16:03:34 <rkukura> #startmeeting networking_ml2
16:03:35 <openstack> Meeting started Wed Jan 21 16:03:34 2015 UTC and is due to finish in 60 minutes.  The chair is rkukura. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:03:36 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:03:38 <openstack> The meeting name has been set to 'networking_ml2'
16:04:16 <rkukura> agenda is at https://wiki.openstack.org/wiki/Meetings/ML2#Agenda
16:04:18 <nuritv> Hi
16:04:48 <rkukura> #topic Announcements
16:05:18 <rkukura> Only announcement from me is that kilo-2 is fast approaching, on Feb 5
16:05:29 <rkukura> Any other announcements?
16:06:13 <rkukura> #topic ML2 Drivers decomposition discussion
16:06:59 <Sukhdev> ODL ML2 decomposition is complete and can be used as an example
16:07:30 <rkukura> Sukhdev: Right. I’ve looked it over, and think it is a good example, with the code using the driver API remaining in the tree.
16:07:36 <Sukhdev> Neutron side as well as stackforge side codes are merged into repos
16:08:22 <rkukura> Sukhdev: I haven’t looked closely at the unit tests - did they need to do much work to split those?
16:08:45 <Sukhdev> Yes, the unit tests are bit tricky
16:09:23 <Sukhdev> I am having some issues with those as well - worked with armax yesterday and made progress
16:09:58 <Sukhdev> the unit tests on the neutron side need to be very thin - no imports from the stackforge side
16:10:18 <Sukhdev> I was making that mistake and hence was having issue
16:10:26 <moshele> sukhdev: can you explain why? because in the mlnx MD I removed all of them from neutron
16:11:27 <Sukhdev> moshele: if you removed them from neutron side, then that is good.
16:12:13 <rkukura> I’d think we’d want UTs for the code that remains in-tree, probably mocking the calls to the external library
16:12:20 <Sukhdev> to get the decomposition right, one has to get one side right first (in this case Neutron side) with no reference to the other side
16:12:40 <Sukhdev> Once the neutron side is happy, then get the other side working
16:13:10 <rkukura> asomya, HenryG, dane_leblanc: Any update on Cisco MD decomposition?
16:13:19 <Sukhdev> once both sides are happy, then we can add the dependencies so that each side can see other side
16:14:45 <rkukura> Or updates on any other drivers?
16:14:57 <Sukhdev> I am hoping to complete the decomposition of Arista MD by this week (hopefully)
16:15:02 <moshele> mlnx is WIP https://review.openstack.org/#/c/148614/
16:15:51 <rkukura> moshele: thanks
16:16:07 <rkukura> Anything else on decomposition before we move on?
16:16:39 <Sukhdev> moshele: that looks good
16:16:54 <moshele> thanks
16:17:11 <HenryG> rkukura: the cisco decomposition is "in progress"
16:17:13 <Sukhdev> moshele: I will look more carefully later...
16:17:15 <asomya> rkukura: Cisco MD decomposition is also WIP https://review.openstack.org/#/c/145793/
16:17:38 <moshele> sukhdev: sure, take you time
16:18:38 <rkukura> asomya: thanks
16:18:50 <rkukura> Anything else on decomposition?
16:19:16 <rkukura> #topic ML2 Sync and error handling (Task Flow)
16:19:55 <Sukhdev> manishg is not here
16:19:57 <rkukura> I don’t see manishg, who Sukhdev and I were hoping would be able to provide an update
16:20:30 <Sukhdev> We have been working on this prior to Paris Summit
16:20:36 <Sukhdev> and wanted to pick it up
16:20:55 <Sukhdev> manishg had volunteered to look into taks flow
16:20:58 <rkukura> Is there still general interest in a plugin-level solution for L, or would people rather deal with it in their own drivers?
16:21:18 <Sukhdev> I was hoping he could provide us the update
16:21:37 <rkukura> Lets keep it on the agenda for next week then, and encourage manishg to attend
16:21:40 <Sukhdev> We discussed this in the PODs at Paris Summit, remember?
16:21:51 <banix> Well we need something for cases where for example update fails in a mech driver
16:22:31 <banix> That cannot be done by mech drivers
16:22:33 <rkukura> Sukhdev: I do remember the discussions, and am happy to help work out a plan for L
16:22:48 <rkukura> Anything else on this before we move on?
16:23:37 <rkukura> #topic Blueprint list consolidation
16:24:08 <rkukura> I believe manishg and shivharis were going to work on this, but neither is here today.
16:24:26 <rkukura> #link https://wiki.openstack.org/wiki/Tracking_ML2_Subgroup_Reviews
16:25:24 <rkukura> My understanding is that the only two non-vendor-specific ML2 BPs accepted for kilo are HPB and portsecurity. Are there others we should track at these meetings?
16:26:36 <rkukura> I’m back working on HPB. Its currently blocked by failures in the DVR tempest tests, and I’m working to figure out what is going on.
16:27:18 <rkukura> HPB is currently targeted for kilo-3, but I’m hopeful to get it merged for kilo-2 if we can resolve the DVR issue.
16:28:16 <matrohon> rkukura : link?
