15:00:39 <mlavalle> #startmeeting neutron_l3
15:00:39 <openstack> Meeting started Thu Nov 16 15:00:39 2017 UTC and is due to finish in 60 minutes.  The chair is mlavalle. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:40 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:00:43 <openstack> The meeting name has been set to 'neutron_l3'
15:01:17 <davidsha> Hey
15:01:29 <mlavalle> davidsha: hey nice to see you!
15:01:41 * haleyb runs in late
15:01:50 <davidsha> you too, how was Sydney?
15:02:11 <davidsha> hey
15:02:28 <mlavalle> very nice place and it is always nice to see all the people
15:03:27 <mlavalle> mhhh, with the change to standard time in the US it might be too early for Swami
15:03:39 <mlavalle> let's start
15:03:54 <mlavalle> #topic Announcements
15:04:51 <mlavalle> The Queens-2 milestone is fast approaching, December 4 - 8:
15:04:58 <mlavalle> #link https://releases.openstack.org/queens/schedule.html
15:05:13 <mlavalle> before we know, we will be in Rocky!
15:05:27 <davidsha> :O
15:05:52 <mlavalle> Next week is Thanksgiving Day in the US, so I think we will not have this meeting
15:06:17 <mlavalle> unless haleyb insists, since he (((meetings))))
15:06:27 <haleyb> mmm, meetings :)
15:06:36 * haleyb will be skiing
15:06:57 <haleyb> err, meeting on the slopes
15:07:07 <davidsha> Nice, enjoy!
15:07:16 <mlavalle> real mountains somewhere in the Rockies or bunny East Coast mountains?
15:08:08 <haleyb> vermont, edge of Canada, but it opens next week so not expecting much
15:08:27 <mlavalle> #topic Bugs
15:08:44 <mlavalle> haleyb: do you want to start with DVR stuff?
15:09:15 <haleyb> hmm, don't see swami, let me get the query
15:09:33 <haleyb> or you can start so i don't waste time
15:10:10 <mlavalle> during the CI meeting, this one was brought up:
15:10:20 <mlavalle> https://bugs.launchpad.net/neutron/+bug/1717302
15:10:23 <openstack> Launchpad bug 1717302 in neutron "Tempest floatingip scenario tests failing on DVR Multinode setup with HA" [High,Confirmed]
15:10:49 <mlavalle> I think last time we said we had our best people (Swami) working on it
15:11:22 <haleyb> yes, although it's unassigned
15:11:40 <mlavalle> that's right
15:11:50 <haleyb> i'll catch-up with Swami next time i see him online and see if he needs help
15:12:04 <mlavalle> ok, cool
15:12:19 <haleyb> there were two new dvr bugs since the last meeting
15:12:31 <haleyb> https://bugs.launchpad.net/neutron/+bug/1728665
15:12:31 <openstack> Launchpad bug 1728665 in neutron "Removing gateway ip for tenant network (DVR) causes traceback in neutron-openvswitch-agent" [Undecided,New]
15:12:57 <haleyb> i have not confirmed yet, it's on my list to try in devstack
15:13:44 <haleyb> other is https://bugs.launchpad.net/neutron/+bug/1731857
15:13:44 <openstack> Launchpad bug 1731857 in neutron "DVR scenario tests fail in default deployment" [Low,New]
15:14:15 <haleyb> seems to just be a config issue in the tests
15:15:14 <haleyb> that's it for hot dvr bugs
15:15:27 <mlavalle> thanks for the update haleyb
15:16:24 <mlavalle> on non DVR bugs, we have https://bugs.launchpad.net/neutron/+bug/1706750
15:16:24 <openstack> Launchpad bug 1706750 in neutron "FlushError for automatic IPAllocations when subnet create is retried" [High,Confirmed] - Assigned to Ihar Hrachyshka (ihar-hrachyshka)
15:16:44 <mlavalle> we fixed a very similar one a few months ago
15:17:06 <mlavalle> but apparently this one is showing up in Red Hat's CI system
15:17:19 <mlavalle> so Ihar assigned it to himself
15:17:27 <mlavalle> will just keep and eye on it
15:18:10 <mlavalle> Also, during the last CI meeting, the following was mentioned:
15:18:14 <mlavalle> http://logs.openstack.org/51/396351/43/check/legacy-neutron-dsvm-fullstack/6f49eec/logs/dsvm-fullstack-logs/TestLegacyL3Agent.test_north_south_traffic/neutron-dhcp-agent--2017-11-14--10-37-54-629926.txt.gz?level=TRACE#_2017-11-14_10_39_57_401
15:18:35 <mlavalle> I think you took a look at it, haleyb?
