14:00:35 <liuyulong> #startmeeting neutron_l3
14:00:36 <openstack> Meeting started Wed Aug 28 14:00:35 2019 UTC and is due to finish in 60 minutes.  The chair is liuyulong. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:00:37 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:00:40 <openstack> The meeting name has been set to 'neutron_l3'
14:01:19 <haleyb> hi
14:01:36 <openstack> liuyulong_: Error: Can't start another meeting, one is in progress.  Use #endmeeting first.
14:01:49 <haleyb> nested meeting? :)
14:01:55 <slaweq> o/
14:01:59 <liuyulong_> bad connection....
14:02:10 <liuyulong> #chair haleyb
14:02:11 <openstack> Current chairs: haleyb liuyulong
14:02:15 <ralonsoh> hi
14:02:15 <liuyulong> #chair liuyulong_
14:02:15 <openstack> Current chairs: haleyb liuyulong liuyulong_
14:02:28 <liuyulong> OK, let's start
14:02:30 <liuyulong> #topic Announcements
14:03:07 <liuyulong> #link https://etherpad.openstack.org/p/Shanghai-Neutron-Planning
14:03:36 <liuyulong> I added some topics last night, hope you guys are interested in that.
14:06:52 <haleyb> liuyulong: instead of centralized DNAT scale-out, why not DVR?  but maybe that's a discussion for the ptg
14:06:59 <liuyulong> The number of PTG participants seems to be very small.
14:08:31 <liuyulong> haleyb, for some deployment, they just do not want all nodes have the public network NICs.
14:09:21 <liuyulong> haleyb, a centralized NAT cluster looks easy to manage and maintain.
14:10:39 <liuyulong> So please add your name and topic there.
14:10:44 <liuyulong> Let's move on.
14:10:54 <liuyulong> #topic Bugs
14:11:47 <liuyulong> We have no bug deputy email this week.
14:12:07 <haleyb> there was a google doc i believe
14:12:26 <liuyulong> #link https://wiki.openstack.org/wiki/Network/Meetings#Bug_deputy
14:12:38 <liuyulong> August-19-2019	Swaminathan Vasudevan (Swami)
14:12:42 <liuyulong> August-26-2019	Akihiro Motoki (amotoki)	replaced by lajoskatona for Aug 26
14:13:22 <liuyulong> So last week should be Swaminathan, and this week lajoskatona
14:14:02 <liuyulong> #link https://bugs.launchpad.net/neutron/+bugs?field.searchtext=&orderby=-datecreated&search=Search&field.status%3Alist=NEW&field.status%3Alist=OPINION&field.status%3Alist=INVALID&field.status%3Alist=WONTFIX&field.status%3Alist=EXPIRED&field.status%3Alist=CONFIRMED&field.status%3Alist=TRIAGED&field.status%3Alist=INPROGRESS&field.status%3Alist=FIXCOMMITTED&field.status%3Alist=FIXRELEASED&field.status%3Alist=INCOMPLETE_WITH_RESPONSE&field.sta
14:14:02 <liuyulong> tus%3Alist=INCOMPLETE_WITHOUT_RESPONSE&assignee_option=any&field.assignee=&field.bug_reporter=&field.bug_commenter=&field.subscriber=&field.structural_subscriber=&field.tag=&field.tags_combinator=ANY&field.has_cve.used=&field.omit_dupes.used=&field.omit_dupes=on&field.affects_me.used=&field.has_patch.used=&field.has_branches.used=&field.has_branches=on&field.has_no_branches.used=&field.has_no_branches=on&field.has_blueprints.used=&field.has
14:14:02 <liuyulong> _blueprints=on&field.has_no_blueprints.used=&field.has_no_blueprints=on
14:14:38 <slaweq> liuyulong: can You maybe use some url shortener and give short url here? :)
14:15:11 <haleyb> https://docs.google.com/spreadsheets/d/1JjcSHL6hFYsAvlrcgK7_Eje1S4vvJdvm363_Mmy557I/edit?usp=sharing
14:15:17 <liuyulong> Let's scan the list by "newest first".
14:15:18 <haleyb> that is the bug report for last week
14:15:56 <liuyulong> I can not open that link right now...
14:16:08 <liuyulong> #link https://bugs.launchpad.net/neutron/+bug/1841753
14:16:09 <openstack> Launchpad bug 1841753 in neutron "Pyroute2 netns.ns_pids() will fail if during the function loop, one namespace is deleted" [Undecided,New] - Assigned to Rodolfo Alonso (rodolfo-alonso-hernandez)
14:16:36 <liuyulong> ralonsoh is working on this
14:16:46 <ralonsoh> I have a patch uploaded
14:16:59 <ralonsoh> https://review.opendev.org/#/c/679026/
14:17:08 <ralonsoh> there is an issue in pyroute2
14:17:17 <ralonsoh> so this is a workaround
14:17:39 <liuyulong> So the fix is a workaround.
