15:02:16 <carl_baldwin> #startmeeting neutron_routed_networks
15:02:17 <openstack> Meeting started Tue Sep 13 15:02:16 2016 UTC and is due to finish in 60 minutes.  The chair is carl_baldwin. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:02:18 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:02:20 <openstack> The meeting name has been set to 'neutron_routed_networks'
15:02:50 <carl_baldwin> #topic Announcemestn
15:02:54 <carl_baldwin> #undo
15:02:54 <openstack> Removing item from minutes: <ircmeeting.items.Topic object at 0x7f23bd8f9210>
15:02:58 <carl_baldwin> #topic Announcements
15:03:02 <carl_baldwin> That's better
15:03:09 <carl_baldwin> RC1 will be today, I think.
15:03:38 <carl_baldwin> If I recall, RC1 is usually when the branch is cut and the master branch opens for the next cycle.
15:04:35 <carl_baldwin> It'd be cool to start getting some things lined up soon. But, I know there are some things still to wrap up  (e.g. tests and docs)
15:04:57 <carl_baldwin> Summit is approaching fast.
15:05:22 <carl_baldwin> Any other announcements?
15:05:29 <mlavalle> PTL nomination ends 2016-09-18, 23:45 UTC
15:06:42 <carl_baldwin> That's this Sunday.
15:06:51 <carl_baldwin> #topic patches
15:06:51 <haleyb> doesn't look like any hats are in the ring yet
15:07:05 <carl_baldwin> #link haleyb nope
15:07:15 <carl_baldwin> Ah
15:07:24 <carl_baldwin> #undo
15:07:24 <openstack> Removing item from minutes: <ircmeeting.items.Link object at 0x7f23bd769410>
15:07:43 <carl_baldwin> #link https://review.openstack.org/#/q/status:open+(topic:bp/routed-networks+OR+topic:bp/neutron-routed-networks+OR+topic:bp/service-subnets)
15:08:11 <carl_baldwin> I think we got everything that I wanted merged for Newton done.
15:08:22 <mlavalle> ++
15:08:23 <carl_baldwin> That's pretty cool.
15:08:35 <carl_baldwin> Does anyone have patches that need some attention?
15:09:24 <carl_baldwin> mlavalle: How are the testing patches coming along?
15:09:36 <mlavalle> we are making good progress
15:09:53 <carl_baldwin> I see that amuller has given them some attention. That is great, thanks amuller
15:10:15 <mlavalle> the scenario test is running reliably now in our vagrant environment
15:10:35 <mlavalle> amuller pointed correcty to the next challenge
15:10:45 <mlavalle> which is to make it runable more generallly
15:10:49 <mlavalle> in any environment
15:11:01 <mlavalle> we will address that this week
15:11:07 <carl_baldwin> Makes sense.
15:11:15 <carl_baldwin> Sounds like progress.
15:11:33 <mlavalle> I also started a conversation with clarkb about getting this running in Jenkins
15:11:51 <mlavalle> expect to address it over the next 2 weeks
15:12:17 <mlavalle> And finally the API tests from janzian are also progressing
15:12:39 <mlavalle> amuller reviewed his patchset yesterday and suggested a couple more test cases
15:12:48 <mlavalle> that should be done over the next 2 days
15:12:51 <carl_baldwin> cool
15:13:06 <mlavalle> that's all this week in that front
15:13:08 <janzian> Yes, thanks amuller for the helpful suggestions
15:14:01 <carl_baldwin> Cool. I am finishing up some improvements to the docs based on the feedback I've received. I'll post it soon.
15:14:06 <carl_baldwin> I think that's it.
15:14:17 <carl_baldwin> #topic Ocata
15:14:50 <carl_baldwin> I have three main things to do for Ocata. I put them on the etherpad around L43
15:15:00 <carl_baldwin> #link https://etherpad.openstack.org/p/routed-provider-networks-notes
15:15:14 <carl_baldwin> mlavalle is all over the first one.
15:15:25 <carl_baldwin> "Integration with Nova Scheduler"
15:15:41 <mlavalle> We are making good progress there
15:15:53 <mlavalle> Actually, we are ahead of the Nova guys
15:16:20 <mlavalle> As of this past Sunday, I have the entire subnet lifcyle debugged and running
15:16:28 * carl_baldwin expects nothing less. :D
15:17:13 <mlavalle> Meaning that we can create / update / delete subnets in routed networks and the corresponding resource pools, agregates and inventories are created / updated and eleted
15:17:14 <carl_baldwin> mlavalle: Do you need any help in that area? Are you ready for reviews yet? Or, would that still be premature.
15:17:59 <mlavalle> not yet
15:18:25 <mlavalle> the Nova team is still making some changes to the inventories api
15:18:45 <mlavalle> and they need to get the association of aggregates with resource pools working
15:19:02 <carl_baldwin> okay
15:19:05 <mlavalle> But that is not stopping our testing
15:19:15 <mlavalle> I am working around those bits
15:19:15 <carl_baldwin> Sounds like this is in good hands.
15:19:35 <mlavalle> I think I will be done testing the functionality next week
15:19:46 <mlavalle> and then I will write unit tests
15:19:46 <carl_baldwin> mlavalle: cool
15:19:57 <mlavalle> and also a tempest scenario test
15:20:24 <mlavalle> Good progress overall
15:20:25 <carl_baldwin> mlavalle: You might consider writing tests along with writing the functionality. I find that helps me focus my efforts when I write the functionality.
15:20:49 <mlavalle> will do. Thanks
15:21:10 <carl_baldwin> The next thing is "Using a routed network as a router:external network"
15:21:18 <carl_baldwin> There are two issues with this right now.
