15:04:07 <mlavalle> #startmeeting neutron_l3
15:04:09 <openstack> Meeting started Thu Mar 24 15:04:07 2016 UTC and is due to finish in 60 minutes.  The chair is mlavalle. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:04:11 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:04:14 <openstack> The meeting name has been set to 'neutron_l3'
15:04:26 <mlavalle> #topic Announcements
15:04:43 <mlavalle> We made it to Newton
15:05:02 <carl_baldwin> Yay!
15:05:09 <mlavalle> Thanks everybody for their hard work during Mitaka.... We'll do it all over again in Newton :-)
15:05:58 <mlavalle> We are a month away from the Austin Summit. If you have talks scheduled, start geeting ready. It will be here before we know
15:06:14 <carl_baldwin> ++
15:06:28 <mlavalle> and I apologize for not updating the agenda. I thought Ryan was doing it ;-(
15:06:35 <carl_baldwin> ++
15:07:00 <mlavalle> any other annoucements
15:07:07 <mlavalle> ?
15:07:34 <carl_baldwin> Not from me
15:07:44 <mlavalle> ok, let's move on
15:07:50 <mlavalle> #topic Bugs
15:08:09 <mlavalle> We have the same 3 bugs from last week.
15:08:18 <mlavalle> First up is https://bugs.launchpad.net/neutron/+bug/1478100
15:08:18 <openstack> Launchpad bug 1478100 in neutron "DHCP agent scheduler can schedule dnsmasq to an agent without reachability to the network its supposed to serve" [High,In progress] - Assigned to Cedric Brandily (cbrandily)
15:08:50 <carl_baldwin> We need to get back to this one.  It is related to DHCP stuff we need to do.
15:09:23 <mlavalle> it was marked as mitaka RC by armax and has a patchset ready for review
15:09:36 <mlavalle> https://review.openstack.org/#/c/205631/
15:09:46 <mlavalle> marked as RC potential
15:10:12 <carl_baldwin> I don't see it marked rc potential anymore.
15:10:47 <mlavalle> thanks for the clarification :-)
15:10:58 <mlavalle> any other comments
15:11:00 <mlavalle> ?
15:11:24 <carl_baldwin> No, just that we need to figure out who is going to handle DHCP for routed networks.
15:11:39 <mlavalle> good point
15:12:11 <mlavalle> ok, next up is https://bugs.launchpad.net/neutron/+bug/1533034
15:12:11 <openstack> Launchpad bug 1533034 in neutron "an unclear error info returned when create an ipv6 subnetpool associate to ipv4 address scope" [High,In progress] - Assigned to Cedric Brandily (cbrandily)
15:12:37 <mlavalle> No change since last week
15:13:00 <armax> carl_baldwin, mlavalle yes, I couldn’t see how it could make the RC cut
15:13:09 <mlavalle> it was moved to Newton-1 milestone
15:13:20 <mlavalle> by armax
15:13:20 <carl_baldwin> I can't figure out why it was bumped up to High.
15:13:46 <mlavalle> I'll ping armax later and asked why
15:14:08 <armax> did I bump something to High without providing a reason?
15:14:15 <carl_baldwin> I'm having trouble staying connected.
15:15:09 <carl_baldwin> armax: https://bugs.launchpad.net/neutron/+bug/1533034
15:15:09 <openstack> Launchpad bug 1533034 in neutron "an unclear error info returned when create an ipv6 subnetpool associate to ipv4 address scope" [High,In progress] - Assigned to Cedric Brandily (cbrandily)
15:15:41 <carl_baldwin> Maybe there is something about it that I don't see yet.
15:15:41 * mlavalle posted bug in Neutron channel for armax to review at his leisure
15:16:34 <mlavalle> ok, let's move on
15:16:34 <armax> I think I did because I noticed the code was using id() rather than whatever it was supposed to do
15:17:08 <armax> and I feared there was something more to it
15:17:09 <carl_baldwin> armax: I'll have a look today.
15:17:42 <armax> it can probably be demoted now if have assessed there’s nothing else broken with the workflow
15:18:01 <armax> but the use of an undefined variable that didn’t get exposed via testing freaked me out at first
15:18:11 <mlavalle> ok, I'll set it at medium at the end of the meeting
15:18:52 <mlavalle> thanks armax for the clarification :-)
15:19:05 <mlavalle> next up is https://bugs.launchpad.net/neutron/+bug/1543094
15:19:05 <openstack> Launchpad bug 1543094 in neutron "[Pluggable IPAM] DB exceeded retry limit (RetryRequest) on create_router call" [High,In progress]
15:19:09 <armax> mlavalle: I did already
15:19:27 <mlavalle> johnbelamaric, pavel_bondar: you made some comments lately about this onw
15:19:48 <carl_baldwin> mlavalle: tidwellr is working on this.
15:19:51 <johnbelamaric> mlavalle: I think ryan took a look and Carl suggested changing the allocation mechanisms.
15:20:08 <carl_baldwin> We're going to try something new which Ryan has in review.
15:20:13 <carl_baldwin> Let me find it.
