15:01:28 <carl_baldwin> #startmeeting neutron_routed_networks
15:01:28 <openstack> Meeting started Tue Jul  5 15:01:28 2016 UTC and is due to finish in 60 minutes.  The chair is carl_baldwin. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:01:29 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:01:31 <openstack> The meeting name has been set to 'neutron_routed_networks'
15:01:35 <carl_baldwin> Anyone around today?
15:01:40 <xiaohhui> hello
15:01:43 <john-davidge> hi
15:01:47 <johndperkins> yo
15:01:57 <mlavalle> o/
15:02:21 <carl_baldwin> Hi all.
15:02:45 <carl_baldwin> #topic Announcements
15:02:55 <carl_baldwin> Etherpad is here:
15:02:58 <carl_baldwin> #link https://etherpad.openstack.org/p/routed-provider-networks-notes
15:03:53 <carl_baldwin> Newton-2 is coming right up.  It'd be very good to get a couple of things merged by then.  Namely DHCP, deferred IP ports in Nova, and service subnets.
15:04:15 <carl_baldwin> But, there are lots of smaller improvements that could go too.
15:04:22 <carl_baldwin> Any other announcements?
15:04:43 * carl_baldwin feels like we were just here. He had a 4 day week-end away from work.
15:04:48 <mlavalle> Remind people of the Neutron mid-cycle, August 17 - 19, Cork Ireland
15:05:13 <carl_baldwin> mlavalle: Thanks.  I've booked my travel.  I'll be there the.
15:06:00 <carl_baldwin> s/ the//
15:06:05 <carl_baldwin> Anything else?
15:06:29 <mlavalle> Talks submission deadline July 13th
15:06:33 <mlavalle> for the Summit
15:07:07 <carl_baldwin> Thanks again.
15:08:03 <carl_baldwin> Moving on...
15:08:24 <carl_baldwin> We've got lots of miscellaneous patches on the etherpad.  Be sure to check there when you can.
15:08:48 <carl_baldwin> I've been trying to make sure that all of the patches are listed on there with their status.
15:09:04 <carl_baldwin> If you have a patch that isn't listed, please list it in the appropriate section.
15:10:23 <carl_baldwin> We're still merging at a decent rate.  So, just double check the status of your patches.
15:11:21 <carl_baldwin> #topic DHCP
15:11:33 <carl_baldwin> blogan said he won't be able to make it today.
15:11:46 * neiljerram slips in late
15:12:00 <carl_baldwin> I'm going to try to pick up where I left off last week and get DHCP working in mlavalle 's fantastic vagrant environment.
15:12:19 * mlavalle blushes
15:13:14 <mlavalle> I will also keep an eye on blogan'a next patchset. Janzian and I will also be testing it
15:13:26 <carl_baldwin> I think we're close.  I think blogan has some work to fix the problem of the agent requesting subnets on other segments and then we've got an RPC thing to fix.
15:13:31 <carl_baldwin> mlavalle: great.
15:14:07 <carl_baldwin> I'm going to try hard to get this fixed by Newton-2
15:14:13 <carl_baldwin> Anything else on DHCP?
15:15:15 <carl_baldwin> #topic Deferred IP allocation
15:15:56 <carl_baldwin> I've got to close on this one
15:15:58 <carl_baldwin> #link https://review.openstack.org/#/c/299591
15:16:34 <mlavalle> I took a look yesterday and made a comment
15:17:10 <carl_baldwin> mlavalle: Thanks for your comment.  I was just replying.
15:17:24 <carl_baldwin> I don't think it should matter if the extension is enabled the way I wrote it.
15:17:38 <carl_baldwin> It will return None which will be interpreted as not deferred.
15:17:47 <mlavalle> Just wanted to make sure it was considered
15:17:55 <carl_baldwin> mlavalle: Thanks.
15:18:19 <carl_baldwin> I still need to digest the rest of the conversation.  I might need to ping Matt today.
