15:00:49 <carl_baldwin> #startmeeting neutron_routed_networks
15:00:50 <openstack> Meeting started Tue Jun  7 15:00:49 2016 UTC and is due to finish in 60 minutes.  The chair is carl_baldwin. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:51 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:00:53 <openstack> The meeting name has been set to 'neutron_routed_networks'
15:00:58 <carl_baldwin> #topic Announcements
15:01:10 <carl_baldwin> #link https://etherpad.openstack.org/p/routed-provider-networks-notes
15:01:34 <carl_baldwin> N-1 is past and I actually think we're pretty close to where I was hoping to be at this point.
15:01:48 <mlavalle> ++
15:01:52 <carl_baldwin> Also, is everyone aware of the mid-cycle planned?
15:02:01 <blogan> yep
15:02:03 <mlavalle> I am
15:02:09 <john-davidge> yep
15:02:12 <mlavalle> 17 - 19 August
15:02:22 <blogan> yall have fun :)
15:02:32 <carl_baldwin> ^ Please take special note of the dates as they have been changed!
15:02:33 <jschwarz> yep
15:03:02 <carl_baldwin> It is going to be Wed - Fri, not Mon - Wed as was originally announced.
15:03:04 <mlavalle> This is the etherpad:
15:03:10 <mlavalle> #link https://etherpad.openstack.org/p/newton-neutron-midcycle
15:03:22 <carl_baldwin> mlavalle: Thanks!
15:04:06 <carl_baldwin> Any other announcements?
15:04:36 <carl_baldwin> #topic Devstack
15:05:09 <carl_baldwin> I got thinking yesterday that there is a fair amount of manual setup that needs to be done to get a devstack system to work with segments.
15:05:22 <carl_baldwin> I started a patch to hopefully provide some help
15:05:35 <john-davidge> great!
15:05:38 <carl_baldwin> #link https://review.openstack.org/326103
15:05:44 <carl_baldwin> It isn't much yet.
15:06:07 <mlavalle> I think this complements well what I've been doing with Vagrant
15:06:14 <carl_baldwin> But, hopefully it can evolve in to something helpful.
15:06:38 <mlavalle> I am creating a Vagrant setup to to create a multi-node devstack with segments
15:06:40 <carl_baldwin> mlavalle: How's that going?  Do you want to add information about what you're doing to this section of the etherpad?
15:06:42 <blogan> sounds like a great idea
15:07:01 <carl_baldwin> mlavalle: Maybe we can merge these efforts.
15:07:29 <mlavalle> carl_baldwin: yes I will add to the etherpad. I expect to port the product to github over the next couple of days
15:07:48 <carl_baldwin> mlavalle: Sounds good.
15:07:51 <mlavalle> as always, progress is slower that originally thought :-)
15:08:03 <mlavalle> but I am making steady progress
15:08:32 <mlavalle> and yes, let's merge these 2 efforts. I'll leverage your patchset
15:08:40 <carl_baldwin> mlavalle: Great.  Try to add to the etherpad when you have something that can be shared.
15:08:59 <carl_baldwin> #topic Associating Subnets to Segments
15:09:26 <carl_baldwin> There are a couple of patches in play
15:09:27 <carl_baldwin> #link https://review.openstack.org/#/c/295173
15:09:48 <carl_baldwin> #link https://review.openstack.org/#/c/315573/
15:10:05 <carl_baldwin> I've +2ed the first, it could use another core's eyes.
15:10:11 <carl_baldwin> The second is a client patch.
15:10:20 <carl_baldwin> Needs an update.
15:10:30 <honghuixiao_> I will udpate it in the following days
15:10:39 * honghuixiao_ is xiaohhui
15:10:58 <mlavalle> lol
15:11:34 * carl_baldwin didn't even notice honghuixiao_ 's changed nick
15:11:46 <carl_baldwin> Thanks, honghuixiao_
15:12:01 <honghuixiao_> It is just a temp name, might change back in another restart :)
15:12:05 <carl_baldwin> #topic Deferred  IP Allocation on Port Update with Binding Info
15:12:17 <carl_baldwin> This is where I'm working lately.
15:12:41 <carl_baldwin> The IPAM work has been approved
15:12:42 <carl_baldwin> #link https://review.openstack.org/320631
15:12:50 <carl_baldwin> Thanks blogan and haleyb for the great reviews.
