15:00:57 <carl_baldwin> #startmeeting neutron_l3
15:00:58 <openstack> Meeting started Thu Jul 17 15:00:57 2014 UTC and is due to finish in 60 minutes.  The chair is carl_baldwin. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:59 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:01:02 <openstack> The meeting name has been set to 'neutron_l3'
15:01:14 <carl_baldwin> #topic Announcements
15:01:26 <carl_baldwin> Juno-2 is in one week.
15:01:49 <carl_baldwin> spec approval deadline (SAD) is in three days.
15:02:00 <carl_baldwin> What an appropriate acronym.
15:02:35 <carl_baldwin> The mid-cycle sprint last week went well.  I enjoyed seeing some of you there.
15:02:52 <pcm_> +1
15:03:23 <carl_baldwin> #link https://etherpad.openstack.org/p/neutron-msp-sprint
15:03:59 <carl_baldwin> We made good progress on DVR among other topics.
15:04:08 <carl_baldwin> Any other announcements?
15:04:35 <carl_baldwin> #topic neutron-ovs-dvr
15:04:59 <carl_baldwin> One patch merged yesterday.  I think we could be close on the L2 patch today.
15:05:44 <carl_baldwin> Swami: hi
15:05:47 <Swami> hi carl
15:05:55 <carl_baldwin> Your patch merged!
15:06:15 <Swami> Great!, Thanks to Carl and Armando
15:06:25 <carl_baldwin> #link https://wiki.openstack.org/wiki/Neutron/DVR/HowTo
15:06:47 <carl_baldwin> With the API patch in, we can work on getting the DB Models patch and the L3 Agent patch in.
15:07:42 <Swami> Is sridharK here
15:07:42 <carl_baldwin> I fell asleep before getting through the latest version of the DB patch but I plan to finish up my review this morning.
15:07:44 <viveknarasimhan> thanks Carl for getting merged  the central patch for DVR merged
15:07:48 <SridarK> Swami: yes
15:08:17 <carl_baldwin> Swami: SridarK: you have the floor.
15:08:19 <armax> carl_baldwin: I pushed another revision for the L2 models DB patch
15:08:26 <SridarK> carl_baldwin: thx
15:08:38 <armax> carl_baldwin: I addressed Anna’s comments
15:08:44 <viveknarasimhan> i would like to mention something
15:08:44 <SridarK> we are working with the DVR team to address FWaaS - DVR integration issues
15:08:53 <SridarK> #link https://review.openstack.org/#/c/106225/
15:09:01 <carl_baldwin> armax: I noticed it this morning.  Thanks.  We should be able to review that quickly.
15:09:03 <SridarK> armax: thanks for the review
15:09:25 <viveknarasimhan> Since we merged Swami's patch which takes l3_dvr_db, we need upgrade the scheduler patch
15:09:31 <armax> carl_baldwin: yeah…my bad I was careful in making sure to address the new needs of db migrations, but not so careful!
15:09:32 <viveknarasimhan> to take the new mixin available there
15:09:49 <viveknarasimhan> we found this out during our Unit Test upgrade for DVR scheduler
15:10:01 <armax> viveknarasimhan: can you be more detailed on the review itself?
15:10:15 <viveknarasimhan> ok, will add review lines there
15:10:17 <armax> viveknarasimhan: I am still waking up, so I am not sure I follow right now ;)
15:10:21 <viveknarasimhan> for dvr scheduler patch
15:10:56 <armax> viveknarasimhan: please put a comment, a -1 if you need to and we’ll look into that
15:11:04 <viveknarasimhan> ok
15:11:05 <carl_baldwin> viveknarasimhan: We’ve got a couple days before Murali’s patch his the top of the review-for-merge list.  But, the sooner the feedback gets in the better.
15:11:15 <viveknarasimhan> doing so now
15:11:15 <Swami> vivek: Are you saying that since we already merged the L3 extension patch , the scheduler patch should no longer have any dependency
15:12:09 <viveknarasimhan> no, am saying
15:12:17 <viveknarasimhan> scheduler should start using the new mixin in l3_dvr_db
15:12:25 <viveknarasimhan> it is using the old NAT mixin
15:12:26 <viveknarasimhan> from l3_db
15:12:33 <carl_baldwin> SridarK: I have your spec at the top of my review list for today.
15:12:39 <Swami> vivek: got it
15:12:43 <SridarK> carl_baldwin: thx
15:13:12 <Swami> carl_baldwin: Just give you heads up on the FWaaS
15:13:34 <Swami> As per our discussion the current blue print will target the North South and not the East-West.
15:13:50 <carl_baldwin> Swami: understood, thanks.
