17:00:51 <cathy_> #startmeeting service_chaining
17:00:52 <openstack> Meeting started Thu Sep  3 17:00:51 2015 UTC and is due to finish in 60 minutes.  The chair is cathy_. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:00:53 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
17:00:56 <openstack> The meeting name has been set to 'service_chaining'
17:01:00 <cathy_> hi everyone
17:01:02 <s3wong> hello
17:01:05 <pcarver> hi
17:01:06 <johnsom> o/
17:01:19 <xgerman> o/
17:01:24 <mohankumar_> Hi
17:01:28 <georgewang> hi
17:01:49 <cathy_> Any topic you would like to discuss today?
17:01:56 <LouisF> hi
17:02:01 <abhisriatt> hi
17:02:12 <s3wong> cathy_: no official agenda?
17:02:22 <Brian__> Hi
17:02:55 <cathy_> I have something to discuss with you
17:02:59 <vikram> hi
17:03:06 <LouisF> vikram: hi
17:03:39 <cathy_> Paul sent an email to openstack-dev about the packaging and release of networking-sfc. Have you read the related emails?
17:04:15 <cathy_> I think we need to sort out clearly what changes we need to put into other OpenStack repo.
17:04:40 <vikram> cathy_:  other OpenStack repo?
17:04:40 <cathy_> First Thanks Paul Carver for raising that question in the alias!
17:04:46 <pcarver> cathy_: +1 That's the main thing that's bothering me right now
17:05:13 <cathy_> vikram: I mean, if we need to make any change in Neutron core component or Neutron client component etc.
17:05:18 <pcarver> We need to modify things like OvS agent, neutronclient, etc
17:05:25 <vikram> cathy_: I will be doing the release for networking-onos .. Will share you the info once i am clear next week
17:05:38 <cathy_> vikram: great!
17:05:42 <pcarver> But we're trying to follow the "big tent/stadium" model with our own repo
17:05:51 <cathy_> for API and DB, we are good since they are all new files
17:05:59 <cathy_> pcarver: yes
17:06:40 <vikram> cathy_: from neutronclient all the required changes we have merged
17:06:42 <cathy_> for CLI part, mohankumar_ and vikram : is there any hook we need to add to existing component?
17:07:20 <vikram> cathy_: no dependency
17:07:29 <cathy_> vikram: great!
17:07:41 <vikram> cathy_: all are merged
17:07:53 <cathy_> vikram: OK, got it. thanks!
17:07:53 <pcarver> vikram: Do you mean merged into the neutron-client repo or the networking-sfc repo?
17:08:04 <vikram> neutron-client repo
17:08:31 <pcarver> Does that mean we don't need the neutronclient directory in the networking-sfc repo anymore?
17:09:23 <vikram> pcarver: I meant the dependent changes for python-neutronclient is merged
17:09:56 <pcarver> vikram: Oh, ok. What I'm concerned with is the neutronclient stuff that we're putting in the networking-sfc repo
17:09:59 <vikram> pcarver: now networking-sfc will not have any dependencies with neutronclient for CLI
17:10:15 <LouisF> vikram: so we still have the neutronclient dir in networking-sfc?
17:10:33 <vikram> LouisF: yes we will have
17:10:45 <LouisF> vikram: ok
17:11:17 <vikram> LouisF: networking-sfc cmds must reside in it's repo
17:11:21 <vikram> like https://github.com/openstack/networking-bgpvpn/tree/master/networking_bgpvpn/neutronclient
17:11:46 <pcarver> So how does it work, when deploying/installing neutronclient code that's in the networking-sfc repo?
17:11:58 <pcarver> And similar question for other parts such as OvS agent
17:12:36 <vikram> pcarver: networking-sfc setup.cfg will have entry-points for the new cli's
17:13:04 <vikram> https://review.openstack.org/#/c/210008/7/setup.cfg
17:13:23 <vikram> pcarver: you can refer these changes
17:15:27 <pcarver> Ok, I think I understand. So neutronclient will pick up the networking-sfc extensions dynamically without needing to make changes to the actual neutronclient code.
17:15:44 <pcarver> Does that also work the same way for the OvS agent?
17:16:19 <vikram> pcarver: +1 for neutronclient
17:16:44 <vikram> pcarver: I am not aware of OvS agent changes .. so can't comment
17:16:53 <pcarver> And does it work the same way for the API? Core neutron will automatically detect and load the networking-sfc API?
17:19:06 <vikram> pcarver: If we check :https://github.com/openstack/networking-sfc/blob/master/requirements.txt
17:19:11 <cathy_> pcarver: yes, we will add hooks to the setup.cfg
17:19:13 <vikram> -e git://git.openstack.org/openstack/neutron.git#egg=neutron
17:19:33 <vikram> + setup.cfg.. will take care of this
17:20:48 <vikram> cathy_: We need the server code for verifying client + horizon code
17:20:51 <cathy_> vikram: you only add new networking-sfc files to neutronclient directory and the neutron client will automaticlaly pick up your new files
17:21:24 <vikram> cathy_: yes.. because we mention the entry points in setup.cfg
17:21:31 <LouisF> bouthors_: hi
17:21:42 <cathy_> bouthors_: hi
17:21:43 <bouthors_> hi
17:21:47 <vikram> cathy_: so when networking-sfc is installed entry-points will be installed
17:22:05 <vikram> https://review.openstack.org/#/c/210008/7/setup.cfg
17:22:15 <cathy_> for service chain plugin and common driver manager part, there is no dependency. So we are good for this part too.