14:17:51 <ralonsoh> that's why I set related-bug in the comment
14:18:00 <ralonsoh> this patch is not a fix for neutron
14:18:08 <ralonsoh> the fix should be in pyroute2
14:18:17 <ralonsoh> (I'll propose a patch this week)
14:19:07 <liuyulong> Make sense, current approach can help solve the problem, we can merge it soon if it break CI too much.
14:19:39 <liuyulong> Next
14:19:43 <liuyulong> #link https://bugs.launchpad.net/neutron/+bug/1841682
14:19:44 <openstack> Launchpad bug 1841682 in neutron "rootwrap sudo process goes into defunct state" [Undecided,New]
14:19:58 <ralonsoh> that was filled yesterday
14:20:13 <liuyulong> zombie process right?
14:20:31 <ralonsoh> this should not be related to the last three patches submitted to neutron-keepalived
14:20:38 <ralonsoh> yes zoombie processes
14:20:58 <ralonsoh> this bug is reported for an environment using rocky
14:21:28 <ralonsoh> so I would like to know if this is happening now (I'll check the logs in the functional and fullstack tests)
14:21:32 <liuyulong> So I wonder what's the real command after that "[sudo]" ?
14:21:43 <ralonsoh> that's the point: we can't know
14:22:07 <ralonsoh> this happens if you spawn a process using "Ip netns"
14:22:15 <ralonsoh> and you kill the process inside the namespace
14:22:25 <ralonsoh> the kernel process will remain in this state
14:23:00 <ralonsoh> I'll take a look at the current logs in the CI
14:23:41 <liuyulong> Great, thank you, may I assign this bug to you?
14:23:45 <ralonsoh> sure
14:24:19 <liuyulong> Rodolfo Alonso (rodolfo-alonso-hernandez) right?
14:24:28 <ralonsoh> yes
14:24:52 <liuyulong> Done
14:24:57 <liuyulong> OK, let's move on
14:25:02 <liuyulong> Next...
14:26:03 <liuyulong> #link https://bugs.launchpad.net/neutron/+bug/1839004
14:26:04 <openstack> Launchpad bug 1839004 in neutron "Rocky DVR-SNAT seems missing entries for conntrack marking" [Undecided,Incomplete]
14:27:06 <liuyulong> Something related to BGP
14:28:20 <haleyb> right, looked like subnet pool issue at first
14:28:21 <liuyulong> Ryan said it is not reproducable.
14:31:12 <liuyulong> Bad connection...
14:31:27 <haleyb> can probably close then
14:31:56 <liuyulong> Let's keep an eye on that bug.
14:32:39 <liuyulong> Most of other bugs related to L3 are marked as fixed or invalid.
14:33:27 <liuyulong> So if you guys have others, go ahead~
14:34:04 <haleyb> https://bugs.launchpad.net/neutron/+bug/1812118
14:34:05 <openstack> Launchpad bug 1812118 in neutron "Neutron doesn't allow to update router external subnets" [Medium,In progress] - Assigned to Brian Haley (brian-haley)
14:34:13 <haleyb> https://review.opendev.org/#/c/670177/
14:34:40 <haleyb> while i can't seem to get the test to pass, let's discuss the one-line change
14:35:54 <haleyb> basically the bug is you can't change the external subnet/IP of a router, you have to remove then re-add
14:36:14 <haleyb> but i think it should be allowed if the network is not changing
14:38:54 <liuyulong> Change external subnet may have some issues of the public network connection. Any why users need to change the external subnet?
14:40:31 <haleyb> i didn't see any issues in testing, and it's possible to do, just have to be admin
14:40:50 <haleyb> if i can delete and add to do the same thing, why not update?
14:42:03 <haleyb> it doesn't seem much different than having multiple external subnets, and your router is on one, but floating IPs are on another
14:42:15 <haleyb> it should still work
14:42:31 <liuyulong> So the original request is whay this command does not work for normal user? "openstack router set --external-gateway 30e25ece-439b-4d9f-a3f7-816d0167d2cd --fixed-ip subnet=72808793-445e-4fe5-b653-097d720304e8 test"
14:43:48 <haleyb> right, because of the missing admin context
14:45:07 <liuyulong> OK, looks like I was overthinking of this.
14:46:42 <haleyb> it's the darn test that is killing me, if anyone has a thought on that i'd appreciate it, seems simple from a high level
14:50:15 <liuyulong> Should call update_router for the test?
14:50:47 <liuyulong> https://review.opendev.org/#/c/670177/7/neutron/tests/fullstack/test_l3_agent.py@139, you call the "update_port" now.
14:52:00 <haleyb> hmm, maybe that's it, i was thinking that test was very similar, i'll take a look, thanks liuyulong
14:53:11 <haleyb> or i just remove the test :)
14:53:47 <liuyulong> Test cases are always needed. : )
14:54:05 <slaweq> liuyulong: nah, tests are only slowing us down :P
14:54:34 <haleyb> it seemed like fullstack was the right way though
14:55:55 <liuyulong> OK, let's stop here today.
14:56:12 <liuyulong> Thank you guys for attending.
14:56:20 <ralonsoh> bye
14:56:26 <slaweq> bye
14:56:30 <liuyulong> #endmeeting