15:01:42 <haleyb> #startmeeting neutron_dvr
15:01:43 <openstack> Meeting started Wed Jun  1 15:01:42 2016 UTC and is due to finish in 60 minutes.  The chair is haleyb. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:01:44 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:01:46 <openstack> The meeting name has been set to 'neutron_dvr'
15:01:54 <haleyb> #chair Swami
15:01:55 <openstack> Current chairs: Swami haleyb
15:01:58 <haleyb> Swami: hi
15:02:15 <Swami> good morning/good afternoon
15:02:37 <haleyb> it's still morning here
15:03:01 <haleyb> #topic Announcements
15:03:04 <Swami> haleyb: fine
15:04:00 <haleyb> only announcement is N-1 is this week
15:04:36 <haleyb> #topic Bugs
15:04:58 <Swami> This week, we had two new bugs filed.
15:05:12 <Swami> https://bugs.launchpad.net/neutron/+bug/1582739
15:05:12 <openstack> Launchpad bug 1582739 in neutron "[DVR][L3 HA] Unable to ping 8.8.8.8 from VM without floating ip" [Medium,In progress] - Assigned to Oleg Bondarev (obondarev)
15:05:31 <Swami> This bug has a patch associated with it.
15:05:44 <Swami> #link https://review.openstack.org/#/c/317541/2
15:05:54 <haleyb> I just reviewed, only issue was small in the test
15:06:00 <Swami> This patch needs another +2. Oleg has already reviewed it.
15:06:11 <Swami> haleyb: ok thanks
15:06:35 <Swami> It seems to be a trivial fix, no issues there to discuss.
15:06:39 <haleyb> there was an import added in the middle of the file, should be at top.  i wish pep8 would flag that
15:06:40 <Swami> Moving on to the next one.
15:06:58 <haleyb> yes, otherwise fix is fine
15:06:58 <carl_baldwin> I can review it too if needed.
15:07:10 <Swami> carl_baldwin: thanks carl.
15:07:25 <Swami> #link https://bugs.launchpad.net/neutron/+bug/1581348
15:07:25 <openstack> Launchpad bug 1581348 in neutron "Can't delete a v4 csnat port when there is a v6 router interface attached" [Medium,In progress] - Assigned to Hong Hui Xiao (xiaohhui)
15:07:44 <Swami> This bug also has a patch associated with it.
15:08:08 <Swami> #link https://review.openstack.org/#/c/315926/ ( patch)
15:08:39 <Swami> I did review this patch.
15:08:53 <Swami> There was a follow on patch for some name change shown below.
15:09:05 <Swami> #link https://review.openstack.org/#/c/319037/1
15:10:08 <Swami> These are the two new bugs for this week.
15:10:17 <haleyb> Swami: i will take a look at those two reviews
15:10:27 <Swami> haleyb: thanks
15:11:33 <Swami> haleyb: carl_baldwin: I have addressed review comments on this patch. Can you take a look at it and provide your feedback. #link  https://review.openstack.org/#/c/300358/
15:11:47 <carl_baldwin> Swami: Yes, I will.
15:12:55 <Swami> carl_baldwin: I also added a comment on this patch. #link https://review.openstack.org/#/c/304905/, I got your point and the only way to address this is to filter this by the tenant id. Please let me know if that would work.
15:17:02 <haleyb> Swami: anything else?
15:17:37 <Swami> On the nova side patch for live migration, hardik is testing the tempest test to validate it, I am seeing some progress, let us see how it goes.
15:17:58 <haleyb> +1 on that
15:18:00 <Swami> That's all on the bugs.
15:18:01 <carl_baldwin> Swami: sorry, I'll look.
15:18:20 <Swami> On the proposed fix for the Allowed_address_pair for unbound ports.
15:19:14 <Swami_> https://bugs.launchpad.net/neutron/+bug/1583694
15:19:14 <openstack> Launchpad bug 1583694 in neutron "[RFE] DVR support for Allowed_address_pair port that are bound to multiple ACTIVE VM ports" [Wishlist,In progress] - Assigned to Swaminathan Vasudevan (swaminathan-vasudevan)
15:19:25 <Swami_> Here are the patches for review.
15:19:41 <Swami_> #link https://review.openstack.org/#/c/320669/ ( Server side patch)
15:20:03 <Swami_> haleyb: I saw your comment on using the config options versus using the device_owner, let me see how I can incorporate it.
15:20:14 <Swami_> Here is the link to the agent side patch.
15:20:28 <Swami_> #link https://review.openstack.org/#/c/323618/
15:21:09 <haleyb> Swami_: yeah, it was really to try and make it portable to stable
15:21:18 <Swami_> haleyb: carl_baldwin: could you provide your feedback on the agent side patch. I did see some comments on the RFE from armax, that he was not comfortable moving in this direction. But I also wanted your opinion on this.
