15:00:08 <carl_baldwin> #startmeeting neutron_l3
15:00:10 <openstack> Meeting started Thu Aug 28 15:00:08 2014 UTC and is due to finish in 60 minutes.  The chair is carl_baldwin. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:11 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:00:13 <openstack> The meeting name has been set to 'neutron_l3'
15:00:20 <carl_baldwin> marun: amuller: devvesa: hi
15:00:33 <pcm_> hi
15:00:35 <amuller> howdy
15:00:39 <devvesa> hi
15:01:00 <carl_baldwin> Swami, mrsmith, viveknarasimhan, armax, safchain:  ping
15:01:06 <armax> carl_baldwin: pong
15:01:23 <carl_baldwin> #topic Announcements
15:01:25 <ajo> hi :)
15:01:33 <carl_baldwin> #link https://wiki.openstack.org/wiki/Meetings/Neutron-L3-Subteam
15:02:01 <mrsmith> hi
15:02:17 <carl_baldwin> Feature freeze in one week.  The gate queues are backing up and reviewer time is at a premium.
15:02:35 <carl_baldwin> Any other announcements?
15:03:06 <carl_baldwin> #topic Bugs
15:03:41 <carl_baldwin> Actually, bugs are looking pretty good overall since last week.  No new Highs (or higher).
15:04:02 <carl_baldwin> Looks like a few Highs have been closed but I’d have to look further to see which ones.
15:04:10 <carl_baldwin> Good progress.
15:04:19 <carl_baldwin> #topic neutron-ovs-dvr
15:05:02 <armax> carl_baldwin: looks like parallel tempest testing for DVR caused a bit of havoc
15:05:07 <carl_baldwin> The backlog is looking better than last week.  A few Highs have closed here as well.
15:05:31 <carl_baldwin> armax: Can you elaborate.  I did see havoc being wreaked yesterday a bit.
15:05:41 <armax> carl_baldwin: because of random failures caused by bug #1356121
15:06:10 <armax> carl_baldwin: it looks like parallel testing exharcebated the effects of that bug
15:06:17 <carl_baldwin> #link https://bugs.launchpad.net/neutron/+bug/1356121
15:06:42 <carl_baldwin> armax: So, we got lucky with the first few runs with parallel on?
15:07:24 <armax> carl_baldwin: to the best of my knowledge all dvr runs without change https://review.openstack.org/#/c/115754/ don’t behave well
15:07:59 <armax> carl_baldwin: but I cannot confirm for sure because we haven’t been very educated in kicking off the experimental job on the patches that are relevant for dvr
15:08:30 * armax scolds
15:08:38 <carl_baldwin> armax: Yes, we have missed running it on some patches.
15:09:13 <carl_baldwin> Everyone please run “check experimental” on all of your patch sets.
15:09:24 <carl_baldwin> … related to dvr at least.
15:09:28 <Rajeev> what outcome should we expect ?
15:09:51 <armax> just patches that may affect DVR yeah, that should suffice
15:09:59 <carl_baldwin> Rajeev: There are four known failures.  Two are fwaas related.
15:10:23 <Rajeev> carl_baldwin: ok, everything else should pass, right?
15:10:24 <armax> carl_baldwin: that said, with parallel tempest we have no known baseline
15:10:35 <carl_baldwin> Rajeev: the other two are test_volume_boot_pattern failures.
15:10:47 <armax> carl_baldwin: with parallel testing I have even seen one fwaas test pass
15:10:51 <armax> carl_baldwin: :)
15:11:04 <carl_baldwin> Rajeev: other transient failures have been spotted but not characterized well.
15:12:08 <Rajeev> carl_baldwin: thanks, that seems good enough to start with.
15:12:09 <carl_baldwin> armax: hopefully we can get https://review.openstack.org/#/c/115754/ through Jenkins check soon.  I will review it today.
15:12:34 <carl_baldwin> armax: the other outstanding change linked to that bug looks ready to go.
15:12:39 <armax> carl_baldwin: ok
15:12:49 <carl_baldwin> #link https://review.openstack.org/#/c/115837/
15:13:22 <carl_baldwin> armax: do you think it will help the situation?
15:13:56 <armax> the review you linked is the one review https://review.openstack.org/#/c/115754/ depends on
15:14:22 <armax> carl_baldwin: change 115754 does seem to address the DB lock timeout error
15:14:32 <armax> carl_baldwin: I ran the experimental parallel job twice
15:14:40 <carl_baldwin> armax: Ah, I see that.  So, I’ll kick it in to free the way for this one.
15:14:45 <armax> and both times I got the usual 4 failures
15:14:57 <armax> on any other patch, the experimental job seems to go nuts
15:15:05 <Rajeev> for bug #1360395 I have a review ready at https://review.openstack.org/#/c/116412/ independent of other defects
15:15:59 <carl_baldwin> armax: Okay, this is pretty important.  Let’s give https://review.openstack.org/#/c/115754/ a high priority today.
