15:00:16 #startmeeting neutron_l3 15:00:22 Meeting started Thu Mar 8 15:00:16 2018 UTC and is due to finish in 60 minutes. The chair is mlavalle. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:23 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:26 The meeting name has been set to 'neutron_l3' 15:00:52 hi there! 15:00:56 hi 15:01:28 hey Swami 15:01:33 Nice to see you! 15:02:06 hi 15:02:47 hi david_chou 15:02:57 haleyb_ you around? 15:03:02 mlavalle: How did the midcycle go. 15:03:19 The meetings went well 15:03:26 it was productive 15:03:33 lots of plans as usual 15:03:44 although the weather didn't cooperate 15:04:05 It snowed so badly that the event was shutdown on Thursday after lunch 15:04:25 we had to improvise meeting rooms at the hotel 15:04:26 mlavalle: that is odd 15:04:48 the snow storm shutdown the entire UK and Ireland 15:04:59 it's the worst they've had in 30 years 15:05:16 mlavalle: Oh 15:05:48 in the UK, the army was rescuing people stranbded in the roads 15:06:14 ok, let's get going 15:06:19 mlavalle: no one can stop the wether. 15:06:28 s/wether/weather 15:06:33 #topic Announcements 15:07:14 Next milestone is Rocky-1, April 16 - 20 15:07:24 so welcome to Rocky! 15:07:57 The entire Rocky delivery schedule is here: 15:08:04 #link https://releases.openstack.org/rocky/schedule.html 15:09:02 Tomorrow or Monday I am going to send a mesaage to the ML with a summary of the PTG 15:09:28 In the meantime you can see the notes in the etherpad: 15:09:33 mlavalle: nice 15:09:40 #link https://etherpad.openstack.org/p/neutron-ptg-rocky 15:10:05 any other announcements? 15:10:16 Execuse me, what is ML? 15:10:34 david_chou: ML - Mailing list 15:10:48 Thanks! 15:10:56 david_chou: sorry 15:11:16 #topic Bugs 15:11:33 Swami: do you want to lead with DVR bugs? 15:12:03 mlavalle: sure I have a couple that I wanted to discuss 15:12:40 #link https://bugs.launchpad.net/neutron/+bug/1751396 15:12:41 Launchpad bug 1751396 in neutron "DVR: Inter Tenant Traffic between two networks and connected through a shared network not reachable with DVR routers" [Undecided,In progress] - Assigned to Swaminathan Vasudevan (swaminathan-vasudevan) 15:12:54 This is a new bug that was filed by me last week. 15:13:03 I have a patch up for review. 15:13:19 #link https://review.openstack.org/#/c/547696/ 15:13:30 Needs review on this patch. 15:13:54 This bug is seen when we have shared network on the internal private network without a gateway. 15:14:14 The next bug in my list is 15:14:35 #link https://bugs.launchpad.net/neutron/+bug/1608400 15:14:36 Launchpad bug 1608400 in neutron "Neutron should not add ARP entry for allowed-address-pair-fixed-ip in DVR router" [High,In progress] - Assigned to Swaminathan Vasudevan (swaminathan-vasudevan) 15:14:58 This is one other issue that we have seen issue internally and was also filed upstream. 15:15:35 The Allowed_address_pair when used with Keepalived even without a FloatingIP is not working with DVR. 15:15:58 The issue was with the router-namespace not able to capture the GARP sent out from the Keepalived. 15:16:37 The reason is we are adding the ARP entry in the router-namespace as a PERMANENT entry. When ARP entry is added as PERMANENT entry it is not updating the entry when the GARP comes in. 15:17:03 The fix for this patch is allow to override the NUD state for the ARP to set the state as 'reachable' 15:17:20 #link https://review.openstack.org/#/c/550676/ 15:17:26 This patch is under review. 15:17:44 I have to add a couple of tests this is still under WIP. 15:17:52 Any early feedback or review would be nice. 15:18:07 it fails unit tests 15:18:29 mlavalle: yes I will fix it and add more tests to it. 15:18:45 mlavalle: Those are the only two new issues that I wanted to bring up. 15:19:02 mlavalle: back to you. 15:19:27 Swami: have you had a chance to look lately at the DVR scenario test that has been failing for some time now? 15:19:48 or is that something that haleyb_ is looking at now? 15:19:56 mlavalle: No I did not notice. I will take a look at it. 15:20:21 Swami: do you remember whhich one I am talking about? 15:20:35 mlavalle: Are you talking about the tempest failure. 15:20:40 yes 15:21:26 mlavalle: In the last meeting haleyb mentioned that he was setting up his environment to test it. 15:21:34 ok 15:21:34 mlavalle: But I have not heard back from him 15:21:43 that's good 15:21:45 mlavalle: definitely I will take a look at it. 15:21:56 Swami: thanks Swami ! 