15:21:35 <carl_baldwin> 1. We can't put floating IPs on a routed network.
15:21:53 <carl_baldwin> 2. We need routing for the floating IPs.
15:22:41 <carl_baldwin> For 1, I think we should be close. My plan is to use service subnets for floating ips.
15:23:12 <mlavalle> ahhh
15:23:13 <carl_baldwin> In other words, if you put a floating IP subnet on a routed network, it *must* have a service subnet of whatever the device owner for a floating IP is.
15:25:04 <carl_baldwin> Then, it should just be a matter of figuring out how to deploy BGP routing. I'm not sure if the BGP routing will need any tweaks.
15:25:32 <carl_baldwin> mlavalle: I'm now thinking it'd be very cool to install quagga on the router in your vagrant environment and play with this.
15:25:49 <mlavalle> carl_baldwin: will start looking at it
15:26:01 <carl_baldwin> mlavalle: I'm not asking you to do it. I hope to dive in to it myself. I've been meaning to brush up on my BGP skills.
15:26:14 <mlavalle> cool
15:26:19 <carl_baldwin> mlavalle: I'm also not trying to discourage you from being involved if you want.
15:26:42 <mlavalle> carl_baldwin: I want to be involved
15:26:52 <carl_baldwin> mlavalle: cool
15:27:12 <mlavalle> I also want to expand my skills ;-)
15:28:53 <carl_baldwin> Are there any questions about this one?
15:29:56 <carl_baldwin> This is the overall plan. I wonder if we'll need a new blueprint for this or if we can continue working under the existing blueprint.
15:30:17 <carl_baldwin> I'm sure a few more details will pop up as we go. But, I think the overall plan is pretty sound.
15:30:34 <mlavalle> was ther a 3rd point?
15:30:45 <carl_baldwin> Yes
15:31:04 <carl_baldwin> The third one builds from the 2nd.
15:31:06 <carl_baldwin> "Floating IPs with no router on routed networks"
15:31:33 <carl_baldwin> GoDaddy is doing this already in their deployment.
15:31:38 <carl_baldwin> I think it is a pretty cool idea.
15:31:58 <mlavalle> "no router" meaning no Neutron router?
15:31:58 <carl_baldwin> This one also has a few steps.
15:32:07 <carl_baldwin> mlavalle: Right. I should clarify that.
15:32:20 <carl_baldwin> mlavalle: good point.
15:33:07 <carl_baldwin> 1. We need to be able to associate floating IPs with ports on the same routed network. Currently, the floating ip association will look for a router to handle the fip.
15:35:25 <carl_baldwin> Sorry, got pulled off for a minute.
15:35:58 <carl_baldwin> 2. The VM port has to have some way to accept the traffic for the floating IP.
15:36:03 <carl_baldwin> ^ This one could be difficult.
15:36:34 <carl_baldwin> At GoDaddy, the VM actually knows it has the floating IP and accepts traffic destined to it.
15:36:49 <carl_baldwin> I think it will be hard to generalize this.
15:36:51 <carl_baldwin> Thoughts?
15:37:14 <carl_baldwin> Another possibility that I thought of is doing NAT at the port.
15:37:25 <carl_baldwin> I think this could use some broader discussion.
15:38:25 <mlavalle> where would we place the iptables for that?
15:39:15 <carl_baldwin> mlavalle: Around the same place as the security groups, I guess. It is a good question. One that I haven't thought through yet.
15:39:34 <mlavalle> yeah, security gorups comes to mind
15:39:53 <mlavalle> how about doing NAT with flows in the switch?
15:39:58 <haleyb> it should work there i think, just in a different table/chain
15:41:54 <carl_baldwin> I think this topic deserves a new RFE / spec. It doesn't really fall in the scope of the original routed networks spec.
15:42:23 <haleyb> no, sounds like a design change
15:42:42 * haleyb will just sit here and state the obvious :)
15:42:45 <carl_baldwin> haleyb: Or, at least more design that wasn't included in the original.
15:43:11 <carl_baldwin> haleyb: It is good to get feedback, even if it is just nodding along. :)
15:43:21 * haleyb nods
15:43:26 <mlavalle> LOL
15:43:55 <carl_baldwin> I'll file an RFE with a brief description. We'll see if a spec is necessary.
15:45:23 <carl_baldwin> Is there anything else we should be doing with routed networks during Ocata?
15:45:46 <mlavalle> I think that's plenty
15:46:04 <mlavalle> Let's keep in mind that Ocata is the cycle where we rally make routed networks operational
15:46:22 <mlavalle> so we should not forget to focus on that
15:46:40 <carl_baldwin> mlavalle: ++ The most important piece is the one that you're working on.
15:47:25 <carl_baldwin> So, I think we have a few action items.
15:47:42 <carl_baldwin> 1. mlavalle continues doing what he's doing
15:48:09 <carl_baldwin> 2. Get quagga installed on the vagrant router and deploy BGP.
15:48:21 <carl_baldwin> 3. Write spec for floating IPs without neutron routers.
15:48:33 <carl_baldwin> These should get us started.
15:48:42 <carl_baldwin> s/spec/rfe/
15:48:42 <mlavalle> Cool
15:49:41 <carl_baldwin> I guess that's the plan then. Please let me know if you'd like to pitch in.
15:49:49 <carl_baldwin> ... and where
15:51:25 <carl_baldwin> I guess that's all I have for today. Anything to add?
15:51:32 <carl_baldwin> Going once...
15:51:38 <mlavalle> Good meeting!
15:52:02 <carl_baldwin> mlavalle: ++
15:52:05 <carl_baldwin> Going twice...
15:52:21 <carl_baldwin> #endmeeting