17:01:23 <cathy_> #startmeeting SFC project
17:01:23 <openstack> Meeting started Thu Jun 11 17:01:23 2015 UTC and is due to finish in 60 minutes.  The chair is cathy_. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:01:24 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
17:01:26 <openstack> The meeting name has been set to 'sfc_project'
17:01:29 <johnsom> o/
17:01:32 <cathy_> hi everyone
17:01:34 <qwebirc1003062> Hi Cathy, This is Ramanjaneya..from vikram team
17:01:35 <pcarver> hi
17:01:39 <igordcard_> cathy_: hello
17:01:43 <igordcard_> hi all
17:01:53 <LouisF> hi Ramanjaneya
17:02:01 <cathy_> ok let's start
17:02:41 <cathy_> #topic update on action items of last meeting
17:03:16 <cathy_> So the repository for this feature development is being created.
17:03:21 <Swami> hi
17:03:38 <cathy_> armax: are you there?
17:03:45 <cathy_> Swami: hi
17:03:52 <armax> hi
17:03:55 <nbouthors> hello
17:04:43 <cathy_> For the network controller driver, I have updated the slide for that
17:04:52 <cathy_> nbouthors: hi
17:05:18 <cathy_> armax: do you have if the repository has been created or the infra team is still reviewing it?
17:05:32 <cathy_> I means do you know?
17:05:48 <armax> cathy_: it’s still in progress…I suspect the infra team is busy dealing with a few project renames
17:05:59 <armax> cathy_: it’ll be a little while longer
17:06:06 <cathy_> armax: ok, thanks
17:07:17 <cathy_> here is the link to the slides https://docs.google.com/presentation/d/1SpVyLBCMRFBpMh7BsHmpENbSY6qh1s5NRsAS68ykd_0/edit?usp=sharing
17:07:35 <cathy_> any comment or questions on the action items?
17:08:10 <armax> cathy_: let me reply to your question regarding agentless architecture in this forum
17:08:18 <cathy_> Ok, let's go to the next topic
17:08:29 <armax> ?
17:08:38 <cathy_> #topic Neutron port chain API
17:09:08 <cathy_> armax: go ahead
17:09:18 <Swami> cathy_: can you post your question here for the benefit of the audience.
17:09:57 <fitoduarte> .
17:10:01 <LouisF> armax: go ahead
17:10:25 <armax> cathy_: ok
17:10:40 <armax> I am assuming that in slide 2
17:11:20 <armax> that the component running the ovs agent is compute host?
17:11:25 <armax> or can it be something else?
17:11:41 <Swami> armax: Yes it is the compute host.
17:11:54 <LouisF> armax: yes
17:12:01 <cathy_> armax: yes
17:12:10 <armax> having said that, I was asking if you guys had considered making the interaction between the controller node and the compute node without relying on the ovs agent
17:12:22 <yamahata> Both traffic classifier and SFF?
17:12:38 <cathy_> what do you mean by the controller node?
17:12:53 <Swami> controller node is the neutron server node.
17:12:55 <LouisF> armax: can you give more details and perhaps and exmaple of where this is done?
17:13:02 <LouisF> an example
17:13:29 <armax> anyhow some for thought, I think it’s premature to talk about the architecture if we don’t first finalize whe most rudimentary API we want to implement this cycle
17:13:54 <armax> LouisF: well, we don’t necessily need to talk to the ovs agent, that’s all
17:14:28 <Swami> LouisF: for example in the current OVN architecture there is no agent in the compute node. So in future if we want to adopt such technology, we should be able to go agentless. That is what armax is pointing out.
17:14:59 <cathy_> In existing OVS, it talks to the neutron Server through OVS agent. Are you suggesting to change this?