15:13:12 <carl_baldwin> I need to work on the Nova side to allow deferred allocation.
15:13:20 <carl_baldwin> And, there is the bug with  ML2
15:13:32 <carl_baldwin> #link https://review.openstack.org/#/c/321152
15:13:53 <carl_baldwin> honghuixiao_: Sounds like you made some progress on this but I haven't had a chance to catch up.
15:14:36 <honghuixiao_> I need to make some change to it, but basically, I can run the test case successfully locally
15:14:38 <saggi> Sorry to barge in but if there is someone here for the Smaug meeting we are in #openstack-meeting-3 . We were sure it's an even week so we assume the room will be ours. Sorry for interrupting.
15:15:02 <carl_baldwin> saggi: No worries.  Thanks.
15:15:45 <carl_baldwin> honghuixiao_: Cool, let me know when you have a patch set up.
15:15:48 <carl_baldwin> honghuixiao_: Great work.
15:16:06 <honghuixiao_> I think the code is self-explanated, it is simple
15:16:07 <carl_baldwin> So, I think this is going well.
15:17:12 <carl_baldwin> I'm also going to get started on adding a flag to the port so that Nova can know when to allow deferred ip allocation.  I think that'll be easy once I get to it.
15:17:39 <carl_baldwin> That's all from me for this.  Anyone else?
15:17:49 <carl_baldwin> #topic Schedule DHCP to Segments and Create DHCP ports
15:17:55 <carl_baldwin> blogan: You're up.
15:18:05 <blogan> so i split the original review into 2
15:18:20 <blogan> #link https://review.openstack.org/#/c/326261/
15:18:38 <blogan> that's pretty straightforward
15:18:51 <blogan> #link https://review.openstack.org/#/c/311931
15:19:07 <blogan> that one is the one where the most discussion is going to happen, as it already has :)
15:19:25 <honghuixiao_> So, for a network with multiple segments, and no segment_subnet in it, we will only have one dhcp service for the network, right?(assuming config to 1)
15:19:36 <blogan> i need to read honghuixiao_'s latest comments more in depth, its morning for me :)
15:19:37 <honghuixiao_> yeah, that is me adding the comments, :P
15:20:22 <blogan> honghuixiao_: you mean multiple segments and no segment has a dhcp enabled subnet?
15:20:52 <honghuixiao_> yes, just the orignal muti-provider network.
15:21:02 <carl_baldwin> honghuixiao_: That sounds about right.
15:21:39 <carl_baldwin> What distinguishes a routed network is the subnets are attached to segments.  If DHCP enabled subnets are not attached to segments, then they should work as before.
15:22:02 <honghuixiao_> So, we should not delete the dhcp port in the segment automatically. Because that will cause the dhcp service unusable for the network.
15:22:26 <blogan> it'll make it unusable just for that segment
15:22:35 <blogan> a dhcp port should be created per segment
15:22:43 <blogan> an agent per segment
15:23:38 <carl_baldwin> honghuixiao_: Is there a comment about this already in the review?  If so, do you have a link to it?
15:23:55 <honghuixiao_> 1 sec
15:24:20 <honghuixiao_> https://review.openstack.org/#/c/317358/4/neutron/plugins/ml2/db.py
15:24:24 <honghuixiao_> line 340
15:24:36 <blogan> it's totally possible, and probable, that i'm uninformed on some of this
15:24:47 <carl_baldwin> #link https://review.openstack.org/#/c/317358/4/neutron/plugins/ml2/db.py@340
15:25:22 <honghuixiao_> learned, the link can be used this way
15:26:27 <carl_baldwin> honghuixiao_: I don't know why they didn't use the standard HTML anchor (e.g. #L340)
15:26:29 * mlavalle also just learned this
15:26:42 <blogan> so segment should be able to be deleted if there is only the dhcp port bound to it correct?
15:27:08 <honghuixiao_> I think the that is the result of discussion in previous mail list
15:27:35 <blogan> honghuixiao_: and your concern is that there is only one dhcp port per network?
15:27:52 <honghuixiao_> yes, that is the original logic
15:28:04 <honghuixiao_> And it is what I see when debug it
15:28:11 <carl_baldwin> I think we might need to consider what kind of network it is (routed or not)
15:28:42 <honghuixiao_> so a routed network, is the one with subnets associated with segments, right?