15:14:00 <Swami> To be backward compatible, the FWaaS will only apply the FWaaS rules to the non-distributed routers.
15:14:15 <SridarK> carl_baldwin: Swami have called that out this is targeting N - S in the spec
15:14:41 <carl_baldwin> Swami: So, no FWaaS for distributed routers at all?
15:14:50 <armax> SridarK: I believe it would be useful to really make it clear in the spec title or commit message
15:15:10 <SridarK> armax: i have addressed ur comments on that in the spec also
15:15:16 <viveknarasimhan> FwaaS for distributed routers but only for
15:15:17 <viveknarasimhan> N-S traffic
15:15:27 <SridarK> viveknarasimhan: +1
15:15:27 <viveknarasimhan> for the first phase carl
15:15:30 <armax> SridarK: thanks for that, but the commit or spec title is unchanged
15:15:46 <Swami> carl_baldwin: sorry if I have conveyed the message properly, FWaaS will be there for distributed routers, but only for the north-south traffic.
15:15:48 <SridarK> armax: ok will change that
15:15:55 <armax> later on we’re gonna file other blueprints to address the other parts
15:16:09 <SridarK> armax: +1
15:16:09 <carl_baldwin> I’m a bit confused now by “FWaaS will only apply the FWaaS rules to the non-distributed routers” but I’ll read the blueprint and take the discussion there.
15:16:31 <carl_baldwin> Swami: I think I understand.
15:16:50 <SridarK> carl_baldwin: we only want to address the N - S cases and make sure we don't break E - W normal forwarding
15:17:17 <armax> SridarK: thanks, it’s only a suggestion but I think it’ll make the scope of the bp a lot more clear
15:17:18 <SridarK> *address N - S FWaaS; no E - W FWaaS
15:17:20 <carl_baldwin> SridarK: That sounds like what I’m thinking now.  So, I think I understand.
15:17:45 <carl_baldwin> armax: Good suggestion.  Making the scope of a bp clear is often the most important part.
15:17:48 <SridarK> armax: i agree too so there is no confusion on what is being addressed
15:18:12 <SridarK> armax: removed the Phase 2 part from the BP based on ur comments
15:18:15 <armax> SridarK: you may want to keep the same topic for all the fw/dvr bp's
15:18:25 <carl_baldwin> So, we’re good to go on this bp?  We will work hard to get this reviewed by the 20th.
15:18:27 <armax> SridarK: but I’d be definitely more descriptive in the commit messafge
15:18:38 <SridarK> armax: will change that
15:19:04 <armax> SridarK: so that’s easy to group them together and see what the overall effort is
15:19:31 <carl_baldwin> I think we should spend a few minutes discussing DVR testing.
15:19:35 <SridarK> carl_baldwin: we have discussed with the DVR team and are on the same page with them on what needs to be done
15:19:50 <SridarK> armax: yes definitely agree
15:19:50 <carl_baldwin> SridarK: great.
15:20:12 <SridarK> carl_baldwin: thanks for some "airtime" here
15:20:33 <SridarK> carl_baldwin: badveli will also be working on this
15:20:41 <carl_baldwin> SridarK: yw, thanks for your work.
15:20:43 <Rajeev_> SridarK: we like how you are approaching the design for DVR
15:20:46 <SridarK> from the FWaaS sub team
15:20:46 <badveli> yes on the implementation
15:21:17 <carl_baldwin> On the testing front.  We’ve got a few things going on, I think.
15:21:22 <SridarK> carl_baldwin: Rajeev_ thx for to the DVR sub team and FWaaS subteam
15:21:51 <viveknarasimhan> carl i am working on enhancing the Unit tests for scheduler
15:21:58 <viveknarasimhan> and ml2 pluin (l2-pop side)
15:22:07 <armax> carl_baldwin: I revised the tempest API tests
15:22:17 <carl_baldwin> First, UT coverage has been called out in a few of the patches.  I spoke with all of the cores and Kyle at the sprint.  We’re okay merging the patches with the current coverage but we’ll need to follow up.
15:22:31 <viveknarasimhan> great... thanks Carl
15:22:49 <viveknarasimhan> but we will be shipping UT stuff as we improve coverge for review to you and armax
15:23:23 <armax> viveknarasimhan: great, sooner the better
15:24:50 <carl_baldwin> We’ll need the new coverage in order to finally say that DVR has been fully implemented.  Yes, the sooner the better.
15:25:09 <carl_baldwin> 2) Tempest.  armax:  it sounds like you’ve got this covered?
15:25:09 <Swami> carl_baldwin: on the testing front, is there any option of introducing the multinode setup in the upstream or are we still targeting a single node.
