14:02:20 <carl_baldwin> #startmeeting neutron_routed_networks
14:02:22 <openstack> Meeting started Tue Apr 19 14:02:20 2016 UTC and is due to finish in 60 minutes.  The chair is carl_baldwin. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:02:23 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:02:25 <openstack> The meeting name has been set to 'neutron_routed_networks'
14:02:28 <carl_baldwin> #topic Announcements
14:02:36 <carl_baldwin> #link https://wiki.openstack.org/wiki/Meetings/Neutron-Routed-Networks
14:02:54 <carl_baldwin> The summit is next week!
14:03:02 <mlavalle> yaay!
14:03:06 <carl_baldwin> We will not be having this meeting next week.
14:03:21 <xiaohhui> do we have discussion at summit about routed network?
14:03:26 <carl_baldwin> Should we reconvene the week after, or the week after that?
14:03:38 <mlavalle> xiaohhui: we have a discussion with Nova
14:03:50 <carl_baldwin> xiaohhui: what mlavalle said ^
14:04:04 <xiaohhui> ok, thanks
14:04:07 <carl_baldwin> mlavalle: That's on Wednesday, right?  At 11:00?
14:04:08 <mlavalle> carl_baldwin: right after the summit.... we have a lot to do
14:04:19 <mlavalle> carl_baldwin: correct, Hilton hotel
14:05:06 <mlavalle> carl_baldwin: 11:00 to 11:40
14:05:42 <carl_baldwin> I'm looking for the schedule link but things are loading very slowly for me today.
14:06:02 <mlavalle> carl_baldwin: for Nova...
14:06:09 <mlavalle> #link https://www.openstack.org/summit/austin-2016/summit-schedule/global-search?t=Nova%3A+
14:06:23 <carl_baldwin> #link https://www.openstack.org/summit/austin-2016/summit-schedule/events/9089
14:06:27 <carl_baldwin> There it is ^
14:06:57 <mlavalle> and specific for Neutron:
14:07:04 <mlavalle> #link https://www.openstack.org/summit/austin-2016/summit-schedule/global-search?t=Neutron%3A+
14:07:06 <xiaohhui> thanks, guys
14:07:09 <carl_baldwin> It will mostly discuss the Nova scheduling aspects.
14:07:59 <carl_baldwin> I think we should plan to meet for smaller discussions.  I will probably be free on Tuesday but I haven't finalized my schedule.
14:08:19 <carl_baldwin> Any other announcements?
14:08:59 <carl_baldwin> Okay, moving on...
14:09:03 <carl_baldwin> #topic Progress
14:09:21 <carl_baldwin> #topic Basic CRUD
14:10:13 <carl_baldwin> After playing around with this a little last week, I found that I don't understand what armax wants.  So, I talked to him yesterday evening.  I can't say that I fully understand yet except that I shouldn't use a mixin.
14:10:35 <mlavalle> that we all know... lol
14:10:47 <carl_baldwin> Anyway, this will be a priority for me this week to figure it out so that we can keep things progressing.
14:10:48 <mlavalle> stay away from mixins
14:11:42 <carl_baldwin> Right, but I'm struggling to find any other pattern in the code that is really relevant to this case.  So, it is going to be some trailblazing, I think.
14:12:11 <xiaohhui> Maybe we should associate segment with network by updating the network's segments
14:12:58 <xiaohhui> Just like the security-group for a port
14:13:07 <carl_baldwin> xiaohhui: I actually think I have a better handle on the segment - subnet association.  It is adding segments in the first place that I'm struggling with.
14:13:31 <carl_baldwin> The extension manager can be used for the segment - subnet association, I think.
14:14:14 <xiaohhui> I agree. That will fit the spec
14:14:27 <xiaohhui> But the change is big
14:14:41 <xiaohhui> based on the current patches
14:14:42 <carl_baldwin> I'll get this figured out this week before the summit.