17:15:09 <armax> all I am saying is that OVS can be controlled remotely
17:15:26 <armax> just like other solutions do
17:16:19 <armax> cathy_: I am simply asking if you had considered it
17:16:26 <armax> cathy_: I am not suggesting anything right now :)
17:16:54 <LouisF> sonds like that is worthwile considering
17:17:02 <cathy_> armax: it is a possible option in the future
17:17:35 <armax> cathy_: I wouldn’t rule it out completely, and consider it a ‘future’ option
17:18:03 <cathy_> Will the OVN architecture available in L release?
17:18:31 <armax> cathy_: because until we figure out what interaction is required with the vswitch everything is fair game
17:18:41 <armax> cathy_: OVN has nothing to do with what I am saying
17:19:24 <cathy_> I see that Swami said that you are referring to OVN architecture.
17:19:44 <Swami> cathy_: I just gave an example for LouisF.
17:19:59 <armax> Swami: no I wasn't
17:20:48 <cathy_> Maybe it is better to discuss this in the email to clarify what you have in mind. Let's move on to next topic
17:21:03 <cathy_> armax: OK with you?
17:21:38 <armax> cathy_: yss
17:21:47 <pcarver> cathy_: other examples would be Contrail and Nuage. Nothing specific to OVN, just lots of options out there that either do or will support service chaining
17:22:12 <pcarver> The service chaining API shouldn't be tied tightly to any one implementation
17:22:33 <armax> pcarver: noone is saying otherwise as far as I can tell :)
17:22:46 <LouisF> pcarver: agree, there should drivers for a wide variety of backend implementations
17:22:52 <cathy_> So if you are talking about different types of SDN controller, then it is through controller driver which is separate from the OVS driver. how about we discuss this in the email since the meeting time is short.
17:22:52 <LouisF> should be
17:23:07 <cathy_> $topic Neutron port chain API for SFC
17:23:19 <cathy_> #topic 2.	Neutron port chain API for SFC
17:23:36 <armax> cathy_: I recall that vikram was going to respin https://review.openstack.org/#/c/177946
17:23:43 <armax> cathy_: any news on that? I have seen nothing popping up
17:24:03 <armax> is vikram around?
17:24:09 <Mohankumar_> HI cathy ..
17:24:17 <Mohankumar_> mohan here from vikram team
17:24:19 <LouisF> armax: i will be updating, vikram is on vavation
17:24:23 <LouisF> vacation
17:24:28 <cathy_> armax: vikram is on vacation. Louis will take care of this
17:24:43 <Mohankumar_> we started neutron client changes
17:24:45 <armax> LouisF: ok, thanks, I think ultimately that spec will have to be moved over to the repo for SFC once we have it up and running
17:25:01 <LouisF> armax: yes thanks
17:25:01 <armax> LouisF: but for now, let’s iterate on that patch
17:25:09 <LouisF> armax: agree
17:25:17 <cathy_> We will update the spec incorporating all the comments/input
17:25:39 <qwebirc1003062> Hi I'm Ramanjaneya from vikram team..
17:25:40 <qwebirc1003062> on behalf of vikram, Mohan started neutron client changes..
17:26:26 <cathy_> let's now discuss the API and see if any questions/comments?
17:26:55 <armax> qwebirc1003062: the neutron client changes are the last thing you wan’t to do
17:27:03 <armax> *want
17:27:35 <cathy_> I know quite some people have given comments on that spec. Any issue/input that we can discuss in this meeting?
17:28:10 <cathy_> To move forward and make it for the L release, we need to reach consensus on the API and finalize the API
17:28:11 <LouisF> we need to get conensus and agreement  on the api first before making any changes
17:28:48 <Mohankumar_> agree
17:28:59 <cathy_> maybe we should put a time line on finalizing the API
17:29:06 <cathy_> ?
17:29:11 <armax> cathy_: reviewing the API in the current status of the spec is a bit painful
17:29:21 <armax> cathy_: as the wrong diff gets in the way
17:29:49 <armax> this might deter some people from reviewing perhaps?
17:30:00 <LouisF> armax: what do you mean by that?
17:30:18 <cathy_> armax: what do you mean by wrong diff?