15:22:15 <carl_baldwin> Swami_: I'll read it.
15:22:21 <Swami_> carl_baldwin: we can still try and see if the option 1 discussed in the RFE is possible, but I thought that would be little more complex to poll on the GARP for every floatingip and take necessary action.
15:22:37 <carl_baldwin> Swami_: Back on https://review.openstack.org/#/c/304905/10/neutron/db/l3_dvr_db.py@262, can't we just go straight to the port ID?
15:23:10 <Swami_> carl_baldwin: we don't have the port_id, we only have the allowed_address_pair port id and not the VM port id.
15:24:00 <carl_baldwin> ok
15:25:18 <Swami_> That's all I had for bugs.
15:25:40 <Swami_> haleyb: Did you add in the bug for the metering that you discussed last week.
15:26:19 <haleyb> Swami_: no, i need to, which brings up a somewhat related change
15:26:23 <haleyb> https://review.openstack.org/#/c/277657/12/neutron/agent/linux/pd.py
15:27:12 <Swami_> haleyb: got it.
15:27:17 <haleyb> That was Ritesh's change for fixing prefix delegation with DVR, but carl had a good point that we maybe need a better way to find what namespace should be used
15:27:40 <haleyb> the metering code has a similar issue in that it needs to use the fip namespace when running on a compute node
15:27:58 <Swami_> haleyb: should we be then checking in for the network and then decide the namespace, because it might not be trivial.
15:28:58 <haleyb> carl_baldwin: what was your thought on that change?
15:29:11 * carl_baldwin looking...
15:30:27 <carl_baldwin> I just thought that this code was needing too much knowledge of DVR internals.
15:31:38 <haleyb> carl_baldwin: yes, it seems to, i'm just thinking where should this abstraction live, namespace code?
15:33:18 <Swami_> haleyb: probably we can add this function in the l3 agent to return the router_gw_ns_name
15:33:43 <haleyb> Swami_: sure
15:33:53 <Swami_> So from the pd standpoint, it can just call the router.router_gw_ns_name and it should return the right namespace.
15:35:39 <haleyb> yes, i'll take a look based on that change
15:35:54 <haleyb> any other bugs/patches to discuss?
15:36:36 <Swami_> haleyb: nothing else.
15:37:17 <haleyb> #topic Gate Failures
15:37:54 <haleyb> gate has calmed recently, don't have any specific dvr issues i've seen
15:38:07 <Swami_> I did see the single node job failure rates are slightly higher than normal.
15:38:42 <Swami_> I just noticed from the graphs, but did not get a chance to look at the details.
15:40:47 <haleyb> #topic Stable backports
15:41:24 <haleyb> Swami_: i think you had some backports that were stuck
15:41:33 <Swami_> haleyb: yes, I did see your comments on it.
15:41:49 <haleyb> i don't think i swayed the voters :(
15:41:57 <Swami_> haleyb: did you get a chance to talk to iharchys
15:42:46 <Swami_> we need those to be pushed in if possible.
15:43:32 <haleyb> Swami_: sorry, which patches?  i might have to go to my review list
15:44:08 <Swami_> #link https://review.openstack.org/#/c/313130/ this is the parent.
15:44:16 <Swami_> there are around six child patches to it.
15:45:13 <haleyb> ok, that was the patch i was looking at, i'll see what i can do
15:45:21 <Swami_> haleyb: thanks
15:46:19 <haleyb> #topic Open discussion
15:47:12 <Swami_> haleyb: I wanted to discuss if there is a way to poll on the GARP, messages on the namespace for a particular IP and report it back to the agent.
15:48:39 <haleyb> meaning if a GARP happens?
15:49:10 <Swami_> haleyb: yes.
15:49:59 <Swami_> haleyb: The reason I ask this is for the allowed_address_pair and fip in DVR, the alternate proposal is to poll on the GARP and then trigger the floatingip from the node, were the mac in the GARP message resides.
15:50:01 <haleyb> I'm not sure we can learn when that happens
15:50:31 <Swami_> haleyb: ok, I knew that it would not be easy, but just wanted to check.
15:51:22 <Swami_> haleyb: kevinbenton mentioned it in the RFE and I have also asked him to help me if he has any ideas on it.
15:51:46 <haleyb> he would be a better resource for that
15:52:31 <Swami_> haleyb: ok let us see what option he has.
15:53:08 <haleyb> anything else?
15:53:18 <Swami_> that's all i had in my list.
15:54:05 <haleyb> #endmeeting