15:21:56 mlavalle: I am pretty much swamped downstream at this point. 15:22:03 I know 15:22:19 if you don't have time to look at it, that's fine 15:22:34 I just don't want that bug to fall through the cracks 15:22:53 mlavalle: no problem, I will check back with haleyb on this 15:23:18 ok 15:24:11 mlavalle: that's all I have for today. 15:24:19 Thank you! 15:24:34 #topic Openflow DVR 15:25:19 Give me one moment. 15:25:27 david_chou: sure 15:25:47 Already submitted the fix for ovs-dpdk with dvr_openflow into the patch https://review.openstack.org/#/c/472289/ 15:26:41 Thomas Morin has a comment on https://review.openstack.org/#/c/472289/ 15:28:32 david_chou: we actually discussed the patch in DUblin 15:29:08 so I am aware of the current status 15:29:35 David Shaugnesey alo responded to Thomas. 15:29:48 What is your idea? 15:29:50 I intend to review it over the next few days and I will leave my comments there 15:30:03 Thanks. 15:30:28 is there a way you could start drafting the devref document? 15:30:39 it would help with the reviews 15:31:26 My peer (John Hinman) is in this meeting too. He will take over my work, because I will retire in the end of this April. 15:31:52 david_chou: retire as in leaving Intel? 15:32:09 Retire. 15:32:26 all right, good luck! 15:33:02 is John Hinman == john-h? 15:33:04 I will work with John to the end of this April. 15:33:46 Yes. 15:34:15 john-h: welcome to the team! 15:35:03 thanks. probably would have been better to have chosen jhinman as nick 15:35:29 john-h: welcome 15:35:35 It will be highly appreciated if john-h and / or david_chou can put together an initial devref document 15:35:57 just add it to the patch 15:36:04 it will help in the review 15:36:11 I searched with "OpenStack devref", it seems not in current master branch for those used to have devref? Can you send same sample or Doc to us. 15:36:44 mlavalle: I need to drop off. Will catch you later. 15:36:58 Swami: go ahead, thanks for the update 15:37:21 david_chou, john-h: here's the entire Neutron devref: https://docs.openstack.org/neutron/latest/contributor/index.html#developer-guide 15:38:36 to add a Openflow DVR doc, you just add another doc here: https://github.com/openstack/neutron/tree/master/doc/source/contributor/internals 15:38:49 it is in the Neutron repo 15:39:09 does it make sense? 15:39:17 Ok, will do. 15:39:37 That is why I say just add it to the patch we already have 15:40:34 is that all for today's update? 15:40:42 John will start to work on unit test, functional test, and tempest for dvr_openflow. 15:41:35 john-h: please keep in mind Tempest tests go to a different repo 15:42:02 #link https://github.com/openstack/neutron-tempest-plugin 15:42:22 ok 15:42:52 I don't think we need tempest API tests 15:43:09 but we will definetely need a scenario test 15:43:12 That is all I have. John, do you have other stuff to discuss. 15:43:25 no 15:43:26 a multinode one 15:43:46 because DVR testing makes sense in multinode environment 15:44:30 I think David has been working on multinode test, right David? 15:45:54 No, I verified the patch in single node as David Shaunessey asid that it is O.K. to verify the patch in single node. 15:46:28 When you said multinode, do you mean hascenario? 15:46:49 john-h, david_chou: here's the section from devref on scenario testing: 15:46:59 #link https://docs.openstack.org/neutron/latest/contributor/testing/testing.html#scenario-tests 15:48:17 Thanks, we will work on it. 15:49:04 david_chou, john-h: if you look at the Zuul section of the patch, you will notice that a "neutron-tempest-plugin-dvr-multinode-scenario" job was executed to test it 15:49:32 it is a job where an environment with several nodes is created and the tests run on that environment 15:49:44 I think it makes sense to have something similar 15:50:33 does it make sense? 15:51:20 We will look it, and ask you question through email or in next meeting. 15:51:40 of course 15:51:49 I expect you to have questions 15:52:36 ok, thanks for the update 15:52:45 #topic Open Agenda 15:52:55 any other topics we should discuss today? 15:54:15 ok, thanks for attending today 15:54:24 Have a nice rest of your week! 15:54:26 Thanks, Bye. 15:54:30 #endmeeting