17:30:18 <armax> LouisF: https://review.openstack.org/#/c/177946/
17:30:38 <armax> it shows that patch 177946 depends on an abandoned patch
17:30:48 <armax> and the actual diff includes two modified files and a deleted one
17:30:59 <armax> for something that should be freshly proposed that hardly makes any sense
17:31:18 <armax> the diff is: +124, -510
17:31:25 <armax> when it should simply be only additions
17:31:28 <cathy_> armax: yes, that is a confusing. Louis could you fix that?
17:31:32 <LouisF> armax: ok
17:33:23 <cathy_> so Louis will clean that up. Quite come people have given comments and there are multiple updated versions. I would suggest that everyone goes to the link and check the latest version and see if you have any more comments.
17:34:37 <armax> cathy_: yes, I’d give us one more week to see if we can pull it together
17:34:45 <cathy_> How about we get the review completed in two weeks from today so that we can start developing the code?
17:35:01 <cathy_> armax: one week is OK with me
17:35:05 <armax> cathy_: once the repo is set up and we move the spec over there, we can have one final push and start iterating on the code itself
17:35:39 <LouisF> armax: i will put the spec in the networking-sfc repo
17:36:15 <armax> cathy_: do we know of anyone who has started coding on a skeleton of the architecture that you put together with Sw?
17:36:20 <armax> *Swami
17:36:31 <Swami> armax:hi
17:36:32 <cathy_> armax: Ok, so let's get the review done in one week and then we will have the spec over the new repo and start implementing it
17:37:32 <cathy_> some people have contacted me to sign up for development of different pieces of the architecture, We will cover this later in this meeting.
17:37:43 <armax> cathy_: cool
17:37:51 <cathy_> that is one of our meeting topics too
17:38:07 <cathy_> now let's go to next topic
17:38:34 <cathy_> #topic Unified API and data model for flow classifier that can be used for SFC, QoS, Packet forwarding etc.
17:39:50 <cathy_> Since Yujin can not join this IRC meeting and Vikram has signed up on driving this and Vikram is on vacation. let's move this topic to next meeting.
17:40:17 <Swami> cathy_: Here is the link to the spec. #link https://review.openstack.org/#/c/186663/
17:40:24 <cathy_> So let's go to next topic
17:40:38 <armax> cathy_: although I am in favor of this initiative, I am not sure it’s the most sensible thing to do right now
17:41:21 <armax> cathy_: I wonder if it’s better striving for some unification once we got some practical experience on how all these initiatives play out
17:41:24 <cathy_> Swami: the flow classifier in https://review.openstack.org/#/c/186663/ is very similar to the flow classifier proposed in https://review.openstack.org/#/c/177946/
17:41:38 <LouisF> armax: is would be useful to have a common api for a flow classifier
17:41:50 <cathy_> armax: agree
17:42:23 <armax> cathy_: especially if these initiatives are tacked separately, it may be difficult to coordinate who does what
17:42:27 <armax> LouisF: agree
17:43:07 <cathy_> Since what Yujin proposed and wanted for packet forwarding is very close to what is proposed by the SFC spec https://review.openstack.org/#/c/177946/, let's implement the flow classifier in this SFC project but make it a independent module which cna be used or extended for other use case
17:44:05 <Swami> cathy_: have you communicated to Yujin on this.
17:44:20 <cathy_> Agree with Armax that let's first play it out and get some practical experience.
17:44:25 <cathy_> Swami: I will via email
17:45:16 <cathy_> Swami: it will be confusing to propose very similar API and data model in two separate spec. Let's merge it into one spec
17:45:41 <Swami> cathy_: agree
17:45:58 <cathy_> Shall we move to the next topic?
17:45:59 <LouisF> lets suggest that to Yuji
17:46:12 <cathy_> LouisF: yes
17:46:43 <cathy_> #topic : functional module breakdown and Module development ownership sign-up
17:47:06 <cathy_> let's reference to the slide and I will go one by one
17:47:17 <cathy_> slide 4
17:48:04 <cathy_> 1. repo creation, Armax has taken this. Thanks Armax!