17:22:23 <vikram> [entry_points]	28
17:22:23 <vikram> neutronclient.extension=	29
17:22:24 <vikram> flow_classifier = networking_sfc.neutronclient.flow_classifier	30
17:22:24 <vikram> port_chain = networking_sfc.neutronclient.port_chain	31
17:22:24 <vikram> port_pair = networking_sfc.neutronclient.port_pair	32
17:22:24 <vikram> port_pair_group = networking_sfc.neutronclient.port_pair_group	33
17:22:26 <vikram> 
17:23:01 <vikram> cathy_: We need the server code for verifying client + horizon code
17:23:24 <vikram> cathy_: We cannot find db migration files in the current review links
17:23:37 <cathy_> The API and DB codes are already posted.
17:23:52 <LouisF> vikram: we will be adding the migration files
17:24:11 <vikram> LouisF: ok.. once added can you please notify..
17:24:21 <vikram> LouisF: we will start our testing
17:24:24 <LouisF> vikram: tomorrow
17:24:30 <vikram> LouisF: great
17:24:44 <mohankumar_> LouisF: that great
17:25:02 <LouisF> we will also be posting unit tests for db and extensions
17:25:41 <pcarver> Ok, so it sounds like the only other thing is how the OvS agent additions work
17:26:22 <vikram> pcarver: do you have the link handy?
17:26:34 <pcarver> We'll be adding code that will add flow table entries to OvS. This code needs to be called by the OvS agent.
17:26:34 <vikram> pcarver: i mean the code changes link?
17:26:56 <pcarver> vikram: I'm not sure there's a review up yet with the OvS agent changes
17:27:24 <pcarver> cathy_:, LouisF: have you got any OvS agent code?
17:27:38 <vikram> pcarver: ok.. as you have mentioned then we might need ofagent to be modified
17:27:50 <cathy_> OVS driver does not have dependency, so we are good. But for OVS agent, we need to add a small amount of codes in existing OVS agent
17:28:03 <LouisF> pcarver: we hope to post ovs agent code soon
17:28:20 <vikram> cathy_, LouisF: do you need the code in neutron repo?
17:28:22 <pcarver> vikram: The actual guts of what networking-sfc does will be putting flows into OvS
17:28:52 <pcarver> All the work so far is just leading up to figuring out what flows to put into OvS
17:29:24 <pcarver> All the API and neutronclient and DB ultimately is there to drive OvS changes
17:29:28 <Anh-NetCracker> Hi, I have a question. How would OVS work with other hardware based ̣network controller?
17:29:34 <pcarver> At least for the initial reference implementation anyway
17:29:38 <cathy_> vikram: we need to make a change to neutron OVS agent.
17:29:54 <LouisF> vikram: there more need to be change to ovs_neutron_agent.py
17:29:58 <cathy_> We will need to get that small patch approved and merged.
17:30:03 <pcarver> Anh-NetCracker: I spoke to broadly. The API and client and DB will support multiple backends
17:30:24 <vikram> cathy_: oh.. then we need to get those changes in.. right
17:30:26 <pcarver> but other backends aren't part of the initial work effort
17:30:56 <cathy_> vikram: yes
17:31:01 <vikram> cathy_: Liberty-3 is planned this week...
17:31:10 <pcarver> The OvS agent will be the initial backend that actually does something in response to the API etc
17:31:17 <vikram> cathy_: I don't know if the code will be allowed after this
17:31:30 <vikram> cathy_: May be we can confirm with armax
17:32:31 <cathy_> vikram: sure.
17:33:05 <LouisF> vikram: this would be similar to how OVSDVRNeutronAgent is activated on ovs_neutron_agent.py
17:33:58 <vikram> LouisF: Ok.. I am not aware with OVS Agent code :(
17:34:26 <vikram> need to read :)
17:34:35 <LouisF> vikram: ok
17:36:06 <vikram> Sorry, need to go for dinner.. My family is waiting ..
17:36:12 <vikram> bye
17:36:39 <cathy_> vikram: thanks! bye
17:36:56 <pcarver> vikram: Thanks
17:37:44 <cathy_> we will explore to see whether there is another way to avoid any change in existing Neutron OVS agent
17:38:30 <cathy_> If anyone has any idea on this, that will be great.
17:38:55 <cathy_> now let's go to horizon work.
17:39:05 <cathy_> mohankumar_: how is the horizon codes going?
17:39:57 <pcarver> I thought I read somewhere that Horizon basically "discovers" stuff by using the neutronclient. I haven't looked into how that actually works.
17:40:29 <cathy_> pcarver: horizon provides the user interface, AFAIK
17:40:34 <LouisF> pcarver: need to design the panels and tabs
17:40:36 <cathy_> we need the horizon code for the demo
17:40:45 <cathy_> mohankumar_: are you there?
17:41:40 <cathy_> mohankumar_: I know it is midnight time for you:-)
17:41:51 <cathy_> OK, let's move to the testing
17:42:44 <cathy_> what tests should we cover: unit test for each part, tempest, integration test. Anything else?
17:45:26 <cathy_> OK, let's talk about the spec and code review.
17:46:34 <cathy_> The API spec has been updated many versions to incorporate comments. If you have more comments, could you please post them by this week? I will upload a new version incorporate Armando's comment on the sequence of commands
17:48:30 <cathy_> For the System and Design Flow spec, I will post another version to clarify Armando's question/comment today. Also please give your final review comments if any.
17:49:54 <cathy_> Same for the other spec and codes. If you have more comments, it will be good to post them this week. We would like to get them merged and move forward.
17:50:34 <cathy_> that's all from my side. Anything else you would like to discuss?
17:51:39 <cathy_> OK, then we can end the meeting earlier today:-)
17:52:02 <cathy_> bye everyone, talk to you in next meeting.
17:52:09 <s3wong> bye
17:52:11 <Brian__> bye
17:52:12 <LouisF> bye
17:52:16 <pcarver> bye
17:52:33 <cathy_> #endmeeting