15:19:31 <haleyb> mlavalle: oh yeah, it seemed strange since the tap device did not have a MAC address
15:20:17 <haleyb> and it's not simple to bail out of init_l3()
15:20:38 <haleyb> was there a bug filed for it?  i don't remember seeing it
15:20:51 <mlavalle> no, a bug was not filed
15:21:04 <mlavalle> just the stack trace was mentioned
15:22:55 <haleyb> i don't remember what that trace was related to, remember something else causing it?
15:23:48 <mlavalle> it is related to fullstack
15:25:10 <mlavalle> http://logs.openstack.org/51/396351/43/check/legacy-neutron-dsvm-fullstack/6f49eec/logs/dsvm-fullstack-logs/TestLegacyL3Agent.test_north_south_traffic/neutron-dhcp-agent--2017-11-14--10-37-54-629926.txt.gz?level=TRACE#_2017-11-14_10_38_37_370
15:25:41 <haleyb> oh, the one thread locking issue?  i think that fix merged
15:25:59 <davidsha> This week?
15:26:44 * haleyb 's brain is full, do not remember the change id :)
15:26:55 <mlavalle> yeah, here: https://review.openstack.org/#/c/514547/
15:27:34 <davidsha> This log is from 2 days ago though, right?
15:27:54 <mlavalle> it is right
15:28:14 <haleyb> so we'd need to wait for new library version
15:28:36 <davidsha> ack
15:28:55 <haleyb> https://review.openstack.org/#/c/520190/
15:28:58 <mlavalle> is there anything we can do in init_l3() in the meantime, though?
15:29:35 <haleyb> mlavalle: i'll see if i had started a hack somewhere
15:29:47 <mlavalle> ok, cool
15:30:43 <mlavalle> finally, I want to mention https://bugs.launchpad.net/neutron/+bug/1709215
15:30:43 <openstack> Launchpad bug 1709215 in neutron "IPv6 addresses is not correctly configured" [Undecided,In progress] - Assigned to Brian Haley (brian-haley)
15:31:03 <mlavalle> a fix is proposed here: https://review.openstack.org/#/c/491409
15:31:27 <haleyb> +2 :)
15:31:30 <mlavalle> this is good to go, haleyb?
15:31:42 <mlavalle> ok, I'll review it today
15:31:53 <haleyb> mlavalle: yes, i had to tweak it since i couldn't understand the original
15:32:02 <mlavalle> perfect
15:32:17 <mlavalle> those are all the bugs I had today
15:32:30 <mlavalle> any other bugs we should discuss?
15:33:24 <haleyb> not related to L3 :o
15:34:19 <mlavalle> ok, moving on
15:34:41 <mlavalle> #topic QoS for floating ips
15:35:17 <mlavalle> The tc lib patch merged earlier this week
15:35:47 <mlavalle> https://review.openstack.org/#/c/453458/
15:35:54 <mlavalle> Thanks haleyb for the reviews
15:36:01 <haleyb> np
15:36:23 <mlavalle> Over the past two days I have reviewed the server side:
15:36:32 <mlavalle> https://review.openstack.org/#/c/424466/
15:37:08 <mlavalle> In fact, last night I +2ed it pointing a few nits in unit tests
15:37:27 <mlavalle> the author fixed them over night
15:37:34 <mlavalle> so I will take a look again today
15:37:50 <mlavalle> if you have time, haleyb, please take a look
15:38:11 <haleyb> sure, will try to finish this week
15:38:18 <mlavalle> The author mentions that he will submit the agent side soon
15:39:03 <mlavalle> #topic Multiple port bindings
15:39:46 <mlavalle> on this I have spent a good chunck of time (including on the flight back from Sydney) looking at the changes I have to do to the ML2 code
15:40:00 <mlavalle> not very simple but I think I am making progress
15:40:11 <mlavalle> will push code soon for review
15:40:31 <mlavalle> any updates from Sean, davidsha?
