15:04:05 <tmorin1> #startmeeting bgpvpn
15:04:06 <openstack> Meeting started Tue Jun 14 15:04:05 2016 UTC and is due to finish in 60 minutes.  The chair is tmorin1. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:04:07 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:04:10 <openstack> The meeting name has been set to 'bgpvpn'
15:04:34 <tmorin1> hi doude, matrohon, pcarver, timirnich, wdeclercq
15:04:40 <pcarver> hi
15:04:45 <matrohon> hi
15:04:47 <mickeys> hi
15:04:51 <tmorin1> hi mickeys
15:05:55 <tmorin1> let's start...
15:05:59 <doude> tmorin1: hi
15:06:45 <tmorin1> #addchair matrohon
15:07:04 <tmorin1> #chair matrohon
15:07:04 <openstack> Current chairs: matrohon tmorin1
15:07:07 <tmorin1> better :)
15:07:13 <tmorin1> ok
15:07:17 <tmorin1> announces ?
15:07:48 <tmorin1> for personal reasons that was a slow week for me, not much to announce
15:08:03 <tmorin1> the work by cedric on the horizon UI has advanced well
15:08:06 <wdeclercq> hey *  (was afk)
15:08:16 <tmorin1> hi wdeclercq :)
15:08:32 <tmorin1> the change is I think ready for a review  (matrohon ?)
15:08:45 <tmorin1> #link https://review.openstack.org/#/c/322134/
15:09:08 <tmorin1> cedric will soon push the change for the non-admin horizon parts
15:10:13 <tmorin1> I've requested a 4.0.1 release to incorporate one bugfix on the bagpipe driver
15:10:20 <tmorin1> #link https://bugs.launchpad.net/bgpvpn/+bug/1589501
15:10:20 <openstack> Launchpad bug 1589501 in neutron "Request Mitaka maintenance release for networking-bgpvpn" [Low,New] - Assigned to Ihar Hrachyshka (ihar-hrachyshka)
15:10:58 <tmorin1> I hope ihrachys will find the time to pull the trigger soon
15:11:14 <tmorin1> other announces ?
15:11:27 <tmorin1> #topic new features
15:11:49 <tmorin1> (reminder: the pad where we discuss is at https://etherpad.openstack.org/p/bgpvpn_advanced_features)
15:11:52 <tmorin1> #link https://etherpad.openstack.org/p/bgpvpn_advanced_features
15:12:10 <tmorin1> doude, pcarver, did you find some time to have a look ?
15:12:58 <tmorin1> we have an audio meeting planned with Jan next Thursday to pursue the discussion
15:13:03 <tmorin1> anyone can join
15:13:04 <pcarver> tmorin1: I took a look and shared with a few folks but don't have any specific feedback yet
15:13:10 <tmorin1> 13:00 UTC
15:13:24 <tmorin1> just ask me for the invite
15:14:23 <tmorin1> pcarver: ok -- would you think it useful to reserve some Q&A time with them on the topic ?
15:14:24 <doude> tmorin1: I started to have a look, no comment for the moment
15:15:10 <pcarver> tmorin1: It would be useful, but I'm struggling to get critical items done before I go on vacation next week. I probably won't get to it before mid-July.
15:16:38 <tmorin1> pcarver: ok -- I'm note sure what to hope for: that we advance fast enough on the topic, or that we do not and we can wait for more feedback   :-/
15:17:00 <pcarver> One use case I am aware of though, is integration with networking-sfc. We're specifically interested in service chains that apply based on which MPLS VPN(s) the network is attached to.
15:17:17 <tmorin1> pcarver: this is a very important topic
15:19:26 <tmorin1> as quickly discussed in Austin, we need to progress the discussion with the -sfc team
15:20:39 <tmorin1> I see two possible ways:
15:20:39 <tmorin1> - BGPVPN--Network association +  SFC classifier referrring to that Network and inferring the presence of a BGVPN
15:20:40 <tmorin1> - have the SFC classifier directly refer to a BGPVPN    (as ingress, egress, or two BGVPNs ingress/egress)
15:22:29 <tmorin1> pcarver: ^^  does the new/future classifier spec already allows refererring to Networks rather than only IP prefixes
15:23:16 <pcarver> tmorin1: sorry. Double booked on voice conference call
15:23:26 <pcarver> be with you in a sec
15:26:03 <pcarver> ok, yes both of those look like potential ideas. In particular there's a desire that when new prefixes are learned via BPG that the flow classifier at the vSwitch level would use them to guide traffic through chains
15:28:22 <tmorin1> pcarver: yes
15:30:49 <pcarver> s/BPG/BGP/
15:32:15 <tmorin1> pcarver: this ^^ will require not only integration in terms of API abstraction, but also something well designed for the dataplane part
15:32:50 <pcarver> tmorin1: yes, agreed. I've only just started thinking about this.