15:28:52 <blogan> honghuixiao_: if each segment gets its own dhcp port would that solve the problem? i thought that already happens, but now you have me doubting myself :)
15:28:57 <carl_baldwin> honghuixiao_: Right, that is the distinguisher.
15:29:28 <carl_baldwin> blogan: I think what you're saying is right *if* the network is a routed network.
15:29:44 <blogan> carl_baldwin: yes
15:30:11 <carl_baldwin> But, if this code is also handling L2 adjacent networks, then it might need to be a little bit smarter.
15:30:29 <carl_baldwin> I'd like to give the review a little more thought though before I'm sure.
15:30:56 <blogan> carl_baldwin: what do you mean?
15:32:11 <carl_baldwin> blogan: I'm not sure this code can assume it is working with a routed network.
15:32:33 <carl_baldwin> blogan: If it isn't, and that DHCP port is the only one serving a multi-segmented network, it might be bad to delete it.
15:32:41 <carl_baldwin> honghuixiao_: ^ Is this kind of what you're thinking?
15:32:50 <honghuixiao_> yes,
15:32:55 <mlavalle> that makes sense
15:33:18 <blogan> carl_baldwin: it should work the same if there are no segments
15:33:25 <blogan> i mean the same as before
15:33:45 <carl_baldwin> blogan: Segments couldn't be deleted before.
15:34:00 <honghuixiao_> yeah, we open the gate to operate segments now...
15:34:21 <blogan> ah
15:34:25 <reedip> sorry , some problem with my IRC, gets connection intermittently
15:34:55 <carl_baldwin> We have to be careful because we're adding the capability to add / delete segments for both routed networks and L2 adjacent networks.
15:35:20 <carl_baldwin> honghuixiao_: Is it okay if we provide feedback on that review a bit later?
15:35:29 <honghuixiao_> sure,
15:35:37 <carl_baldwin> I, for one, would like to let it sink in a little and then comment.
15:35:41 <blogan> okay i think i see where the problem might be, ill need to run through some testing of this to get a better grasp on it
15:36:05 <carl_baldwin> honghuixiao_: Thanks for bringing it up.  It will be good to think this through.
15:36:12 <blogan> +1
15:36:18 <honghuixiao_> :)
15:36:34 <carl_baldwin> blogan: honghuixiao_ : Anything else on DHCP?
15:36:50 <blogan> nope
15:36:52 <carl_baldwin> blogan: Thanks for updating the etherpad, that will be very helpful.
15:36:56 <honghuixiao_> I also leave a comment in the review, but I think it is OK to go on there
15:37:08 <manjeets> etherpad link please ?
15:37:24 <carl_baldwin> manjeets: ^^^ It is the first link in the meeting.
15:37:41 <carl_baldwin> #link https://etherpad.openstack.org/p/routed-provider-networks-notes
15:37:48 <manjeets> thanks carl_baldwin:
15:37:58 <carl_baldwin> manjeets: glad to help
15:38:01 <carl_baldwin> #topic Integration with Nova Scheduler
15:38:16 <carl_baldwin> I should've announced that the spec was merged!
15:38:26 <carl_baldwin> mlavalle: Thanks for all of your help on that.
15:38:35 <mlavalle> glad to help :-)
15:38:53 <blogan> good job!
15:39:00 <carl_baldwin> We have a few dependencies.  Any way we can help there?
15:39:08 <mlavalle> the challenge was not to miss the boat for Newton
15:39:47 <mlavalle> The focus shifts now from specs to tracking code
15:40:22 <mlavalle> I am tracking the new placement api
15:40:28 <mlavalle> #link https://review.openstack.org/#/c/293104/
15:40:45 <carl_baldwin> mlavalle: Are there some gerrit topics that we could link to the etherpad.
15:41:07 <mlavalle> Yes I will add them
15:41:18 <mlavalle> essentially, 2 topics
15:41:33 <mlavalle> #link https://review.openstack.org/#/q/topic:bp/generic-resource-pools
15:42:00 <mlavalle> This is for the generic resource pools
15:42:14 <mlavalle> and the changes to allocate_for_instance:
15:42:22 <mlavalle> #link https://review.openstack.org/#/q/topic:bp/prep-for-network-aware-scheduling,n,z
15:42:46 <mlavalle> last check, johnthetubaguy hasn't posted any code yet for this one
15:42:58 <carl_baldwin> mlavalle: Excellent.  Let's stay on top of those.  I'll try to work them in to my review routine.