15:40:51 <davidsha> None I'm afraid :(
15:41:37 <mlavalle> cool
15:41:49 <mlavalle> #topic Kick ass DVR
15:41:55 <davidsha> :)
15:42:22 <mlavalle> we discussed this in the last drivers meeting
15:42:42 <mlavalle> we concluded that this is an idea that we want to pursue
15:42:56 <mlavalle> haleyb insisted on it
15:43:09 <mlavalle> but we need to spend sometime fixing the DVR jobs
15:43:21 <mlavalle> before we want to merge it
15:43:47 <mlavalle> is this a fair depiction of the situation, haleyb?
15:44:27 <haleyb> mlavalle: yes
15:44:51 <davidsha> thanks to both of you, I do think the decision was fair.
15:45:04 <davidsha> And I understand why it was made.
15:45:32 <mlavalle> davidsha any updates on your side?
15:46:25 <davidsha> I haven't made any progress unfortunetly, I have been focusing on the OpenFlow Manager, I'd neglected it a bit until now.
15:47:11 <davidsha> But I have made time next week for drafting tests for dvr_kickass
15:47:24 <mlavalle> cool
15:47:49 <mlavalle> davidsha, haleyb: indulge me for a bit and let me go crazy...
15:48:05 <davidsha> fire away!
15:48:17 <mlavalle> we have been fighting the DVR failures for good chunck of time now
15:48:30 <mlavalle> two cylces maybe....
15:49:29 <mlavalle> could we use this openflow DVR PoC as a starting point to simplify DVR?
15:49:58 <mlavalle> so we have a better handle of it
15:51:32 <haleyb> mlavalle: perhaps, i guess i haven't looked at it that way
15:51:45 <davidsha> Are most of the problems in the gate being cause by weirdness with network namespaces? Or how do you mean?
15:52:19 <mlavalle> not necessarilly that the issues are namespaces related
15:53:02 <mlavalle> but we can use this PoC to try to look at DVR in a different way
15:53:16 <mlavalle> in case we continue struggling with DVR failures
15:53:25 <mlavalle> I am just putting two things together
15:54:24 <mlavalle> remember, I asked for your indulgence
15:55:24 <mlavalle> if we continue struggling, maybe is time to look at a serious simplification effort and in the way use openflow to implement routers
15:55:51 <davidsha> Sure, the only issue at the moment is IPv6 doesn't work yet.
15:56:14 <mlavalle> yeah, there is stuff still to be done
15:56:17 <haleyb> yes, and other projects (like OVN) are using openflow to implement routers
15:57:01 <davidsha> OpenFlow can't modify Neighbour discovery packets enough to fake a router interface for IPv6, ODL and OVN both use controllers for that.
15:57:05 <haleyb> anything to reduce the number of moving parts would be good (namespaces, etc)
15:57:18 <mlavalle> ok, just wanted to plant the seed, let's leave it at that today
15:57:50 <mlavalle> haleyb: and yes, openflow would be just an element of a broadar simplification
15:58:01 <mlavalle> let's move on
15:58:07 <mlavalle> #Open Topics
15:58:13 <davidsha> Ok.
15:58:19 <mlavalle> #topic Open Agenda
15:58:31 <mlavalle> anything else to share with the team
15:58:32 <mlavalle> ?
15:59:15 <haleyb> not from me
15:59:26 <davidsha> I think we may be out of time.
15:59:36 <mlavalle> ok team, thanks for your attendance, see you in two weeks !
15:59:42 <davidsha> Thanks!
15:59:44 <mlavalle> #endmeeting