17:48:26 <cathy_> 2 Integration with Neutron/devstack, CLI, Horizon, Heat, Anyone wants to take this?
17:49:33 <cathy_> Mohankumar_: would you like to take this work?
17:50:02 <Mohankumar_> yes cathy
17:50:17 <cathy_> great, Thanks Mohankumar_ !
17:50:47 <cathy_> now Service chain API Extension. Anyone takes this?
17:50:49 <LouisF> i can take that
17:50:57 <cathy_> Thanks Louis!
17:51:11 <Mohankumar_> thanks :)
17:51:19 <cathy_> Service chain Plugin: API handling and Data Base. Anyone?
17:51:42 <LouisF> i'l do that also
17:51:48 <cathy_> I can take this
17:51:56 <cathy_> Thanks Louis!
17:52:44 <cathy_> OVS Driver  ?
17:52:53 <LouisF> we need to also have a driver manager
17:53:20 <cathy_> LouisF: do you mean the common service chain API shim layer?
17:53:24 <LouisF> to handle the different backend drivers eg ovs, odl....
17:53:31 <LouisF> yes
17:53:51 <cathy_> common service chain API shim layer. Anyone wants to take this piece?
17:54:14 <LouisF> i will look at that
17:54:25 <cathy_> ok, thanks louis!
17:54:49 <cathy_> OVS agent on Host. Anyone taking this piece?
17:54:52 <armax> cathy_: it’s probably better to ask what LouisF doesn’t want to do :)
17:55:03 <cathy_> armax: :-)
17:55:06 <armax> cathy_: for that I’d wait a little longer, as we might not  even need it
17:55:27 <LouisF> please volunterer and i am glad to hand off
17:55:40 <LouisF> you mean use ovn?
17:55:40 <cathy_> armax: OK, let's go to the next piece
17:55:45 <armax> this task assigment is going to be in flux anyway…in my experience people come and go so we’d need to be prepared to step in :)
17:55:51 <armax> LouisF: forget about OVN :)
17:56:11 <armax> LouisF: I never mentioned it and blame Swami for putting it in people’s mind
17:56:12 <armax> s
17:56:18 <Swami> LouisF: If you want me to take up the ext and plugin part I can take it.
17:56:23 <LouisF> :]
17:56:27 <cathy_> armax: 100% agree with that we need to be prepared to step in.
17:56:44 <Swami> armax: I swear that I just gave an example
17:56:53 <cathy_> Service chain Classifier on host. Anyone?
17:57:23 <armax> Swami: I know, I know :)
17:57:31 <IlyaG> Hi! anyone tried to run CI on top of Fuel Juno using ESX?
17:57:33 <LouisF> may nicolas?
17:57:36 <LouisF> maybe
17:57:37 <cathy_> I know that Nicolos will take this piece. He has told me that
17:58:13 <cathy_> Service function forwarder with NSH Type 2 encapsulation. Anyone?
17:58:31 <cathy_> Swami: would you like to take this?
17:58:54 <cathy_> I can help with this piece too
17:59:30 <cathy_> Swami: ?
17:59:50 <Swami> cathy_: sorry I was looking at some other information.
17:59:50 <igordcard> What are you going to use as NSH-aware VNFs?
17:59:51 <nbouthors_> cathy: we can provide a service classifier
17:59:57 <Swami> cathy_: yes i am here
18:00:12 <Swami> cathy_: I can help with the classifier.
18:00:51 <cathy_> igordcard: not NSH-aware VNF for the first release
18:01:02 <cathy_> let's wrap the meeting up.
18:01:11 <cathy_> and continue in next meeting.
18:01:17 <cathy_> Thanks everyone. Bye now
18:01:30 <igordcard> bye
18:01:35 <armax> bye
18:01:45 <Mohankumar_> bye !
18:01:48 <cathy_> #endmeeting