15:25:12 <armax> carl_baldwin: the Tempest API tests are now better designed and the review passed the existing test suite (w/o dvr)
15:25:41 <carl_baldwin> armax: Great.
15:26:15 <armax> carl_baldwin: checking if they pass now with DVR now that swami’s patch merged
15:26:16 <carl_baldwin> Swami: There is an infra bp to add multi-node testing.  It will take a little time so, for now, a single node is all we have.
15:26:42 <armax> carl_baldwin: we also have a number of patches to add the CI jenkins job to address the single node tests
15:26:50 <armax> but no love from the infra team yet
15:27:23 <carl_baldwin> #link https://review.openstack.org/#/c/106495/
15:27:32 <carl_baldwin> ^ The infra spec for multi-node.
15:27:50 <armax> #link https://review.openstack.org/#/c/106902/
15:28:08 <armax> #link https://review.openstack.org/#/c/106901/
15:28:20 <armax> #link https://review.openstack.org/#/c/103180/
15:28:45 <armax> the last tree links should give us the ability to run an experimental job at will that has DVR enabled
15:29:18 <armax> before we can really do that though we’d need the entire dvr feature in the tree
15:30:14 <carl_baldwin> armax: Thanks for the links.  Right, we need at least the basic set of 7 patches merged, right?  1 down 6 to go.
15:30:34 <armax> carl_baldwin: yes
15:31:56 <carl_baldwin> Hopefully we can get the infra team’s attention before then.  I think they might be meeting in Germany this week.  They may be heads down.
15:32:16 <armax> ok
15:32:26 <Swami> good to know.
15:33:00 <carl_baldwin> I have one more thing for DVR.  Is there a plan to enable GRE tunneling?  I’m told that would be easy.  What about VLAN?
15:33:17 <viveknarasimhan> GRE tunneling can be enabled
15:33:38 <viveknarasimhan> did a basic investigation and it needs change on the L2 OVS agent for most part
15:34:21 <carl_baldwin> viveknarasimhan: Can you start a placeholder patch for that?
15:34:27 <viveknarasimhan> yes, will do carl
15:34:46 <Swami> carl: we have a couple of items in the backlog and we need to prioritize
15:35:12 <Swami> Migration, Services support, support VLAN, GRE tunnelling etc.,
15:35:31 <Swami> I have a list I can pass it on to you offline.
15:35:48 <carl_baldwin> Swami: great.  Let me know how that backlog gets prioritized.
15:36:10 <carl_baldwin> Swami: It will be great to have that list.
15:36:11 <Swami> carl_baldwin: sure
15:36:17 <carl_baldwin> Anything else for DVR?
15:36:23 <Swami> Do you want it right now.
15:36:31 <carl_baldwin> Swami: If you have it, yes.
15:36:38 <Swami> hold on a minute.
15:36:57 <carl_baldwin> Swami: sure just email it.
15:37:00 <carl_baldwin> #topic l3-high-availability
15:37:25 <Swami> 1.	Migration  a.	Legacy to Distributed Router Migration i.	With no Gateway and FloatingIP ii.	With Gateway only iii.	With no FloatingIP and  iv.	With FloatingIP v.	With FloatingIP and Gateway vi.	Allow removing a distributed router from an agent binding even if it has FloatingIP.  b.	Migration CLI Commands  i.	Use existing CLI commands ii.	Or Create new CLI commands to address the SNAT migration  2.	Services a.	LBaaS Sup
15:37:30 <armax> safchain:  udere?
15:37:31 <carl_baldwin> safchain: armax: How are we doing here?  I saw that the bp merged.
15:37:41 <carl_baldwin> Swami: thanks.
15:37:42 <armax> we’re at risk here
15:37:47 <Swami> Sorry it is not well formated.
15:37:52 <Swami> I will add one by one
15:38:05 <carl_baldwin> Swami: no worries.
15:38:07 <armax> I don’t see Assaf
15:38:39 <armax> carl_baldwin: safchain said he didn’t have cycles to work on it then he came back work on it a little and  then disappeared again
15:38:56 <safchain> armax, carl, hi sorry to be late
15:39:04 <carl_baldwin> safchain: hi
15:39:25 <armax> carl_baldwin: to make sure we make consistent progress here we’d need someone who steadily work on this feature
15:39:49 <carl_baldwin> I was thinking about HA + DVR the other day.  The thing that came to my mind was that HA would need to be aware of the SNAT port that DVR allocates.
15:40:01 <safchain> armax, we are working on it with assaf
15:40:03 <Swami> carl_baldwin: I will send an email with the complete list.