15:43:18 <mlavalle> Yes, we can help with reviews
15:43:39 <mlavalle> and I have offered to johnthetubaguy to help with code if he thinks he needs it
15:44:04 <johnthetubaguy> yeah, I hope to start on that ASAP
15:44:12 <mlavalle> In the meantime, as soon as I have the vagrant multinode test environment ready, I will strat playing with the code that is in gerrit
15:44:35 <carl_baldwin> mlavalle: Sounds like a good plan.
15:45:07 <carl_baldwin> mlavalle: johnthetubaguy:  Anything else for discussion?
15:45:17 <mlavalle> not from me
15:45:40 <johnthetubaguy> not from me either really
15:45:53 <carl_baldwin> Thanks!
15:45:57 <carl_baldwin> #topic L2 Adjacency Extension
15:46:28 <carl_baldwin> reedip doesn't seem to be around.  I'll ping him out of band.
15:46:41 <carl_baldwin> #topic Enable Create / Delete of Segments on Existing Network
15:46:50 <mlavalle> he pushed a revision to his patchset
15:47:01 <carl_baldwin> #undo
15:47:01 <openstack> Removing item from minutes: <ircmeeting.items.Topic object at 0x7fe2376b8cd0>
15:47:05 <carl_baldwin> mlavalle: He did?
15:47:10 <mlavalle> yeah
15:47:17 <mlavalle> give me sec
15:47:55 <carl_baldwin> I guess I've already reviewed it so I didn't think it was new.  But, it has been in the last week.
15:48:50 <carl_baldwin> #topic Enable Create / Delete of Segments on Existing Network
15:49:13 <carl_baldwin> honghuixiao_: We've already started in to this a bit.  Anything else?
15:49:27 <honghuixiao_> just need more reviews to it
15:49:28 <honghuixiao_> https://review.openstack.org/#/c/317358
15:49:57 <honghuixiao_> Then nothing else
15:51:13 <carl_baldwin> honghuixiao_: ack and thank you.
15:51:36 <carl_baldwin> #topic Client
15:52:18 <carl_baldwin> rtheis: Anything need attention?
15:52:18 <rtheis> hi
15:52:51 <rtheis> I merged a fix to the OSC client for network segments
15:53:04 <rtheis> I would like to get to create/delete later this week
15:53:17 <rtheis> and look at associating subnets to segments
15:53:44 <rtheis> that's all for now
15:54:10 <carl_baldwin> rtheis: Thanks, I am trying to watch the etherpad, keep it up to date as things change.
15:54:15 <carl_baldwin> #topic OVN plugin
15:54:24 <carl_baldwin> Anything to discuss here?
15:54:31 <rtheis> OVN is now ML2 driver as of Friday
15:54:47 <carl_baldwin> Awesome!
15:54:54 <rtheis> I'm working on some ML2 related fixes and enhancements now.
15:55:12 <rtheis> and https://review.openstack.org/#/c/302623/ is out for review...lots of activity on it
15:55:51 <honghuixiao_> yeah, that is for geting the segment_host_maping for ovn now. Kyle is helping push the code forward
15:56:04 <carl_baldwin> Thanks, mestery !
15:56:13 <mestery> yw carl_baldwin :)
15:56:19 <mestery> Though see comments on that patch honghuixiao_ ....
15:56:26 <mestery> regXboi brough up an issue
15:57:00 <carl_baldwin> time check, we're almost done.
15:57:13 <carl_baldwin> Anything else needs attention here?
15:58:00 <carl_baldwin> #topic Router External Gateways on Routed Networks
15:58:21 <carl_baldwin> john-davidge has shown some interest in looking in to this.
15:58:35 <john-davidge> Yeah, it would be good to get some eyes back on #link https://review.openstack.org/#/c/300207 and the thread on the ML
15:58:52 <john-davidge> we have a dependency on that spec going ahead
15:59:11 <carl_baldwin> Yeah, that thread kind of died.  Do you have a link to it?
15:59:36 <john-davidge> i resurrected it a couple hours ago, but let me dig up the link
15:59:51 <carl_baldwin> john-davidge: I need to get caught up on the mailing list today.
16:00:09 <carl_baldwin> We're out of time.
16:00:12 <carl_baldwin> Thanks!
16:00:17 <carl_baldwin> #endmeeting