15:18:57 <mlavalle> I will also try to comment on the broader conversation, now that I have a dded a patchset to that seires
15:19:07 <mlavalle> series^^^
15:19:56 <carl_baldwin> mlavalle: What do you mean by "added a patchset ..."?
15:20:50 <mlavalle> I added a new patchset for the interaction with routed networks from nova. We can comment in the NOva update
15:20:57 <mlavalle> doesn't interfere with yours
15:21:08 <carl_baldwin> mlavalle: ok
15:22:13 <carl_baldwin> #topic Service Subnets
15:22:30 <carl_baldwin> john-davidge: haleyb:  Hi, anything on this?
15:23:22 <haleyb> carl_baldwin: i am trying to get the WIP out so john can iterate on it, too much FTO for me last week :(
15:23:36 <haleyb> it's my priority today
15:23:57 <john-davidge> haleyb: Thanks, please do everything you can to get it uploaded today, I'm ready to work on it fulltime
15:24:29 <carl_baldwin> I'll be happy to take a look at it too.
15:24:35 <carl_baldwin> ... when it is posted.
15:25:55 <carl_baldwin> Anything else?
15:26:15 <haleyb> not from me
15:26:52 <carl_baldwin> It is probably a stretch at this point to get this merged by Newton-2.  That is fine but it'd be nice to close on it soon after.  I'm willing to help anytime you need another perspective.
15:27:25 <carl_baldwin> #topic Integration with Nova Scheduler
15:27:35 <mlavalle> First of all, many thanks to johnthetubaguy and the Nova core team for merging the patchsets in the prep for network aware scheduling series up to here:
15:27:54 <mlavalle> #link https://review.openstack.org/#/c/329851/
15:28:41 <mlavalle> Following johnthetubaguy advise from last week's meeting, I added this patchset on top of it:
15:28:59 <mlavalle> #link https://review.openstack.org/#/c/337387/
15:29:34 <mlavalle> This patchset adds the ineraction from allocate_for_instance in Nova with the Routed Networks API
15:30:28 <mlavalle> I made some assumptions as to the precise features in the Neutron side, because some of that is not ready yet
15:31:05 <carl_baldwin> mlavalle: Do you think there is any chance to get this reviewed?
15:31:43 <carl_baldwin> I have to admit I don't understand this patch's purpose yet.  I need to look at it.
15:32:05 <mlavalle> carl_baldwin: Well, it was John's advise. So I am assuming they will review it. I will certainly give it visibility on Monday's during the scheduler meeting
15:33:15 <carl_baldwin> I'll take a look at it.
15:33:45 <carl_baldwin> mlavalle: Do you know what the time table for the scheduler changes is?
15:33:54 <carl_baldwin> mlavalle: Still targeted for Newton?
15:34:10 <mlavalle> carl_baldwin: it is still targeted for Newton
15:34:38 <mlavalle> Last night I started working in the Vagrant environment adding 2 aggregates
15:35:12 <mlavalle> On top of that I want to deploy this week the g-r-p series of patchsets
15:35:20 <carl_baldwin> Sounds like good progress.  Keep up the good work and let me know if / when we need to help.
15:35:36 <mlavalle> the purpose is two fold
15:35:55 <mlavalle> 1) Get a sense of how ready g-r-p is ready
15:36:16 <mlavalle> 2) Start prototyping in Neutron the interaction with that API
15:37:10 <mlavalle> That's all for today
15:37:26 <carl_baldwin> mlavalle: Good plan.
15:37:36 <reedip> hi
15:37:43 <carl_baldwin> #topic Open Discussion
15:37:49 <carl_baldwin> What have we left out?
15:38:17 <reedip> carl_baldwin, john_davige : wanted to know if l2_adjacency would be false if we have a single segment
15:39:02 <mlavalle> tah is when l2_adjacency is true, as far as I understand it
15:39:11 <carl_baldwin> reedip: I wouldn't count segments.  I think if subnets are associated with segments then it should be false.
