15:03:31 #startmeeting neutron_l3 15:03:32 Meeting started Thu Jun 1 15:03:31 2017 UTC and is due to finish in 60 minutes. The chair is mlavalle. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:03:33 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:03:35 The meeting name has been set to 'neutron_l3' 15:03:47 #chair haleyb Swami_ 15:03:48 Current chairs: Swami_ haleyb mlavalle 15:03:49 hi back in the right channel 15:04:00 Sorry for the confusion 15:04:13 #Announcements 15:04:26 #topic Announcements 15:05:01 We almost reached Pike-2. It is next week 15:05:10 #link https://releases.openstack.org/pike/schedule.html 15:05:14 mlavalle: too close 15:05:56 any other announcements from the team? 15:06:22 mlavalle: I don't have any thing to announce 15:06:59 ok, moving on 15:07:08 #topic Bugs 15:07:26 Swami_: please fire away with DVR related bugs 15:08:06 mlavalle: thanks 15:08:19 mlavalle: haleyb: I found that the dvr is broken in upstream. 15:08:45 mlavalle: haleyb: was there any new devstack changes that occurred in the pike1 15:09:11 Swami_: i don't know. what about the neutron change you noticed? 15:09:16 I did see that a patch that was merged from kevin broke the dvr. The snat namespace are not getting created. 15:09:50 #link https://review.openstack.org/#/c/457592/ 15:10:00 After this patch, the snat namespaces are not getting created. 15:10:11 The snat ports are not getting bound to the actual host. 15:10:48 Why aren't all the DVR jobs broken then? 15:11:05 haleyb: mlavalle: that's my confusion right now. 15:11:51 haleyb: The agent side test cases that are specific to snat namespaces may pass. But I still wonder why the tempest tests are not showing any errors. 15:12:11 * haleyb is loading the grafana page now 15:12:15 haleyb: that is reason I was checking with you, if there was any devstack change beyond this patch that makes it work in upstream. 15:13:02 haleyb: mlavalle: I suspect that this might be one of the reason for the address scopes match to log the errors while trying to process the routers. 15:13:03 I am not aware of relevant devstack changes 15:13:49 Swami_: is there an error in the server or l3 log? 15:13:52 mlavalle: Yes it is easy to reproduce, if you can go to a patch before this posted patch you can see both the snat and fip namespace and after this patch, I don't see the snat namespace. 15:14:08 haleyb: I did see one rpc error in the neutron-server. 15:14:32 haleyb: I don't see any error on the agent side, since the port is not bound. 15:15:03 so it just never gets bound, hmm 15:15:04 haleyb: I wanted to confirm first and then I will file a bug. I also posted a message to kevin last night. 15:15:10 haleyb: yes 15:15:37 I did see the rpc handler is trying to bound the port to a given host, but the port update fails. 15:16:32 haleyb: mlavalle: I will try to see if I can collect more information on this issue today, having said that there is no configuration changes in the devstack. 15:16:33 even if we missed to update one of the calls, it is still used as a mixin, so it should work, right? 15:17:11 mlavalle: We need to identify, why the port is not getting bound to the snat host? 15:17:35 mlavalle: I am not clear on why this mixin change would cause this issue. 15:17:50 yeah, seems pretty innocent 15:17:53 Swami_: i would open a bug with the port update failure, i am creating a new stack here 15:18:09 haleyb: ok, thanks 15:18:17 * haleyb still dealing with broken laptop, recreating everything sucks 15:18:27 it does :_) 15:18:32 haleyb: Meanwhile I will also check for the path on why it is failing. 15:19:43 haleyb: mlavalle: So I am suspecting the patch that was reverted for error logs, might be an after effect of this issue. #link https://review.openstack.org/#/c/457592/ 15:20:26 ack 15:20:29 Swami_: for the address scopes? 15:20:29 haleyb: mlavalle: so the plan is we fix this snat issue first and then I can drill down to see if there is any check required in the address scopes patch to prevent those logs, if there is no existing route. 15:20:41 haleyb: yes 15:22:14 These are the only hot issues I had with DVR. There are no new bugs added to the list this week. Other than the one that we would be filing today to address the snat issue with the mixin change. 