17:01:44 <Sukhdev> #startmeeting networking_l2gw 17:01:47 <openstack> Meeting started Mon Mar 14 17:01:44 2016 UTC and is due to finish in 60 minutes. The chair is Sukhdev. Information about MeetBot at http://wiki.debian.org/MeetBot. 17:01:48 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 17:01:50 <openstack> The meeting name has been set to 'networking_l2gw' 17:02:20 <Sukhdev> #topic: Announcements 17:02:35 <Sukhdev> oferby: Hi 17:02:50 <oferby> Hi 17:02:50 <Sukhdev> anybody from India team here? 17:03:27 <Sukhdev> #topic: Agenda 17:03:34 <Sukhdev> #link: https://wiki.openstack.org/wiki/Meetings/L2Gateway#Meeting_March_14.2C_2016 17:03:44 <Sukhdev> oferby : I see you have been busy 17:03:52 <oferby> :-) yes I was. 17:04:00 <Sukhdev> #topic: patches under review 17:04:25 <Sukhdev> #link: https://review.openstack.org/#/q/status:open+project:openstack/networking-l2gw,n,z 17:04:49 <Sukhdev> I approved some of the patches - some are in merge conflict 17:05:06 <oferby> one of mine? 17:05:54 <Sukhdev> oferby : no I do not think yours is in merge conflict - but, it is failing the tests 17:07:00 <Sukhdev> #topic: L2GW cloud inter-connect 17:07:14 <Sukhdev> oferby : last week irenab was looking for you 17:07:22 <Sukhdev> irenab : are you here? 17:07:32 <oferby> yes 17:07:55 <oferby> we did not have meeting last week, did we? 17:08:13 <Sukhdev> no - it was the previous week 17:08:38 <oferby> anyway, I responded to all her comments 17:09:34 <Sukhdev> oferby : I did not have chance to review the latest version of the patch 17:09:46 <Sukhdev> I reviewed the previous version 17:10:23 <oferby> any comments? even on the previous one? 17:10:46 <Sukhdev> I wanted to understand the overall flow - 17:11:10 <Sukhdev> so, I was reviewing with the intent to understand the flow - 17:11:39 <Sukhdev> I will review the latest one later this week - 17:11:53 <Sukhdev> oferby : one question for you 17:12:01 <oferby> ? 17:12:23 <Sukhdev> any update/word on the ovslib change that you submitted? 17:12:36 <Sukhdev> I should post the link of that patch to this wiki 17:12:47 <Sukhdev> do you mind sharing that link? 17:13:03 <Sukhdev> oferby: and second question 17:13:34 <Sukhdev> well - I will hold off and let you respond to the first one :-) 17:15:44 <oferby> There are 2 patches to the ovs - 17:16:30 <oferby> first one is a change to the schema. This one was committed. 17:16:43 <Sukhdev> you mean it merged? 17:16:51 <oferby> yes 17:17:19 <oferby> The second one is the change I made to the IDL to support passive mode - 17:18:04 <Sukhdev> do you happen to have the links to both of these? 17:18:26 <oferby> sec, I will look for it 17:19:09 <oferby> http://openvswitch.org/pipermail/dev/2016-February/065667.html 17:19:25 <oferby> this is the schema change that was merged. 17:20:17 <Sukhdev> #link: http://openvswitch.org/pipermail/dev/2016-February/065667.html 17:20:24 <oferby> the second one which is the change to IDL to support passive mode is more problematic ... 17:20:49 <Sukhdev> oferby : How is it problematic? 17:21:45 <oferby> I run the code review with Russell Bryant and after he approved the code, Ben Pfaff said that he doesn't like the way I and Neutron people use the IDL ... 17:22:05 <oferby> He said that the IDL was intended to read the schema from a file 17:22:24 <oferby> the way Neutron and I use it is by reading it from the server 17:22:56 <Sukhdev> hmmm.... 17:23:29 <oferby> I'm trying to convince him that we don't want to have the local file to be able to work with the IDL as we can read it from teh schema 17:23:46 <oferby> read the schema from the server 17:24:09 <oferby> if he will not approve, we will have to use the vtep.schema file locally 17:24:26 <Sukhdev> oferby : can you dig up the link ? I wanted to add to this wiki - I was trying to find it and could not locate it - hence, I thought I will ask you :-) 17:24:41 <irenab> hi, I am here 17:24:59 <oferby> http://openvswitch.org/pipermail/dev/2016-February/065918.html 17:25:01 <oferby> hi irena 17:25:08 <irenab> hi oferby 17:25:18 <Sukhdev> irenab : welcome - I was telling oferby that you were looking for him last week 17:25:46 <irenab> I reviewed the spec and have only one last commment left with regards to the mac, ip db table 17:25:50 <Sukhdev> #link: http://openvswitch.org/pipermail/dev/2016-February/065918.html 17:25:55 <irenab> Sukhdev: thanks :-) 17:26:52 <irenab> We can follow up the discussion overt the patch if you prefer 17:27:21 <Sukhdev> irenab : hold that thought for a sec 17:27:27 <irenab> I think this can be resolved during the implementation, just wanted to get your and Sukhdev understandings 17:28:08 <Sukhdev> oferby : this link you posted is a ML thread, do you happen to have a link to the patch itself? 