14:15:19 <xiaohhui> Sounds great
14:15:51 <carl_baldwin> I'm going to skip the segment - subnet association topic for now since I won't be able to work on it this week.
14:16:07 <carl_baldwin> #topic Nova Scheduler
14:16:25 <carl_baldwin> mlavalle: hi
14:16:29 <mlavalle> hi
14:16:46 <carl_baldwin> Apparently there has been some discussion on this in Nova and they don't like something.
14:17:12 <carl_baldwin> I haven't seen any feedback on the spec yet so I really don't know what they're thinking yet.
14:17:39 <mlavalle> I am going to ping jaypipes today to make sure we have another conversation this week
14:18:00 <carl_baldwin> I'm inclined to take kevinbenton 's feedback and split out the deferred IP address stuff and work on that separately.
14:18:11 <reedip_> the meeting started late? or early?
14:18:51 <mlavalle> I suggest listen to jaypipes feedback this week. He might not have been aware of the last revision to the spec
14:18:55 <carl_baldwin> reedip_: I did start a minute late.
14:19:07 <mlavalle> so I am going to make sure he is aware of that
14:20:02 <carl_baldwin> The only thing that sticks out to me from Dan's feedback is that he doesn't want Nova's allocation to be conditional on an IP already being allocated in Neutron for the port.
14:20:03 <mlavalle> in any case, IMO, it is best if we smooth out any disagreements this week, so we can reach agreements next week
14:20:33 <carl_baldwin> Which is what Jay has been saying from the beginning.
14:21:02 <carl_baldwin> mlavalle: I agree.
14:22:05 <carl_baldwin> We also need to make progress on the other Nova part.
14:22:13 <mlavalle> so I say let's focus this week on that high bandwidth conversation
14:22:29 <carl_baldwin> That is, allowing deferred IP address on a port.
14:22:42 <mlavalle> carl_baldwin: agree
14:23:02 <carl_baldwin> That discussion hasn't gotten any discussion lately as far as I know.
14:23:25 <mlavalle> yeap
14:23:45 <carl_baldwin> mlavalle: Anything else?
14:23:53 <mlavalle> that's it this week
14:24:26 <carl_baldwin> mlavalle: Thanks!
14:24:35 <carl_baldwin> #topic Host / Segment Mapping
14:24:47 <mlavalle> hi again
14:25:06 <mlavalle> so this week I pushed another revision to the patchset
14:25:39 <mlavalle> #link https://review.openstack.org/#/c/285548/
14:26:14 <carl_baldwin> I actually took a peek at it yesterday with armax.
14:26:22 <mlavalle> I followed sugestion from xiaohhui to pass the bridge mapping to the function
14:26:50 <mlavalle> any feedback?
14:27:09 <carl_baldwin> I thought armax was going to add feedback.  But I can pass it along on the review.
14:27:09 <xiaohhui> why not just pass the phynet to the function?
14:27:49 <mlavalle> he hasn't added any yet
14:28:56 <mlavalle> xiaohhui: I get the idea.... Make it work for non agent based drivers. I'll just move to it gradually, becuase I am working on another piece with the SR-IOV driver
14:29:22 <xiaohhui> We plan to extact hostname/physical_net in networking-ovn, but that can be changed.
14:29:24 <carl_baldwin> mlavalle: I just added comments.
14:29:47 <carl_baldwin> I think my comment might be along the same lines as xiaohhui 's if I understand correctly.
14:29:47 <mlavalle> carl_baldwin: I saw that, thanks!
14:30:36 <carl_baldwin> I understand that it is a WIP and you may have had the intention to evolve it in this way already.  But, since we had the feedback, I offer it to you.
14:30:37 <carl_baldwin> :)
14:30:52 <mlavalle> perfectly welcome. Thanks!
14:31:03 <carl_baldwin> mlavalle: Anything else on this?