15:16:42 <armax> I think bug wise, DVR looks under control, I can work with Rajeev and the other folks to make further progress, I think for now we can move to the next agenda item
15:17:19 <carl_baldwin> Rajeev: great.  don’t be concerned if it doesn’t get attention right way.  It must be prioritized after the experimental job issues and the High bugs.
15:17:33 <carl_baldwin> One more bug:
15:17:36 <carl_baldwin> #link https://bugs.launchpad.net/neutron/+bug/1358554
15:17:40 <Rajeev> carl_baldwin: sounds good.
15:17:49 <carl_baldwin> rossella_s has posted a patch for this but I have not yet look at it.
15:18:41 <armax> carl_baldwin: it’s on my backlog, it looks like vivek had a first pass
15:18:59 <carl_baldwin> armax: good.  Thanks.
15:19:20 <carl_baldwin> otherwise, I agree with armax.  I think the backlog is under control now.
15:19:47 <carl_baldwin> #topic l3-high-availability
15:20:03 <carl_baldwin> We had a meeting on this this morning at 1300UTC.
15:20:23 <carl_baldwin> I have seen good improvement in the patches recently and I need to do another pass through the reviews.
15:20:28 <armax> carl_baldwin: that was early even for me :)
15:21:01 <amuller> Very reasonable time for us EMEA folks :)
15:21:15 <devvesa> indeed
15:21:45 <carl_baldwin> With DVR under control, this topic has hit the top of the priority list for feature freeze.
15:22:32 <carl_baldwin> In this morning’s meeting, we agreed to merge this with known bugs that make it incompatible with DVR.
15:22:52 <carl_baldwin> Bugs will be filed to address those and they will be worked after feature freeze.
15:23:06 <carl_baldwin> amuller: What do you have to add?
15:23:47 <amuller> Just waiting for reviews here... Maru added comments today to the l3 functional test patch and the keepalived patch
15:23:50 <amuller> which I'm sorting now
15:24:09 <amuller> the server side patches could use another pass (I fixed all of your comments Carl)
15:24:40 <carl_baldwin> marun: Thanks for the reviews.
15:25:13 <carl_baldwin> amuller: I did look through how you’ve addressed my comments.  I’m satisfied with how you’ve done that.
15:25:42 <carl_baldwin> I would like to run the code and I’m starting work on a two NN devstack system.
15:25:52 <amuller> cool
15:27:04 <carl_baldwin> amuller: You’ve made a lot of progress with these.  Keep up the good work.
15:27:47 <amuller> That's it from my end
15:27:54 <amuller> carl_baldwin: Thanks for the reviews
15:28:01 <amuller> If we get this merged I'm going to break in tears
15:28:10 <armax> amuller: have we checked with amotoki if patches need to be reworked after his rpc callback refactoring?
15:28:23 <amuller> armax: the first one in the series was merged and I rebased
15:28:27 <amuller> needed a small change
15:28:28 <armax> amuller: I am going to break in tears if this does not merge :)
15:28:40 <amuller> armax: ahh, 100% of tears then, good
15:29:05 <armax> amotoki: ping
15:29:11 <amuller> I hope Sylvain will have a nice surprise when he gets back from his summer vacation and he sees his baby merged :)
15:29:27 <amuller> it would be nice if Akihiro could go through the latest change in the first server side patch to make sure it fits with this plan
15:29:46 <armax> amuller: I am reaching out to him
15:30:33 <carl_baldwin> amuller: armax: +1  that will be good to get his perspective.
15:31:52 <carl_baldwin> amuller: Any more to add?
15:31:58 <amuller> nope
15:32:28 <carl_baldwin> amuller: marun: armax:  Thanks.  You’re great.
15:32:36 <carl_baldwin> #topic bgp-dynamic-routing
15:33:15 <carl_baldwin> devvesa: I want to recognize all the work that you’ve put in to this.  I’ve been really excited to see the work getting done.
15:33:36 <devvesa> thanks
15:33:42 <carl_baldwin> Unfortunately, review priorities have kept me from giving it the attention it deserves.
15:34:00 <devvesa> I understand, time is critical now for reviews
15:34:56 <devvesa> now I'm maintaining the patches (i hate rebase every-day migrations) and I got some reviews
15:35:05 <devvesa> so it goes on
15:35:28 <devvesa> also I made the wiki page you asked me for testing (i sent a mail, but I don't know if you saw it)
15:35:35 <devvesa> #link https://wiki.openstack.org/wiki/Neutron/DynamicRouting/TestingDynamicRouting
15:36:03 <carl_baldwin> devvesa: I did see that link.  Thanks for that.
15:36:17 <carl_baldwin> I think we have to brace ourselves for the reality that this work may have to wait until Kilo opens up.  What are your thoughts?
15:36:51 <devvesa> it not depends on me anymore
15:37:20 <carl_baldwin> devvesa: That is true and fully recognized.
