14:04:30 <mlavalle> #startmeeting neutron_routed_networks
14:04:31 <openstack> Meeting started Tue Jun 28 14:04:30 2016 UTC and is due to finish in 60 minutes.  The chair is mlavalle. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:04:32 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:04:35 <openstack> The meeting name has been set to 'neutron_routed_networks'
14:04:43 <mlavalle> #chair carl_baldwin
14:04:44 <openstack> Current chairs: carl_baldwin mlavalle
14:05:08 <carl_baldwin> Hi.  Sorry I'm late
14:05:19 <mlavalle> carl_baldwin: you are chair
14:05:45 <carl_baldwin> I'm mobile now.  At my desk in 5.
14:06:06 <blogan> glad to see i'm not the only one late :)
14:06:25 <mlavalle> so let's get moving
14:06:31 <mlavalle> #topic Annoucements
14:06:43 <mlavalle> Team etherpad is here:
14:06:51 <mlavalle> #link https://etherpad.openstack.org/p/routed-provider-networks-notes
14:07:45 <mlavalle> Also want to remind everybody that Newton-2 is approaching fast
14:08:22 <mlavalle> Here's the official calendar
14:08:29 <mlavalle> #link http://releases.openstack.org/newton/schedule.html
14:08:50 <mlavalle> We have until July 11 - 15 for Newton-2
14:09:07 <carl_baldwin> Wow
14:09:32 <mlavalle> Finally, I want to remind everybody about the Neutron mid-cycle in Cork Ireland.
14:09:39 <mlavalle> Etherpad is here:
14:09:46 <mlavalle> #link https://etherpad.openstack.org/p/newton-neutron-midcycle
14:09:47 <carl_baldwin> I booked my flight
14:09:57 <mlavalle> August 17 - 19
14:10:17 * mlavalle passes the mike to carl_baldwin
14:10:47 <carl_baldwin> Anyone else going for sure?
14:10:57 <mlavalle> I am going for sure
14:11:19 <mlavalle> and I know john-davidge will be there
14:11:20 <rtheis> I am
14:11:25 <john-davidge> yep, I'm booked
14:11:27 <carl_baldwin> mlavalle: cool.  It is hard to tell where we'll be in this project at that time.
14:11:35 <carl_baldwin> rtheis: john-davidge:  great
14:11:49 <carl_baldwin> Okay, I'm back at my desk.  Any more announcements?
14:12:22 <carl_baldwin> Okay.  Let's get going.
14:12:31 <carl_baldwin> #topic development environment
14:12:44 <carl_baldwin> mlavalle has done great work on the vagrant env.
14:13:06 <carl_baldwin> I've used it and it has helped me make some progress that I don't think I would've made yet otherwise.
14:13:28 <carl_baldwin> I encourage you to check it out to get a sense for what multiple segments is like.
14:13:40 <carl_baldwin> #link https://github.com/miguellavalle/routednetworksvagrant
14:13:53 * john-davidge has been checking it out this morning :)
14:14:17 <mlavalle> the lat remaining improvements, which is to enable routing in the iprouter vm will be finished be EOD today
14:14:31 <mlavalle> so if you test it today, please keep that in mind
14:14:43 <carl_baldwin> I think we should have DHCP worked out soon.  It is really just breaking on the agent getting a port on the network.
14:15:20 <mlavalle> the automatic configuration of the agent is compute2 will be added also today
14:15:31 <mlavalle> EOD meaning Texas time EOD
14:15:52 <carl_baldwin> mlavalle: ok.  I'll grab it when it is ready.
14:16:17 <carl_baldwin> As always, feedback and improvements are encouraged and welcome.
14:16:27 <blogan> thansk for the work on that
14:17:04 <carl_baldwin> #topic Segment CRUD
14:17:07 <mlavalle> in any form or shape. You can send me an email with improvment requests.... you can fork the repo and sed a pull request... please feel free
14:17:14 <carl_baldwin> #link https://review.openstack.org/#/c/293305
14:17:37 <carl_baldwin> #topic associating subnets to segments
14:17:48 <carl_baldwin> #link https://review.openstack.org/#/c/315573/
14:18:05 <newt_> A
14:18:11 <carl_baldwin> ^ client patch just needs a +2 in the neutron client.