16:28:16 <rkukura> If anyone cares to look through the logs at http://logs.openstack.org/22/116122/16/check/check-tempest-dsvm-neutron-dvr/9247857/, please let me know if you come up with any theories!
16:29:09 <rkukura> matrohon: There are three patches, but the DVR failure is introduced in the 2nd, which changes the DB schema: https://review.openstack.org/116122
16:29:45 <banix> 
16:30:17 <matrohon> but the dvr tempest test is non voting
16:31:46 <rkukura> matrohon: Still need to at least understand the issue before it merges. It consistently fails.
16:31:54 <rkukura> The other ML2 BP is for portsecurity, and is targeted for kilo-2 I believe.
16:32:18 <yamahata> rkukura: yes. K-2 is still the target
16:32:21 <rkukura> current portsecurity patch is https://review.openstack.org/#/c/126552/
16:33:11 <yamahata> Sheta is working on extension driver side
16:33:14 <rkukura> shwetaap: I see you updated this, and I gave you a pointer to the GBP extension driver test which should be a good example of how these work
16:33:19 <yamahata> which is not uploaded yet.
16:34:31 <shwetaap> rkukra: thanks! I am making changes to my code right now. I shall work with Yalei and get my changes up and the same patch.
16:34:39 <rkukura> I think this is good candidate to implement as an extension driver, but would not object to building it into ML2 if doing it as a driver turned out to be problematic.
16:35:15 <rkukura> shwetaap: OK, I’ll try to promptly review the patches
16:36:14 <rkukura> There may be a few more BPs that touch ML2 even if they aren’t focused on it, that we should track going forward. Hoping the wiki page will facilitate this
16:36:36 <rkukura> Anything else on kilo BPs before we move on?
16:37:19 <rkukura> #topic Bugs
16:37:36 <rkukura> shivharis usually handles this, but isn’t here
16:38:02 <rkukura> Lets go over the 3 linked in the agenda…
16:38:16 <rkukura> https://bugs.launchpad.net/neutron/+bug/1179223
16:38:37 <rkukura> romilg: Any update on this?
16:39:17 <romilg> I have posted the patch-set long back need reviewers to review it
16:39:40 <romilg> review link:  https://review.openstack.org/#/c/136106/
16:39:44 <rkukura> romilg: I just added it to my queue
16:39:53 <romilg> Thanks :)
16:40:16 <rkukura> Looks like a couple of cores have reviewed previous versions, so should be getting there
16:40:48 <rkukura> next bug: https://bugs.launchpad.net/neutron/+bug/1179223
16:40:59 <rkukura> sorry, that’s the same one
16:41:26 <rkukura> how about matrohon’s: https://bugs.launchpad.net/neutron/+bug/1179223
16:42:03 <matrohon> rkukura : I just wanted to know the status of this work?
16:42:06 <rkukura> That’s mine actually - back to working on that this week as soon as the HPB DVR issue is understood
16:42:37 <rkukura> I’ll have an update for next meeting
16:42:39 <matrohon> rkukura : ok. still targeted for k 2
16:42:43 <matrohon> ?
16:42:47 <rkukura> I’m hopeful
16:43:43 <rkukura> The HPB DVR issue seems to also have to do with the kind of DB schema changes planned for this bug, so these are closely related
16:43:54 <matrohon> rkukura : FYI this would be very useful for : https://review.openstack.org/#/c/141114/
16:44:22 <rkukura> matrohon: understood, and will look closely at that
16:44:55 <matrohon> ther previous patch try to workaround the lack of multi-bound port in ML2
16:45:24 <matrohon> so we wonder if you'd better wait or not
16:45:32 <matrohon> s/you/we
16:46:01 <rkukura> matrohon: Lets add discussion of this to next week’s agenda, when we’ll hopefully have a better idea about this bug and the HPB/DVR issue
16:46:17 <matrohon> rkukura : ok
16:47:31 <rkukura> Any other bugs to discuss?
16:48:32 <rkukura> #topic Open Discussion
16:48:36 <GLaupre_> (I missed the begining of the meeting, is there a record of the irc bot?)
16:49:10 <matrohon> this patch is ready to merge but need some core : https://review.openstack.org/#/c/102543/
16:49:11 <rkukura> GLaupre_: bot seems to be working
16:50:58 <banix> GLaupre_: yes check the meeting wiki page please
16:51:06 <rushiagr> GLaupre_: there is. Find logs herehttp://eavesdrop.openstack.org/meetings/cinder/2015/
16:51:23 <rkukura> matrohon: Maybe see about bumping the bug to medium
16:51:59 <rkukura> GLaupre_: http://eavesdrop.openstack.org/meetings/networking_ml2/
16:52:00 <matrohon> rkukura : good idea, i'll do
16:52:10 <rkukura> Anything else today?
16:52:16 <GLaupre_> tks a lot :)
16:52:21 <banix> GLaupre_: #link http://eavesdrop.openstack.org/meetings/networking_ml2/
16:52:38 <rkukura> lets wrap up a bit early then...
16:52:44 <rkukura> Thanks everyone!
16:52:45 <banix> thanks all
16:52:48 <rkukura> #endmeeting