15:41:01 <carl_baldwin> safchain: How do you feel about hitting juno-3?
15:41:31 <armax> safchain: no much progress the last couple of days, is there anything we can do?
15:41:34 <armax> to help?
15:42:28 <safchain> carl_baldwin, armax Assaf reviewed all the patches, and we split the tasks in two part, API/scheduler - agent
15:42:47 <safchain> armax, yes any help is welcome
15:43:23 <safchain> armax, please tell me if you want to work on a specific part, patch
15:43:23 <armax> safchain: well we don’t want to step on each other toes
15:43:46 <armax> safchain: we cannot work on the same patches at the same time
15:43:51 <safchain> armax, sure
15:44:02 <armax> safchain: where do you feel you need help?
15:44:59 <safchain> armax, I think the current task split with assaf is ok, and I wil submit a new patch-set for API tomorrow
15:45:11 <armax> ok
15:46:02 <carl_baldwin> We’re short on time.  Anything else for HA?
15:46:26 <safchain> armax, If you have any time to work on the scheduler
15:46:29 <safchain> carl_baldwin, ok for me
15:46:42 <armax> safchain: I can have a look
15:46:46 <carl_baldwin> safchain: armax: I’ll let you coordinate offline.
15:46:53 <carl_baldwin> #topic l3-svcs-vendor-*
15:47:06 <carl_baldwin> pcm_: Some progress this week.  Anything to discuss?
15:47:17 <pcm_> Sure :)
15:47:28 <pcm_> Reference implementation is in.
15:47:43 <pcm_> Could use review of Cisco implementation (https://review.openstack.org/#/c/107473/1)
15:48:13 <pcm_> Since the Cisco one was dependent on the reference one, do I need to rebase the commit to master now>
15:48:14 <pcm_> ?
15:48:36 <carl_baldwin> pcm_: My view of rebasing is that it should be done whenever Jenkins fails to merge it.
15:48:45 <carl_baldwin> Others may not share that view (yet).
15:48:48 <carl_baldwin> ;)
15:48:55 <carl_baldwin> pcm_: Anything else?
15:49:18 <pcm_> No. Later (next week), I'll probably want to see if there are other services that need this validation too
15:50:01 <carl_baldwin> pcm_: Thanks.
15:50:02 <pcm_> Can work with vendors on applying this method to their service.
15:50:12 <carl_baldwin> #topic bgp-dynamic-routing
15:50:16 <pcm_> thanks for the reviews!
15:50:35 <carl_baldwin> I don’t see devvesa or nextone92 on.
15:50:53 <carl_baldwin> I do know that there has been some progress on this topic.
15:50:59 <carl_baldwin> The bp merged.
15:51:05 <yamamoto> i have a short update on ryu bgp
15:51:23 <yamamoto> outfilter patch has been merged to ryu master
15:51:23 <carl_baldwin> yamahata__: Great.  You have the floor.
15:52:18 <carl_baldwin> good to know.  Do you know if devvesa has been working with your code yet?
15:52:19 <yamamoto> i think we completed features requested by devvesa
15:52:43 <yamamoto> more requests and questions are welcome
15:52:54 <carl_baldwin> I know that he has been working on implementation.  We discussed some model simplifications yesterday.  I’m not sure how much he has worked with Ryu code yet.
15:53:00 <yamamoto> i don't know about devvesa's progress
15:53:13 <carl_baldwin> I’ll ping him later today for progress.
15:53:29 <carl_baldwin> Anything more to add?
15:53:35 <yamamoto> nothing from me
15:54:44 <carl_baldwin> #topic neutron-ipam
15:55:29 <carl_baldwin> SAD to say that IPAM won’t make the cut for Juno.
15:56:21 <carl_baldwin> This doesn’t mean that work has to stop by those who are interested.  I’d like to get an early start on this for K
15:56:44 <carl_baldwin> Just means that review time will be diverted to other things.
15:57:19 <carl_baldwin> I don’t see any of the ipam guys around.  So, that is all for now.
15:57:24 <carl_baldwin> #topic Open Discussion
15:59:11 <carl_baldwin> Thanks everyone for the hard work that has been done.  I’ve seen good progress being made.
15:59:22 <pcm_> bye
15:59:24 <yamamoto> thank you
15:59:26 <SridarK> Thanks DVR folks for all the help on the FWaaS - DVR issues and special thanks to Swami for helping us understand DVR and for those fantastic pictures (although i can't convince him to do it in ASCII)
15:59:27 <carl_baldwin> That’s all I have.  Bye.
15:59:29 <SridarK> bye
15:59:51 <carl_baldwin> #endmeeting