15:02:49 <Swami> #startmeeting distributed_virtual_router
15:02:49 <openstack> Meeting started Wed Sep 24 15:02:49 2014 UTC and is due to finish in 60 minutes.  The chair is Swami. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:02:50 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:02:52 <openstack> The meeting name has been set to 'distributed_virtual_router'
15:03:30 <Swami> carl_baldwin: you have asked about the critical bugs for DVR, do you have any questions on it
15:03:59 <carl_baldwin> I think all of my questions were answered in the thread.
15:04:09 <Swami> carl_baldwin: thanks
15:04:15 <carl_baldwin> yw
15:04:19 <Swami> just wanted to make sure that we are in the same page
15:04:40 <Swami> #info Juno RC-1 is going to be cut tomorrow
15:05:17 <Swami> Hope all our critical ones should go in by tomorrow.
15:05:54 <Swami> #topic Bugs
15:06:06 <Swami> #link https://bugs.launchpad.net/neutron/+bugs?field.tag=l3-dvr-backlog
15:06:41 <Swami> There are three critical bugs right now and I think two have been already addressed and fix commited.
15:07:05 <Swami> #link https://bugs.launchpad.net/neutron/+bug/1373100
15:07:07 <uvirtbot> Launchpad bug 1373100 in neutron "New race condition exposed when cleaning up floating ips on router delete" [Critical,In progress]
15:07:18 <carl_baldwin> That is the only critical still open.
15:07:23 <haleyb> I might have another new one, will wait until you're through the others
15:07:26 <Swami> This above bug is currently in progress, I am not sure if it can make it to the Juno RC-1 cut date.
15:08:00 <Swami> carl_baldwin: haleyb: We also found another bug where we cannot delete the "snat_namespace".
15:08:44 <Swami> But this fix is already part of the Mike's patch for the snat_move. So I have not raised an additional bug on this.
15:09:38 <haleyb> We have seen "DVR: Failed to obtain local DVR Mac address" occasionally on our internal CI system, a fix of the openvswitch agent fixes things.  I'll file a bug but didn't know if others had seen it
15:10:20 <Swami> haleyb: no I don't think we recall any such bug internally
15:10:41 <Swami> haleyb: please go ahead and file the bug.
15:10:46 <viveknarasimhan> that is known bug
15:10:52 <viveknarasimhan> bhailey
15:10:55 <viveknarasimhan> it is being addressed by
15:11:01 <Swami> viveknarasimhan: Have we filed a bug on this.
15:11:17 <viveknarasimhan> one sec
15:11:21 <viveknarasimhan> yeah
15:11:24 <viveknarasimhan> retrieving bug ID now
15:12:12 <haleyb> is that the race condition one you've been chasing?
15:12:25 <viveknarasimhan> https://bugs.launchpad.net/bugs/1364215
15:12:26 <uvirtbot> Launchpad bug 1364215 in neutron "L2 Agent switch to non-dvr mode on first RPC failure" [High,Confirmed]
15:13:06 <viveknarasimhan> a restart of openvswitch agent is the workaround for that problem
15:13:38 <Swami> Vivek, is there a fix for this bug right now.
15:14:02 <Swami> haleyb: can you check with vivek and see both your fixes addresses the same problem.
15:14:36 <haleyb> Yes, vivek feel free to ping me afterwards as we see this all the time now
15:14:37 <viveknarasimhan> i was working on a fix
15:14:43 <viveknarasimhan> i haven't completed the fix testing
15:15:03 <viveknarasimhan> if node is started before controller rabbitmq cluster comes up
15:15:09 <Swami> viveknarasimhan: Yes sure, you can check with haleyb on your proposal and then complete the patch.
15:15:13 <viveknarasimhan> this problem will be seen, as the first dvr rpc call from l2-agent on node will fail
15:15:20 <viveknarasimhan> and it will throw Failed to obtain DVR Mac address
15:16:38 <viveknarasimhan> if you often see it , it might be controller problem rather than oopenvswitch aget problem
15:16:49 <Swami> ok haleyb and viveknarasimhan will sort this offline and push the code to address the bug 1364215. If the symptoms are different then go ahead and file a new bug.