15:22:47 mlavalle: back to you 15:23:03 Swami_: Thanks 15:23:40 This one https://bugs.launchpad.net/neutron/+bug/1684682 15:23:41 Launchpad bug 1684682 in neutron "DHCP namespace doesn't have IPv6 default route" [High,Fix released] - Assigned to Brian Haley (brian-haley) 15:23:46 merged recently 15:24:05 so I will remove it from the etherpad 15:24:19 the bug itself shows fix released, so we are good 15:24:29 yup, i proposed backports yesterday 15:24:55 Thanks haleyb 15:25:16 Next one is https://bugs.launchpad.net/neutron/+bug/1655567 15:25:17 Launchpad bug 1655567 in neutron "DHCPv6 failures with "address already allocated in subnet" error" [High,In progress] - Assigned to Miguel Lavalle (minsel) 15:25:53 For this one I confirmed with more recent hits the pattern I explained last week 15:26:46 subnet creation and dhcp port creation / update interfere each other when the subnet is dhcpv6-stateless 15:27:09 Last night proposed a fix: https://review.openstack.org/#/c/469669 15:27:34 I forgot to add matching () 15:27:41 so it failed a lot of tests 15:28:13 I updated earlier today. As soon as it passes Jenkins I will remove WIP and it is good for reviews 15:28:30 thanks mlavalle 15:29:24 haleyb, Swami_: please note that I left a comment / question in the code 15:29:45 I would like some input on hat 15:29:49 that^^^ 15:29:52 mlavalle: ok 15:30:03 ok 15:30:29 Next one is https://bugs.launchpad.net/neutron/+bug/1683227 15:30:31 Launchpad bug 1683227 in neutron "test_connection_from_same_address_scope failed with: Cannot find device "qg-74372988-7f"" [High,Confirmed] - Assigned to Miguel Lavalle (minsel) 15:30:52 I just started working on this one right before this meeting 15:31:29 as usual, struggling a little bit coming up with the right query for Kibana 15:31:39 I'll keep the team posted 15:32:42 Finally, this one https://bugs.launchpad.net/neutron/+bug/1694764 15:32:43 Launchpad bug 1694764 in neutron "test_metadata_proxy_respawned failed to spawn metadata proxy" [High,Confirmed] 15:32:47 mlavalle: I hope this is for a non-dvr 15:33:15 LOL 15:33:36 mlavalle: :) 15:33:40 I just came across this one before the meeting 15:34:30 that is newton as well 15:34:34 yes 15:35:32 any other comments or bugs from the team? 15:35:59 not from me 15:35:59 mlavalle: nothing from me. 15:36:02 Hey, is there an open discussion section for the meeting? 15:36:13 davidsha: yes there is 15:36:24 hang on just a few more minutes 15:36:24 mlavalle: I'll wait then, thanks! 15:36:45 #topic DNS 15:38:00 I recently pushed the last patchset of the series that adds dns_domain attribute and related functionality to ports 15:38:11 #link https://review.openstack.org/#/c/468697/ 15:38:23 it is passing all the existing unit tests 15:38:33 so I didn't break anything 15:38:51 and I am working through the unit tests for the new functionality 15:38:59 should be done early next week 15:39:25 any questions? 15:39:49 mlavalle: no 15:39:51 mlavalle: so will dnsmasq send back this domain? i haven't looked closely yet 15:39:52 ok 15:40:18 haleyb: no, this dns_domain is only for external dns service 15:40:38 or in case the user want a dns_domain attribute assigned to its port 15:40:58 ah, so someone can host different domains with a single project 15:41:09 correct 15:41:33 (light bulb ON) 15:41:39 LOL 15:42:08 #topic Open Discussion 15:42:18 davidsha: the floor is yours 15:42:43 mlavalle: Thanks, I've been working on a PoC for dvr in user space 15:43:38 davidsha: is this to get rid of the namespaces. 15:43:50 It uses OpenFlow rules to make the switch act like a router, at the moment it supports east west traffic and I'm almost done with north south. 15:44:01 Swami_: To provide an alternative yes. 15:44:48 davidsha: are you aware of https://wiki.openstack.org/wiki/Dragonflow 15:45:39 ? 15:46:05 mlavalle: I was, do you think this would be a better topic for DragonFlow? 15:46:30 davidsha: not necessarily. I just want you to have visibility of it 15:46:40 please carry on 15:46:55 mlavalle: ack, I'll take a closer look :) 15:47:39 I'm hoping to wrap up north south soon, would you all be interested in taking a look? 15:47:51 davidsha: I would 15:47:58 me too 15:48:23 and thanks for the effort and for sharing :-) 15:49:03 No problem! I'll try to have it up asap! 15:49:58 davidsha: you don't have to look at dragonflow. But if you find it interesting, you might want to watch https://www.youtube.com/watch?v=2zrMZFnBQuw 15:50:52 mlavalle: thanks, I'll check it out! 15:51:21 anythong else? 15:51:44 davidsha: will take a look at it. 15:51:45 anything^^^^ 15:52:04 mlavalle: haleyb: next week I may be on vacation. 15:52:21 Swami_: enjoy! are you going somewhere? 15:52:27 mlavalle: haleyb: moving my home to a new home. 15:52:44 Swami_: good luck! 15:52:46 bigger, nicer, shinner? 15:53:03 mlavalle: bigger closer to high school for my kid 15:53:20 that makes sense. Good luck, moves are not fun 15:53:39 you realize how much stuff you have accumulated over the years 15:53:45 a lot of it useless 15:53:55 mlavalle: yes it is. 15:54:08 mlavalle: 11 years store - too much to throw out. 15:54:37 thanks for joining the meeting guys 15:54:44 Thanks! 15:54:45 enjoy the rest of the week 15:54:54 #endmeeting