14:18:28 <carl_baldwin> #topic segment host mapping
14:18:42 <carl_baldwin> #link https://review.openstack.org/#/c/329309/
14:18:49 <carl_baldwin> ^ Oh, that one merged last night.  Great.
14:18:53 <xiaohhui> https://review.openstack.org/#/c/329540/
14:18:54 <mlavalle> yes!
14:19:08 <xiaohhui> I think this one need some focus
14:19:30 <carl_baldwin> xiaohhui: Thanks for bringing it up.
14:19:51 <xiaohhui> no problem
14:20:33 <carl_baldwin> xiaohhui: I'll look today.  That looks like a good follow up.
14:20:53 <xiaohhui> Thanks
14:20:56 <carl_baldwin> #link https://review.openstack.org/#/c/334648
14:21:24 <carl_baldwin> ^ I need to follow up on this one and test it more thoroughly.  It is kind of a hack in its present state.
14:22:10 <carl_baldwin> But, with this patch and blogan's I was able to create a new network with running agents and all the mappings got created and DHCP was scheduled to both segments.
14:22:14 <carl_baldwin> Pretty cool.
14:22:38 <blogan> very cool
14:23:12 <carl_baldwin> I think segment aware IPAM is complete and working until we get to haleyb and john-davidge 's floating IP stuff.  So, we'll skip that for now.
14:23:41 <carl_baldwin> Guys, we've done a ton of good stuff in these topics ^^ .  It is pretty awesome.
14:23:53 <carl_baldwin> #topic DHCP
14:24:36 <carl_baldwin> blogan: How is it going?
14:25:02 <blogan> carl_baldwin: good, was going to check out the scenarios we discussed yesterday
14:25:04 <carl_baldwin> This merged, a great step forward:
14:25:04 <carl_baldwin> https://review.openstack.org/#/c/311931
14:25:30 <carl_baldwin> #link https://review.openstack.org/#/c/334672
14:25:37 <blogan> carl_baldwin: auto schedule still needs to be done too, ill be helping aradhana with that
14:25:44 <carl_baldwin> ^ I put this hack in my environment so that I could get DHCP.
14:26:29 <carl_baldwin> blogan: What do you think the right way to go about requesting DHCP ports is?
14:27:06 <blogan> carl_baldwin: honestly, i'm not sure yet, need to really dive into it.  i'll have a better answer later today :)
14:27:30 <carl_baldwin> Should the agent be smart enough to request the right subnets?  Or, should the server know that for a DHCP port, it should figure out the right subnets.
14:27:32 <carl_baldwin> ?
14:27:50 <blogan> i think i tshould be the server
14:28:09 <blogan> the server will already know the host
14:28:35 <blogan> so it can figure out the segment_id from that and thus, subnets
14:28:36 <carl_baldwin> Part of me leans that way too.
14:29:03 <carl_baldwin> The other part of me wonders if that is embedding too much specialized knowledge in the port create and IPAM logic.
14:29:14 <carl_baldwin> But, what you say makes sense to me.
14:29:50 <carl_baldwin> I wonder if we should pose the question to the mailing list.
14:30:03 <blogan> well my first gut reaction is that this is part of the ipam process, narrowing down the subnets, and the agent shouldn't know anything about it
14:31:08 <mlavalle> I say follow blogan's suggestion and submit to review quickly
14:31:09 <carl_baldwin> blogan: That does make some good sense.
14:31:37 <mlavalle> feedback tends to be more meaningful when there is code to discuss about :-)
14:32:36 <carl_baldwin> blogan: I'll leave it in your hands.  How would you like to get this feedback?
14:33:04 <blogan> carl_baldwin: i like gettign a review up first, and then doing an ML thread
14:33:22 <carl_baldwin> blogan: That sounds like a good plan.
14:33:24 <blogan> if it looks like there's some disagreement
14:34:11 <carl_baldwin> I will say that the changes on the server side will be more localized and probably easier to get done.
14:34:25 <carl_baldwin> ... because the server knows all it needs to know to find the right subnets.
14:34:44 <blogan> a good point too
14:34:51 <carl_baldwin> blogan: I look forward to seeing your patch and discussing.