15:16:50 <uvirtbot> Launchpad bug 1364215 in neutron "L2 Agent switch to non-dvr mode on first RPC failure" [High,Confirmed] https://launchpad.net/bugs/1364215
15:16:54 <viveknarasimhan> since no controller is servicing the rpc, dvr rpc will timeout and post the error
15:17:30 <Swami> ok moving on
15:17:53 <Swami> For the other "snat_namespace" not getting deleted bug we should probably use the patch shown below.
15:18:02 <Swami> #link https://review.openstack.org/#/c/122024/
15:18:28 <Swami> But if this is critical we can split this up into a separate patch so that it does not hold the RC1 build.
15:20:12 <Swami> carl_baldwin: Another bug that should be targetted for RC1 is https://bugs.launchpad.net/neutron/+bug/1356467
15:20:13 <uvirtbot> Launchpad bug 1356467 in neutron "VPNaaS behavior undetermined with DVR router" [Low,In progress]
15:20:52 <Swami> This has a "low" priority but still it makes sense since we don't support VPN with DVR, we don't want anyone to get panic when VPN does not work with DVR.
15:21:33 <Swami> #link https://review.openstack.org/#/c/121038/
15:21:43 <carl_baldwin> Swami: It is noted.  It may be difficult to make a case for this now unless the implementation is completely solid already.
15:21:58 <carl_baldwin> Even then…
15:22:09 <Swami> carl_baldwin: Yes I think right now it is in review for a while with issues on UT.
15:23:59 <Swami> carl_baldwin: Ok what ever is possible we should do.
15:24:06 <carl_baldwin> Feedback needs to be addressed ASAP.
15:24:24 <Swami> Yes I will ask vikas to address the review comments.
15:24:30 <carl_baldwin> otherwise it is a not-starter.
15:24:37 <Swami> ok
15:25:54 <Swami> That's all we have for the bugs.
15:27:20 <Swami> #topic Open Discussion
15:28:27 <viveknarasimhan> i want to let know
15:28:42 <viveknarasimhan> that I have started implementation of VLAN support for dvr
15:28:46 <Swami> viveknarasimhan: yes
15:28:54 <viveknarasimhan> expect to post a WIP code early next week for review
15:29:10 <viveknarasimhan> but may not be able to make it to Juno due to timelines
15:29:28 <Swami> viveknarasimhan: that is fine. This is not a feature that we targeted for Juno
15:29:36 <Swami> You can push the wip code.
15:29:45 <viveknarasimhan> sure, will do
15:30:15 <Swami> You can add a backlog item in DVR and then push a wip code with the bug ID.
15:30:34 <Swami> Does that sound reasonable.
15:30:40 <viveknarasimhan> ok. sure
15:30:46 <Swami> viveknarasimhan: How big is this change.
15:31:07 <viveknarasimhan> predominantly L2 only change
15:31:22 <viveknarasimhan> no plugin change (or) L3 Side changes
15:31:44 <Swami> viveknarasimhan: Probably we need more testing in these area along with our existing code.
15:31:46 <viveknarasimhan> have not completed all changes, so will not be able to give right LOC
15:31:53 <viveknarasimhan> yes, agreed
15:32:52 <Swami> next week we will revisit the topic for the DVR for Kilo.
15:33:28 <Swami> #link https://etherpad.openstack.org/p/kilo-neutron-summit-topics
15:34:11 <Swami> The above link is the Kilo etherpad link for neutron. Any new features for Kilo should go in into the etherpad for internal discussion
15:35:04 <viveknarasimhan> ok
15:35:09 <Swami> Thanks all for joining the meeting
15:35:32 <Swami> See you all next week with a plan for the Kilo.
15:35:43 <Swami> bye
15:35:49 <Swami> #endmeeting