15:39:34 <reedip> hmm ... okay
15:39:50 <carl_baldwin> A multi-segmented network might start out as a single segment but with the intention that it become multi-segmented soon.
15:40:06 <mlavalle> That makes sense
15:40:16 <reedip> john-davidge had this comment, so wanted to know for sure
15:40:46 <mlavalle> The intention being signaled be associating subnets to the segment
15:40:51 <carl_baldwin> I don't want the value of l2adjacency to suddenly change when a new segment is added.
15:41:12 <john-davidge> carl_baldwin: Perhaps I misunderstood the spec, but that's how I read it
15:41:41 <carl_baldwin> mlavalle: ++
15:42:01 <john-davidge> https://specs.openstack.org/openstack/neutron-specs/specs/newton/routed-networks.html#l2-adjacency
15:42:04 <reedip> john-davidge : thanks for the comment though, made things clear :)
15:42:25 <john-davidge> according to the above ^^^ l2 adjacency should be true only if there is a single segment
15:42:41 <carl_baldwin> john-davidge: Unfortunately, we didn't get some updates in to the spec.
15:42:42 <carl_baldwin> :(
15:43:16 <carl_baldwin> That is my mistake.  I meant to get that updated but did not.
15:43:21 <john-davidge> carl_baldwin: Ok, so the new expectation is that l2_adjeacency will be true only if the netowkr has zero segments?
15:43:26 <mlavalle> john-davidge: yeah, that is how I read it a few months ago. But I think that we should expect this evolution in thinking happen, as we develop the functionality
15:43:56 <reedip> mlavalle, john-davidge, carl_baldwin : can you review and let me know how to move forward with https://review.openstack.org/#/c/304647/ , whenever you have time?
15:44:00 <mlavalle> carl_baldwin, john-davidge: wrestling with the code always enhance our understanding
15:44:00 <carl_baldwin> john-davidge: Not really.  It will be true only if there are no subnets associated with segments.
15:44:42 <john-davidge> carl_baldwin: Ah, ok
15:45:25 <carl_baldwin> john-davidge: Sorry about that confusion and thanks for your diligence.  This one really is on me.  I forgot that was not properly changed in the spec.
15:45:48 <mlavalle> Yeah, in both cases the question is what subnets are associated to
15:45:54 <john-davidge> carl_baldwin: No worries, thanks for the clarification. I guess I missed the discussion where that chnaged
15:45:55 <carl_baldwin> I'll post an update to the spec but those aren't always given much priority.  But, at least we'll have an update to reference.
15:46:58 <carl_baldwin> #action carl_baldwin will post update to clarify l2adjacency in the spec.
15:48:08 <carl_baldwin> reedip: I will review the update.
15:48:27 <carl_baldwin> reedip: Thanks for your work there.
15:50:22 <carl_baldwin> Anything else?
15:50:51 <mlavalle> Late last week janzian, yb and I had a conversation about adding tempest tests for routed networks
15:51:25 <mlavalle> janzian is going to follow up soon adding segment api support to the neutron driver in tempest
15:51:36 <mlavalle> and the api tests
15:51:49 <mlavalle> yb will be working on a scenario test
15:52:40 <carl_baldwin> mlavalle: It will be great to have well tested code in Newton.  I hope we can set a good example.
15:52:59 <janzian> Yes, please tear my code apart when I get it up for review :)
15:53:07 <carl_baldwin> Speaking of good examples, we also need good documentation.  I started writing some but haven't completed it.
15:53:32 <carl_baldwin> janzian: :)  I hope that's not what I'm doing when I review.  But, I like to be thorough.
15:54:12 <carl_baldwin> I will recommit to creating some documentation -- at least a complete draft -- by the end of Newton-3.
15:55:47 <carl_baldwin> Anything else?
15:56:00 <carl_baldwin> Going once...
15:56:31 <carl_baldwin> ... twice ...
15:57:02 <carl_baldwin> #endmeeting