14:35:06 <blogan> carl_baldwin: do you want me to take over that patch you have or just do antoher one?
14:35:13 <carl_baldwin> blogan: Just do another one.
14:35:32 <blogan> carl_baldwin: will do
14:35:37 <carl_baldwin> blogan: I'll keep mine handy as a work-around until the real solution is done and then abandon
14:35:50 <carl_baldwin> blogan: anything else?
14:36:01 <blogan> nope thats all i got
14:36:15 <carl_baldwin> blogan: Thanks!  Nice work on the scheduling.
14:36:20 <carl_baldwin> #topic Integration with Nova Scheduler
14:36:28 <mlavalle> hi
14:36:31 <carl_baldwin> mlavalle: How we doing?
14:36:47 <mlavalle> There are 2 pieces that we need to track of
14:37:15 <mlavalle> The first part is the series of patches by johnthetubaguy that starts here:
14:37:26 <mlavalle> #link https://review.openstack.org/#/c/328307
14:37:47 <mlavalle> This is a refactoring of the code in Nova that process ports for instances
14:38:09 <mlavalle> johnthetubaguy has been making incremental progress
14:38:17 <carl_baldwin> mlavalle: wrong link?
14:38:33 <johnthetubaguy> yeah, I think this patch is a better way to look at it: https://review.openstack.org/#/c/332712/
14:38:53 <mlavalle> thanks johnthetubaguy
14:39:22 <johnthetubaguy> I hope that patch before that WIP patch would be enough to unblock the initial work / prototype
14:39:37 <johnthetubaguy> I got some bits of the chain merged, but quite a bit more to go
14:39:55 <johnthetubaguy> there were comments on the earlier patches, and I hope those have now been addressed
14:40:01 <mlavalle> yeah, we haven't gotten to the point where we can call from the conductor yet....
14:40:19 <mlavalle> I thionk we are close, though
14:40:22 <johnthetubaguy> I don't think thats too important for the non-scheduler involved case
14:40:38 <johnthetubaguy> the main crux of the split is there though
14:41:37 <johnthetubaguy> I should share me thoughts on that
14:41:37 <carl_baldwin> johnthetubaguy: I'll take another review pass today to keep up.
14:41:45 <johnthetubaguy> I think there are two phases here
14:41:57 <johnthetubaguy> 1) assume infinite IPs, make routed networks work
14:42:06 <johnthetubaguy> 2) deal with the scheduling concerns
14:42:26 <johnthetubaguy> (1) may only be a prototype, but that should be unblocked now
14:42:52 <mlavalle> johnthetubaguy: is this series of patches going to deal with both phases?
14:43:47 <johnthetubaguy> so the freeze is really, really soon for Nova, June 30th
14:44:06 <johnthetubaguy> I would like to get in what is up for review, minus the WIP patch
14:44:24 <carl_baldwin> johnthetubaguy: What does this freeze mean?
14:44:41 <johnthetubaguy> that gets us to a place I think mlavalle could prototype the basic launch with routed networks (i.e. minus the scheduler bits)
14:44:59 <johnthetubaguy> carl_baldwin: basically no more code merged for non-priority efforts, which includes my patches
14:45:36 <johnthetubaguy> http://releases.openstack.org/newton/schedule.html#nova-non-priority-feature-freeze
14:46:43 <johnthetubaguy> I was hoping to get further, but its taken more patches than I had hoped to get the code into some half decent shape
14:47:55 <carl_baldwin> johnthetubaguy: I see
14:49:58 <carl_baldwin> Well, let's go to Open Discussion.
14:50:04 <carl_baldwin> #topic Open Discussion
14:50:52 <carl_baldwin> What did we leave out?
14:52:04 <mlavalle> I will go back to this tomorrow: https://review.openstack.org/#/c/331497/
14:52:10 <carl_baldwin> johnthetubaguy: Is the generic resource pool and scheduler work a priority?
14:52:30 <johnthetubaguy> carl_baldwin: yes, that is a priority effort
14:53:45 <carl_baldwin> johnthetubaguy: thanks
14:56:35 <carl_baldwin> Well, I think that's about all I have.
14:57:09 <carl_baldwin> Going once...
14:58:29 <carl_baldwin> Going twice...
14:59:18 <carl_baldwin> #endmeeting