron_l3' 15:01:21 #topic Announcements 15:01:28 hi 15:01:34 #link https://wiki.openstack.org/wiki/Meetings/Neutron-L3-Subteam 15:02:08 #info Kilo-2 will be on February 5th 15:02:22 Any other announcements? 15:03:20 #topic Bugs 15:04:01 Looks like I need to triage a little but I don’t see any new bugs to discuss 15:04:20 #action carl_baldwin to triage new bugs 15:04:48 #topic L3 Agent Restructuring 15:05:20 I’ve been making a lot of progres on this this week. 15:06:14 #link https://review.openstack.org/#/c/143733 is about ready to merge I think. 15:06:26 amuller: ping 15:08:29 Yesterday, I started teasing out fip namespace logic. I would like eyes from Rajeev, mrsmith, and other dvr expertise here. 15:08:49 #link https://review.openstack.org/#/c/145565/ 15:08:51 sure - no prob 15:09:22 I will take a look at it 15:09:45 mrsmith: Swami: thanks. 15:10:43 Can use review of FW refactoring: #link https://review.openstack.org/140884 15:11:05 Will then work on event handlers with SridarK 15:11:19 I think my patch changes the logic a bit. I think it better acknowledges multiple external networks. 15:11:56 Swami and I are working thru issues already with multiple ext nets and dvr 15:12:12 all these refactors make for busy reabses 15:12:16 :) 15:13:20 mrsmith: I know. That’s why I want eyes on it right away. It proved too difficult to do the floating ip refactoring without first working on the fip namespace. 15:13:37 pc_m: I will look today. 15:13:44 carl_baldwin: Thanks! 15:14:08 #action carl_baldwin will review https://review.openstack.org/#/c/140884/ 15:18:01 mrsmith: Swami: feel free to ping me with your concerns. How has multiple ext-net progress been going? 15:18:19 carl_baldwin: we have a couple of patches that we are working on right now. 15:18:32 good - definitely need some dvr agent changes 15:18:37 The final one is the "floatingip-namespace" correction. I am currently working on it. 15:19:34 Do you have links to the patches handy? 15:20:11 #link https://review.openstack.org/#/c/142674/ 15:20:24 #link https://review.openstack.org/#/c/143567/ 15:20:43 These two above patches fixes the current exceptions that are raised. 15:21:10 As I mentioned we found an issue with the FIP namespace creation issues when there are more than one external network. 15:21:19 the main issue is handling separate fip-namespaces for each ext net 15:21:24 We are working on it and when we have a wip I will push it in. 15:22:33 mrsmith: You will definitely want to check out this patch in that context: https://review.openstack.org/#/c/145565/ 15:23:00 Hopefully we’re not going in incompatible directions. 15:23:03 carl_baldwin: okay 15:23:50 as you know, refactoring and bug fixing can be hard to do at the same time :) 15:24:31 but we have to make progress on the refactoring 15:25:16 mrsmith: I do know. That is why I’m trying to call attention to what I’m doing very quickly rather than waiting to post the final result later. 15:25:30 +1 15:25:46 mrsmith: Swami_: Yesterday afternoon, I added this: https://review.openstack.org/#/c/145565/3/neutron/agent/l3/dvr.py L46 15:26:49 +1 15:27:13 The new fip namespace class takes the external network id. Multiple instances are created. This was something that the old code wasn’t careful about. 15:27:38 Anyway, we should move on. Please ping me with your concerns. 15:27:48 yes - or just plain ignorant... adding an array is something Swami and I were thinking of 15:28:08 yup - thanks for working on this as always carl_baldwin 15:28:15 pc_m: Anything else? 15:28:24 carl_baldwin: Just looking at your changes in the fip patch. 15:28:25 no, i'm all set. 15:28:36 That was exactly what we were thinking on 15:28:53 I will review your patch completely, test it and will provide my inputs. 15:29:22 Swami_: At least we’re thinking along the same lines. I will appreciate that. 15:29:41 #topic neutron-ipam 15:30:04 johnbelamaric: hi 15:30:14 salv-orlando: ping 15:30:15 carl_baldwin: hello 15:30:47 All the relevant blueprints are in good shape, I think. 15:31:34 yes. what do we need to get your interface merged? 15:31:58 #link https://review.openstack.org/#/c/134339/ 15:32:26 most of our team was out for the holidays but we should be able to get going on refactor soon 15:32:45 johnbelamaric: I was hoping to see salv-orlando’s work show up as a dependant patch sometime soon. 15:33:20 johnbelamaric: Also, seeing a start on the refactoring of neutron as a dependent patch soon would be nice. 15:33:59 carl_baldwin: ok 15:35:09 I was thinking that starting both those efforts may show where the interface is deficient. 