15:34:57 <tmorin1> pcarver: ok
15:35:25 <tmorin1> pcarver: would you have a pointer on the status of the discussion on the new classifier ?
15:36:41 <pcarver> tmorin1: It hasn't really started yet. I've just had discussion with someone internal who doesn't really get involved with OpenStack. I need to shape my scribled handwritten notes into some sort of use cases or other public discussion
15:38:02 <tmorin1> pcarver: well, the team has had a few weekly meetings, I just need to find the logs/pad to have a look, I couldn't attend because of timezone issues
15:38:37 <pcarver> tmorin1: sorry, comprehension fail due to ongoing voice conference call. Thought we were still on the previous topic.
15:39:12 <pcarver> Yes, there has been some discussion on flow classifier but I don't have a good summary. I think there's been some confusion over meeting time.
15:39:12 <tmorin1> pcarver: sorry, my mistake as well
15:39:22 <tmorin1> pcarver: ok
15:40:28 <pcarver> http://eavesdrop.openstack.org/#Neutron_Common_Classifier_meeting
15:40:51 <pcarver> The link to meetings logs is broken. I'll see about fixing that
15:41:10 <tmorin1> pcarver: ok, thanks
15:41:19 <tmorin1> what else do we have to discuss ?
15:43:41 <mickeys> The patch set for BGP EVPN IP Prefix Advertisement has been updated to respond to comments, adding content as requested
15:43:45 <mickeys> #link https://review.openstack.org/#/c/322654/
15:45:35 <mickeys> An RFE was submitted to add VNI in the bgpvpns table
15:45:41 <mickeys> #link https://bugs.launchpad.net/bgpvpn/+bug/1590908
15:45:41 <openstack> Launchpad bug 1590908 in bgpvpn "[RFE] Add 'VNI' column in bgpvpns table in Neutron database." [Wishlist,Confirmed] - Assigned to Siddanagouda Khot (siddanmk)
15:47:13 <tmorin1> yes
15:47:17 <matrohon> mickeys : I'm not sure siddamk will be able to provide an implementation for the bug
15:47:41 <matrohon> mickeys, do you have any other volunteer?
15:47:43 <tmorin1> mickeys: I did a first review, and plan to revisit the change perhaps this week
15:47:48 <mickeys> steve_ruan will take over the bug. siddanmk sent him the patch in progress.
15:47:57 <tmorin1> mickeys: good
15:48:10 <matrohon> mickeys, great!
15:48:16 <tmorin1> I commented on the bug to give a link to the spec we had already written for the VNID
15:49:05 <mickeys> tmorin1: Thanks for the pointer. I will review.
15:49:27 <tmorin1> mickeys: good
15:49:53 <tmorin1> mickeys: there one thing I noticed, but haven't yet provided as comment on https://review.openstack.org/#/c/322654/
15:50:26 <tmorin1> mickeys: we may want to have on a said setup, both the DR E-VPN use case, and the BGPVPN use cases
15:50:55 <tmorin1> mickeys: to allow this, it means that the BGPVPN service plugin would need to have multiple drivers be enabled at the same time
15:51:11 <tmorin1> this is currently not supported (but not hard to support)
15:51:41 <mickeys> Anything specific to EVPN, or is this a general enhancement for any set of drivers?
15:51:58 <tmorin1> not specific to E-VPN
15:52:09 <tmorin1> general enhancement for the service plugin
15:52:14 <mickeys> tmorin1: Understood
15:53:29 <mickeys> I guess we also need an addition to the type value in bgpvpns for evpn. Should this be a separate BGPVPN RFE?
15:54:10 <tmorin1> mickeys: worth thinking about whether we need one or not
15:54:53 <tmorin1> mickeys: have you considered 'type' 'l3' with 'technique' that would be set to 'evpn-prefixes' -- that would seem to me as a nice fit
15:55:19 <tmorin1> http://docs.openstack.org/developer/networking-bgpvpn/future/attributes.html#technique-attribute
15:55:22 <mickeys> tmorin1: I forgot about the 'technique' that you showed me in Austin. Let me take a look at that.
15:55:44 <tmorin1> ok !
15:56:21 <tmorin1> like VNI, it's an attribute that we specified, but which has not been implemented yet
15:57:09 <mickeys> tmorin1: A related issue is the different options with regard to route types in EVPN prefix advertisement. My preference is to leave that to config, once more than one set of types is supported.
16:00:08 <tmorin1> perhaps
16:00:26 <tmorin1> I'm not sure what you need to control
16:00:52 <tmorin1> it's time to wrap up
16:00:59 <tmorin1> by everyone
16:01:04 <tmorin1> bye everyone
16:01:14 <tmorin1> many thanks
16:01:17 <tmorin1> #endmeeting