14:31:20 <mlavalle> This week I intend to work on the changes to the SZR-IOV mechanism driver
14:31:28 <mlavalle> SR-IOV^^^^
14:32:14 <mlavalle> The idea is for it to conform to the same interface as the other in-tree drives, so they all have check_segment_for_agent
14:32:15 <carl_baldwin> Need anything?
14:32:34 <mlavalle> that way we can use this method to check connectivity
14:32:39 <mlavalle> that's it this week
14:32:46 <carl_baldwin> Thanks.
14:33:04 <carl_baldwin> #topic Create / delete segment in ML2
14:33:13 <xiaohhui> yeah, no code change last week, but some investigation.
14:33:16 <carl_baldwin> xiaohhui: I think I've seen some activity on this.
14:33:42 <carl_baldwin> Maybe it was in the form of feedback on other patchs.
14:33:51 <carl_baldwin> xiaohhui: Do you need anything?
14:34:03 <xiaohhui> The currect patch only works for DB, I found there are gaps in ml2.
14:34:09 <xiaohhui> The segment now is created/deleted with network in ml2.
14:34:31 <xiaohhui> The gap is to enable updating the segment of network in l2-agent. There were no such code for now. I think it would be better to code after the CRUD of segment is settled.
14:34:49 <carl_baldwin> xiaohhui: That is probably wise.
14:35:09 <carl_baldwin> I don't think this is one of the more urgent pieces.
14:35:09 <xiaohhui> I think I need to set WIP to the patch, if it is not done yet
14:36:09 <xiaohhui> yeah, But since we now have the chance to update the segment of a network/subnet, we might need to know how to handle it. I will still look into this one.
14:36:14 <carl_baldwin> I wonder if it would be good to continue trying to get OVN on par with create / delete segments with the network and leave this for a while.
14:36:35 <carl_baldwin> xiaohhui: okay
14:37:34 <carl_baldwin> #topic Segment aware DHCP
14:37:56 <carl_baldwin> blogan isn't here today.  I think we can skip this.
14:38:04 <carl_baldwin> #topic Client
14:38:34 <carl_baldwin> rtheis: reedip_ : hi
14:38:39 <rtheis> hi
14:38:45 <reedip_> yup
14:38:49 <carl_baldwin> I updated the API based on your feedback.
14:39:03 <rtheis> carl_baldwin: thank you
14:39:17 <rtheis> I've update the OSC patches accordingly.
14:39:33 <carl_baldwin> rtheis: reedip_ : thanks for the feedback.
14:39:34 <reedip_> I am waiting for the final API conclusion to update the NeutronClient patch
14:39:35 <rtheis> carl_baldwin: and thank you for reviewing the devref in neturonclient
14:40:09 <rtheis> I am also waiting the same.
14:40:33 <carl_baldwin> rtheis: Do you have links handy?
14:40:37 <rtheis> I'm also working with OSC team on devref for beta commands: https://review.openstack.org/#/c/305436/
14:41:07 <rtheis> OSC patches are: https://review.openstack.org/#/c/290138/ and https://review.openstack.org/#/c/302395/
14:41:30 <rtheis> SDK patches are: https://review.openstack.org/#/c/290054/ and https://review.openstack.org/#/c/302223/
14:42:14 <carl_baldwin> rtheis: Thanks.
14:42:17 <rtheis> yw
14:42:26 <carl_baldwin> Anything else?
14:42:31 <rtheis> nothing else
14:42:47 <reedip_> https://review.openstack.org/#/c/293305/13 was updated
14:42:55 <carl_baldwin> reedip_: Thanks.
14:42:58 <reedip_> But boden had some comments which I can take care of
14:43:24 <carl_baldwin> reedip_: I'll take a look at it
14:43:30 <reedip_> however the rackspace maintenance(?) has slowed the gate down a bit , so , waiting for the callback from gate :)
14:43:49 <carl_baldwin> #topic L2 adjacency
14:44:20 <reedip_> I havent worked on this patch from a WIP I published earlier , and had some doubts
14:44:35 <reedip_> carl_baldwin: I will email you the doubts ( I have them on my office PC)
14:44:39 <reedip_> is that ok ?