15:20:20 <mlavalle> carl_baldwin: ah ok... yeah I know he played with it last week
15:20:26 <johnbelamaric> One suggestino would be to rollback the changes done in Mitaka. this would leave open a concurrency issue for Galera, but would be better than what is there now
15:20:42 <mlavalle> I'll assign it to Ryan
15:20:42 <johnbelamaric> I would suggest then that we do the new implementation as a separate driver
15:20:51 <carl_baldwin> #link https://review.openstack.org/#/c/292207/
15:21:09 <carl_baldwin> johnbelamaric: Why a new driver?
15:21:21 <johnbelamaric> Hmm. looking at this it is in the non pluggable code?
15:21:39 <johnbelamaric> Well, a new driver leaves current behavior (sequential IPs) in place.
15:21:43 <carl_baldwin> johnbelamaric: It is in both.
15:21:54 <johnbelamaric> Ah. ok
15:21:56 <carl_baldwin> Ryan started playing with the non-pluggable code first, I think.
15:22:16 <johnbelamaric> As I see it, the pluggable infrastructure is ideal for experimenting wiht this kind of change.
15:22:34 <carl_baldwin> For those that aren't up to date.  The idea is to try dropping the availability ranges table entirely.
15:23:11 <carl_baldwin> Then, avoid congestion in allocation by picking a new IP randomly from a window of next available IPs.
15:23:31 <carl_baldwin> ^ Turns out this is the tricky part.  Many UTs depend on predictable IP allocation.
15:23:42 <carl_baldwin> (end summary)
15:24:05 <johnbelamaric> So, UTs that test things other than the actual IPAM can still use the sequential driver
15:25:16 <carl_baldwin> johnbelamaric: interesting.  We have to weigh that against the difficulty of supporting more drivers.
15:25:55 <carl_baldwin> Anyway, I'm sure Ryan will have more for us next week.
15:26:20 <mlavalle> OK, moving on... any other bugs I might have missed?
15:27:08 <mlavalle> if not...
15:27:09 <carl_baldwin> johnbelamaric: I'll pass it on to Ryan.  It is an interesting point.
15:27:15 <johnbelamaric> carl_baldwin: ok
15:27:17 <carl_baldwin> mlavalle: No, not that I can think of now.
15:27:27 <mlavalle> #topic Routed Networks
15:28:05 <carl_baldwin> We have lots to do!
15:28:11 <mlavalle> we do!
15:28:26 <carl_baldwin> We need to commit someone to looking at DHCP.
15:28:52 <carl_baldwin> I'd also like to see if there is someone who could handle create / delete of segments through the segment API for ML2.
15:29:30 <carl_baldwin> Yesterday, I finished splitting up my work so that I have a patch that allows reading segments through the API but they still must be created through the multiprovidernet extension.
15:29:51 <mlavalle> right
15:30:02 <carl_baldwin> I broke out a WIP patch where I started to work on create / delete but I realized that we could let someone else work on that while we make progress in other areas.
15:31:03 <carl_baldwin> This person would hopefully already be familiar with ML2 and segments.
15:32:24 <carl_baldwin> I'm going to proceed with associating subnets with segments and look at IPAM changes.
15:32:48 <carl_baldwin> That's where I am.
15:33:26 <mlavalle> I am working on the patchset to move the segments table out of ml2
15:33:42 <mlavalle> I am refining the migration script
15:34:02 <mlavalle> Also working on the host segments mapping
15:34:27 <carl_baldwin> mlavalle: Oh, I also have some rebasing in progress locally that is dependent on that moving table patch.
15:35:06 <mlavalle> And also have a pretty good idea on where to start on the Nova side. We need to get to agreement on the nova neutron synch up when port is created in  Neutron with an IP
15:35:39 <mlavalle> Later today I will review the spec with my proposal on how to address that
15:36:13 <carl_baldwin> mlavalle: I updated the spec and I put a diagram in one of my comments.
15:36:15 <mlavalle> It wasn't very well received during the Tuesday meeting but I think it will enrich the discussion
15:36:59 <mlavalle> that's all I have
15:37:05 <mlavalle> anything else?
15:37:37 <carl_baldwin> Nope
15:37:44 <mlavalle> ok, moving on
15:38:08 <mlavalle> #topic BGP dynamic routing
15:38:37 <mlavalle> I don't see vikram around, so maybe we should skip this topic
15:38:53 <carl_baldwin> ++
15:38:59 <mlavalle> #topic DNS
15:39:28 <mlavalle> This topic is mostly done. janzian is enriching the unit tests, adding test cases for floating ips
15:39:45 <mlavalle> janzian: anything to share with the team?
15:40:10 <janzian> Not really, just chugging along with it
15:40:30 <mlavalle> ok, ping us when the patchset is up for review
15:40:36 <janzian> Will do
15:41:30 <mlavalle> #topic Open agenda
15:41:49 <mlavalle> any other topics the team might want to bring up?
15:43:12 <mlavalle> ok, thans all for attending. tty next week!
15:43:20 <mlavalle> #endmeeting