15:37:31 <devvesa> I understand it, but is better to know it now
15:38:08 <matrohon> it seems to  be targeted for neutron-incubator too?
15:38:40 <devvesa> matrohon: yes, it was proposed as one of the extensions
15:39:05 <carl_baldwin> matrohon: it was proposed as such but that decision has not been made.
15:39:17 <matrohon> ok thanks
15:39:30 <carl_baldwin> First, neutron-incubator is not yet set up for it.  We need more discussion around the incubator repo.
15:39:36 <devvesa> i think incubator may be a good idea to add features without fit in the tight schedule of integrated release
15:40:22 <carl_baldwin> devvesa: If done well, it could be a good thing for features like this.  However, I don’t have my head totally wrapped around how it will be done.
15:40:49 <devvesa> anybody has? :)
15:41:33 <carl_baldwin> devvesa: I think that is why it is still in discussion.
15:42:23 <carl_baldwin> devvesa: Some of my work I have planned for kilo will integrate with dynamic routing.  I look forward to working on it.
15:42:59 <matrohon> same for me :)
15:43:21 <carl_baldwin> devvesa: anything more?
15:43:36 <devvesa> nothing else
15:44:27 <carl_baldwin> devvesa: Thanks.  Keep up the good work.
15:44:39 <devvesa> carl_baldwin: thanks to you, carl
15:45:08 <carl_baldwin> #topic Open Discussion
15:45:45 <matrohon> can ml3 team give me an opinion on this bug : https://bugs.launchpad.net/neutron/+bug/1361540
15:45:58 <pcm_> Can you put  https://review.openstack.org/#/c/114345/ and https://review.openstack.org/#/c/116018/ on your review radar?
15:47:14 <devvesa> second one has already +2, right?
15:47:22 <pcm_> I think they look pretty solid, IMHO.  The latter will introduce the requests-mock package which can be used on other commits
15:47:32 <pcm_> devvesa: yes, just need another +2.
15:47:55 <devvesa> oh, I thought second +2 only would be necessary in specs
15:48:09 <carl_baldwin> pcm_: I have put them on my radar.
15:48:29 <carl_baldwin> devvesa: confirm 2 +2s are necessary.
15:48:31 <pcm_> carl_baldwin: Thanks! I know you're really busy. Appreciate it!
15:49:02 <devvesa> matrohon: I'll take a look at yours, I'm more free now :)
15:49:09 <carl_baldwin> matrohon: I have put that bug on my radar as well.  I will try to get to it.
15:49:19 <matrohon> thanks
15:50:09 <marun> pcm_: I wonder why the CSR REST client isn't outside of the tree, frankly.
15:50:09 * pcm_ has to duck out...
15:50:28 <marun> pcm_: release on pypi and save yourself (and us) the headache of in-tree maintenance
15:50:40 <pcm_> marun: It was done in tree. probably should be out.
15:51:06 <marun> pcm_: not to say that it needs to happen immediately, but please keep it in mind for future.
15:51:20 <pcm_> marun: yes. will definitely do that! THanks!
15:51:26 <marun> pcm_: :)
15:51:34 <carl_baldwin> marun: pcm_:  Sounds like it is worth looking in to.  Good suggestion.
15:51:47 <devvesa> isn't a discussion about put out the tree all the vendor drivers?
15:51:52 <pcm_> I have to drop off. Thanks all!!!
15:52:01 <devvesa> please do :)
15:52:23 <marun> devvesa: there is, but having thinner plugins is a win even if we had a separate plugin repo
15:53:31 <amotoki> I am thinking to add HA field for router panel in Horizon. It is admin specific feature and once DVR support is merged, it should be really small. thought?
15:53:33 <marun> devvesa: easier maintenance when we change neutron if vendors don't mix their neutron integration with python integration
15:55:46 <devvesa> sorry, what do you mean 'mix neutron integration with python integration'?
15:56:58 <matrohon> amotoki : it seems that DVR is not mandatory for HA support?
15:57:06 <carl_baldwin> amuller: can you address amotoki ’s question?
15:57:36 <amuller> yeah
15:57:47 <amuller> amotoki: HA and 'distributed' routers are not connected
15:57:48 <carl_baldwin> matrohon: no, dvr is not mandatory.
15:58:00 <amuller> I mean you could have a router that's not HA not distributed, HA and not distributed... all 4 permutations
15:58:01 <amotoki> matrohon: not dependent on directly. i am talking about horizon support code.
15:58:09 <amuller> ahh
15:58:27 <amuller> the CLI currently shows only for admins the router attributes (distributed and HA) it would be good if Horizon did this too
15:58:58 <amotoki> amuller: yes. i am aware of both reviews in neutron and neutronclient.
15:59:26 <amotoki> it is one of important community features so I will talk with david horizon ptl.
15:59:39 <amuller> Great :)
16:00:09 <carl_baldwin> Our time is up.  Thanks all.
16:00:11 <carl_baldwin> #endmeeting