15:35:42 johnbelamaric: What do you think? 15:35:42 carl_baldwin: yes, good point - we'll get started ASAP 15:36:46 johnbelamaric: great! 15:36:52 carl_baldwin: might be conflicts with salv-orlando's work, but we will resolve when we see that 15:37:58 johnbelamaric: I’ll watch out for conflicts. Initially, there shouldn’t be much conflict as salv-orlando will be working on mostly new code. 15:38:36 johnbelamaric: anything else to discuss? 15:38:38 carl_baldwin: right, but depending if he works on integrating it to the existing Db plugin. anyway, the bulk should be non-conflicting 15:38:45 carl_baldwin: nope 15:38:51 I am not sure I am working on new code. I am moving existing code into new places 15:39:25 also - I am doing work dependent on patch #134339 and contextually removing existing IPAM logic in db_base_plugin_v2 15:39:54 nevertheless we should not stop each other fearing conflicts. Let's work independently and fix the conflicts when they arise 15:40:06 salv-orlando: ok, will do 15:40:50 salv-orlando: ok. I misunderstood a bit. I guess I thought that you would create new code without moving out the old implementation at first. But, I’m okay with this too. 15:41:41 salv-orlando: agree about conflicts. 15:41:50 carl_baldwin: If I did so then we would have the new and old implementation at the same time - with the new one untestable beyond unit test. Nobody would let us merge that. 15:42:13 But on the other hand we can still have a 3rd patch in the same patch series which does the cleanup in neutron and hooks the new logic 15:42:26 I mean does the "glueing" 15:42:50 salv-orlando: I’ll let you go about your business. Don’t want to tell you how to get it done. 15:43:14 salv-orlando: Anything else to update? 15:43:15 carl_baldwin: you can I won't get offended 15:44:08 carl_baldwin: I have nothing else on IPAM or anything l3 15:44:34 salv-orlando: thanks for the update. I look forward to the ipam patch(es) 15:44:37 I am dealing with the aftermath of setting dnsmasq minimum requirement to 2.67, but that's not something that should concern you here ;) 15:45:30 salv-orlando: I saw that going on. Hope you get that straightened out without too much trouble. 15:45:39 salv-orlando: johnbelamaric thanks. 15:45:53 #topic neutron-ovs-dvr 15:45:57 carl_baldwin: there is no trouble unless you have consumers insisting on using ubuntu 12.04 ;) 15:46:19 carl_baldwin: I need to drop off, mike will provide the update on dvr 15:46:38 Swami_: thanks 15:46:40 we already talked about multi-ext-nets 15:46:58 Rajeev and I are making progress on dvr-l3-ha 15:47:12 keeping up with the refactoring 15:47:54 Links to patches : https://review.openstack.org/#/c/139686/ 15:48:03 * carl_baldwin is just making life difficult everywhere for you, isn’t he? 15:48:05 https://review.openstack.org/#/c/143169/ 15:48:23 just keeping things interesting carl_baldwin ;) 15:48:40 dvr-l3-ha is still wip 15:48:43 carl_baldwin: our turn will come too :) 15:48:45 but we are close 15:49:07 the l2pop patch seems stalled 15:49:20 to fix l3-ha running with l2pop 15:49:33 we will need that to get fixed for dvr-l3-ha to work 15:50:17 mrsmith: link handy? 15:50:33 no.... let me try to find it 15:51:03 Rajeev_: Did you see the discussion earlier on https://review.openstack.org/#/c/145565 ? 15:51:50 carl_baldwin: no, but I will take a look at it and catch up from mrsmith 15:51:56 Rajeev_: I would value your feedback on it. 15:52:24 carl_baldwin: will do 15:53:05 here is the stalled l2pop patch: 15:53:23 Rajeev_: regarding HA/DVR, have you seen https://review.openstack.org/#/c/143733 ? 15:53:45 https://review.openstack.org/#/c/141114/ 15:54:49 carl_baldwin: This is the refactoring into derived classes for dvr agent side. 15:55:05 #action carl_baldwin will review https://review.openstack.org/#/c/141114/ today 15:55:30 Rajeev_: Yes. I’d value some feedback in the context of enabling dvr with ha. 15:56:33 carl_baldwin: will take a look with mrsmith as he is focusing on agent side ha dvr work. 15:56:34 I’d also like amuller to take a look but I’ve just heard he’ll be out for a week. 15:57:12 carl_baldwin: fyi I think I've already hit some oddities while using mixins and trying to implement ha and dvr 15:57:51 so the derived classes may help 15:59:38 We’re about out of time. I’ve got to be offline to get in to the office but should be around the rest of the day. 16:00:11 Thanks for all of your work. Let’s communicate often during this refactoring. We’ll get through it and I think we’ll be better off for it. 16:00:22 #endmeeting