14:02:33 #startmeeting neutron_l3 14:02:34 Meeting started Wed Aug 26 14:02:33 2020 UTC and is due to finish in 60 minutes. The chair is liuyulong. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:02:35 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:02:37 The meeting name has been set to 'neutron_l3' 14:02:44 Sorry, a bit late 14:03:12 liuyulong: i have many conflicts this morning, but will try and respond if you use my nick 14:03:16 hi 14:03:31 Sure 14:03:33 Hi 14:03:37 #topic Announcements 14:03:41 * slaweq has another meeting in 30 minutes so will be just lurking here 14:03:45 #link http://eavesdrop.openstack.org/meetings/networking/2020/networking.2020-08-25-14.01.log.html#l-10 14:04:09 This should be a short meeting, so dont worry 14:04:17 This is the team annouuncements. 14:04:46 No other announcements from me, so let's go through the bugs. 14:04:53 #topic Bugs 14:05:22 #link http://lists.openstack.org/pipermail/openstack-discuss/2020-August/016559.html 14:05:29 #link http://lists.openstack.org/pipermail/openstack-discuss/2020-August/016761.html 14:05:34 hi (sorry for being late) 14:05:45 These are the last 2 weeks bug deputy report list. 14:05:48 Hi ralonsoh 14:06:00 #link https://bugs.launchpad.net/neutron/+bug/1891673 14:06:01 Launchpad bug 1891673 in neutron "qrouter ns ip rules not deleted when fip removed from vm" [High,In progress] - Assigned to Edward Hope-Morley (hopem) 14:06:48 Bence had confirmed the bug. 14:07:07 IMO, this is related to the mixed deployment between compute node dvr and dvr_snat. 14:07:37 That is saying the user is running VM in a dvr_snat agent mode node. 14:07:52 And then try to bind/unbind the floating IP from the instance. 14:08:05 #link https://review.opendev.org/#/c/746336/ 14:08:15 This is the fix of the bug. 14:08:56 * haleyb will review again later 14:10:17 But according to the fix, it is a bug of the cache of route rule priority. 14:10:53 Yes, let's try/review that fix. 14:11:04 Next 14:11:06 #link https://bugs.launchpad.net/neutron/+bug/1891360 14:11:07 Launchpad bug 1891360 in neutron "Floating IP agent gateway IP addresses not released when deleting dead DVR L3 agents" [Wishlist,Confirmed] 14:11:26 Yes, neutron should do this cleanup work. 14:11:46 agree 14:11:49 A potential way is add some port_delete notification recivier which check the port device_owner. 14:12:14 Sorry, agent delete notification 14:13:00 Then dvr DB plugin then can do the related port removal. 14:14:33 Beyond of this, maybe delete the dvr_agent should take care more cautiously. 14:15:07 For now, neutron seems to allow delete the agent without any warning or check. 14:15:59 If a dvr L3 agent has running router with floating IPs, delete the related FIP agent gateway port may have a potential mis-op risk 14:16:57 right, should make sure things get to the dvr_snat node in that case 14:19:13 So, a standard operation document can be supplied to user when they want to offline a L3 (dvr) agent. 14:19:30 as a first step work for this bug 14:21:26 OK, next one 14:21:33 #link https://bugs.launchpad.net/neutron/+bug/1891448 14:21:34 Launchpad bug 1891448 in neutron "L3 agent mode transition between dvr and dvr_no_external" [Wishlist,New] 14:22:59 I submitted this since we meet such request locally. 14:24:56 A way in my mind is to add some notifications like: dvr (change to no_external) -> server -> dvr_snat 14:25:15 Or, dvr_no_external (change to dvr) -> server -> dvr_snat 14:25:48 Everytime the agent mode is changed, the L3 agent will re-process the routers. 14:26:50 During this re-processing procedure, check the floating IP host attributes. 14:26:57 and i can imagine there are some edge cases in transitions 14:30:10 haleyb, cool, good to know, please add them to the LP. We can discuss in there later. 14:30:34 Last open one: 14:30:35 #link https://bugs.launchpad.net/neutron/+bug/1892405 14:30:36 Launchpad bug 1892405 in neutron "Removing router interface causes router to stop routing between all" [High,New] 14:31:40 I'm confused about "VMs can't reach the HA port (IP)". 14:32:10 Is that saying the VM is trying to ping the HA port in the snat-namespace? 14:32:52 The HA port is just used for VRRP of the router HA mechanism. 14:32:58 i wonder if this is related to the HA to centralized issue slawek is fixing? not a lot of info there, like a backtrace :) 14:33:03 It is not visible to the end user. 14:34:53 maybe is talking about the GW port, I don't know 14:35:05 let's ask for more info in the LP bug 14:35:59 yes, slaweq has marked it to High 14:36:44 OK, last 3 fixed bugs. 14:36:45 #link https://bugs.launchpad.net/neutron/+bug/1892489 14:36:47 Launchpad bug 1892489 in neutron "[Prefix delegation] When subnet with PD enabled is added to the router, L3 agent fails on waiting for LLAs to be available" [High,Fix released] - Assigned to Slawek Kaplonski (slaweq) 14:36:49 #link https://bugs.launchpad.net/neutron/+bug/1892364 14:36:50 Launchpad bug 1892364 in neutron "L3 agent prefix delegation - adding new subnet to the router fails" [Medium,Fix released] - Assigned to Brian Haley (brian-haley) 14:36:54 #link https://bugs.launchpad.net/neutron/+bug/1892362 14:36:55 Launchpad bug 1892362 in neutron "Restarting L3 agent when PD is used fails due to IPAddressAlreadyExists error" [Medium,Fix released] - Assigned to Slawek Kaplonski (slaweq) 14:37:02 All thanks to slaweq, great work! 14:37:07 ++ 14:37:23 thx, all were small issues but in overall caused impossible to run dibbler-client 14:38:17 Wait, one more bug 14:38:19 #link https://bugs.launchpad.net/neutron/+bug/1892200 14:38:20 Launchpad bug 1892200 in neutron "Make keepalived healthcheck more configurable" [Wishlist,New] 14:38:26 It is approved. 14:39:16 So, cool, it is configurable, not fixed. So I'm fine with it. : ) 14:40:03 OK, no more bugs from me now. 14:41:29 OK, let's move on. 14:41:34 #topic On demand agenda 14:42:23 I noticed that Flavio is trying to backport the OVN port forwarding to stable/ussuri. 14:42:46 Any thougths? 14:43:04 IMO is possible 14:44:26 #link https://review.opendev.org/#/c/747299/ 14:44:41 We have some discussion in this link. 14:45:00 #link https://docs.openstack.org/project-team-guide/stable-branches.html#active-maintenance 14:45:13 There are some rules that prevent from doing such backport. 14:45:49 well, you are right on this 14:45:51 ralonsoh, IMO, every commit is possible to backport. The main concern is should/should not. : ) 14:46:20 let's raise the discussion in gerrit 14:46:29 but as you said, this is a new feature 14:46:49 Yes, it is new to OVN backend. 14:47:39 From my experiences, the L3 native implementation of port forwarding takes at least 2 or 3 dev cycles to make it robust. 14:53:58 OK, let's end here. 14:54:01 Thank you guys. 14:54:08 thanks! 14:54:09 bye 14:54:14 #endmeeting