14:44:47 <carl_baldwin> reedip_: Yes, that is okay.
14:45:01 <reedip_> would do them tomorrow morning ( twelve hours from now maybe )
14:45:14 <carl_baldwin> reedip_: That will be fine.
14:45:16 <mlavalle> reedip_: you attening the Summit next week?
14:45:23 <reedip_> okay, so that would help me a bit in working on the L2 adjacency
14:45:42 <reedip_> my Visa is stuck ... I mean I dont know what its status is
14:45:47 <reedip_> so I am not sure
14:46:02 <reedip_> if they have approved it or not, and hence my attendance :|
14:46:11 <mlavalle> good luck with it!
14:46:21 <reedip_> i need it :) thanks mlavalle :)
14:46:32 <carl_baldwin> reedip_: good luck, hope to see you.
14:46:36 <carl_baldwin> #topic OVN and routed networks
14:46:40 <carl_baldwin> xiaohhui: Do you have anything?
14:46:55 <xiaohhui> We are working on reading hostname and physical _net from ovn. Next step depends on the patch segment/host mapping.
14:47:06 <xiaohhui> That is all for the ovn in routed network
14:47:20 <carl_baldwin> xiaohhui: Thanks.
14:47:22 <mlavalle> xiaohhui: you coming to Summit?
14:47:36 <xiaohhui> yeah
14:47:47 <rtheis> xiaohhui: is this networking-ovn patch a WIP or good to go? https://review.openstack.org/#/c/302623/
14:47:47 <mlavalle> xiaohhui: great, let's talk there :-)
14:47:52 <xiaohhui> my visa got approval lat friday
14:48:18 <rtheis> If ready to go, I'll try to review this week
14:48:25 <xiaohhui> I think it is good to go.
14:48:31 <rtheis> thank you
14:48:46 <xiaohhui> But it is not that urgent.
14:48:53 <rtheis> ok
14:50:01 <carl_baldwin> #topic Open Agenda
14:50:19 <carl_baldwin> I had a discussion about BGP and routed networks with tidwellr last week.
14:51:03 <carl_baldwin> BGP will need to be able to associate speakers with segments.
14:52:06 <carl_baldwin> I also had a great discussion with kris and mike from GoDaddy last week.  From our discussion, it sounds like what we're doing is going to work for them.
14:52:32 <mlavalle> that is actually great feedback. Thanks for sharing
14:52:50 <carl_baldwin> They'll have a pretty crazy migration from their current solution but it will happen.
14:53:29 <mlavalle> should we try to get some early feedback from other operators during Summit?
14:53:29 <carl_baldwin> They also might be able to help with enabling floating IPs on a routed network (with no Neutron router).
14:53:44 <carl_baldwin> mlavalle: yes, good seque
14:54:04 <carl_baldwin> They are having a large deployers team session right after the keynotes on Monday.
14:54:12 <carl_baldwin> I am planning to attend and do just that.
14:54:40 <mlavalle> I'll attend with you. Would you share time and location?
14:55:03 * carl_baldwin looking...
14:55:25 <carl_baldwin> https://www.openstack.org/summit/austin-2016/summit-schedule/events/9515
14:55:26 <mlavalle> carl_baldwin: don't worry, I'll just stick to you :-)
14:55:39 <carl_baldwin> #link https://www.openstack.org/summit/austin-2016/summit-schedule/events/9515
14:56:21 <mlavalle> Thanks!
14:56:37 <carl_baldwin> Anything else?
14:56:47 <carl_baldwin> once...
14:56:58 <carl_baldwin> twice...
14:57:05 <carl_baldwin> Okay, thanks everyone!
14:57:11 <carl_baldwin> #endmeeting