17:28:16 <irenab> I apology, but I have only few mins, but we can follow up on the patch 17:28:37 <oferby> this is the link - 17:29:04 <oferby> openvswitch manage their patches in mailing list... look at the bottom of the page 17:29:18 <Sukhdev> oferby : got it - thanks 17:29:27 <Sukhdev> irenab : sorry - go ahead, please 17:30:03 <irenab> I just had last comment on the Remote GW Spec regarding persistency of the mac, ip tuples 17:30:16 <irenab> otherwise I think Spec in the very good shape 17:30:42 <oferby> before talking about the new API, Sukhdev - are you OK with the status of the IDL change? 17:31:08 <irenab> I have to drop now, so lets continue on the patch discussion 17:31:34 <Sukhdev> oferby : I am going through the link to see the comments on the patch 17:31:49 <oferby> ok 17:31:50 <Sukhdev> I wonder if we have to switch to the file - how much of the work will that be? 17:32:02 <oferby> very minor 17:32:16 <oferby> it is just that we will have to keep a copy of the file 17:33:31 <oferby> Neutron do not do this. they connect to the server, read teh schema, close the connection and connect again using IDL. silly and will not work for passive mode - 17:33:42 <oferby> this is what I was trying to explain to Ben 17:34:25 <Sukhdev> hmmm - and we need this support in L2GW to keep it backward compatible 17:34:55 <Sukhdev> so, if we made this change, we could keep it local to L2GW repo 17:35:56 <oferby> we can use a local file for now, just to have the passive mode merged and then I can send a second patch to use it properly. this way we will be able to continue in L2GW 17:36:02 <oferby> what do you think? 17:36:21 <Sukhdev> that is a good idea 17:37:02 <oferby> ok. so I will change the code as Ben wanted hoping that this will produce a quick merge - 17:37:33 <oferby> than I will send a second patch while meanwhile we will be able to merge my code to L2GW 17:37:34 <Sukhdev> cool - thanks - can you add me to the email as well - this way I can keep in the loop 17:37:57 <oferby> I always put you as a requester - 17:38:03 <Sukhdev> oferby : that sounds good - thanks 17:38:04 <oferby> you should get mails 17:38:24 <oferby> look at the end of the patch text before the code 17:38:28 <oferby> your email is there 17:40:20 <Sukhdev> ahh... I see it now - thanks a lot 17:40:22 <Sukhdev> got it 17:41:22 <Sukhdev> oferby : now to my second question and I think irenab wants to discuss it as well - where does your L2GW patch stands? 17:41:58 <oferby> I believe irenab is talking about the API, right irena? 17:42:46 <oferby> It looks like we are done. only a minor comments from irena 17:43:21 <oferby> irenab are u here? 17:44:18 <Sukhdev> oferby : If I want to test your patch - can you give me specific steps to set up the test system 17:44:34 <Sukhdev> I would like to play with your patch and test it for you 17:45:15 <oferby> I can send you detailed test setup to your email. is this ok for you? 17:45:34 <Sukhdev> yup - that will be great 17:45:55 <oferby> this will include the L2GW, the new patch, OVSDB server and the VTEP emulator. 17:46:15 <Sukhdev> VTEP emulator is for what? 17:46:58 <oferby> I guess your switches does not support the new feature so you will have to use OVS for the test. 17:47:36 <Sukhdev> oferby: To test L2GW, I create L2GW on Arista Switch and use neutron to fire up couple of VxLAN based VMs - 17:47:58 <Sukhdev> then I use L2GW API to create gateway tunnel and test the functionality that way 17:48:32 <oferby> right, but you do not have the new HARDWARE VTEP schema - 17:48:34 <Sukhdev> Arista Switches run OVSDB server 17:49:00 <Sukhdev> correct - I need one 17:49:00 <oferby> so you can't create tunnels with different tunnel id - 17:49:34 <Sukhdev> Actually, if you can send me step-by-step instructions in the email, it will be helpful 17:50:10 <oferby> this is the change that was needed in the schema to support connecting to a remote neutron network using tunnel key that is different from the one that the local neutron is using 17:50:11 <Sukhdev> I will test it for you and let you know if I find any issues - this will help us get these cleaned up 17:50:24 <oferby> great 17:50:53 <oferby> I will send you detailed explanation on how to build a test setup. 17:50:56 <Sukhdev> oferby: I assume you are testing it as well 17:51:16 <Sukhdev> are you making sure that the backward compatibility is maintained? 17:51:25 <oferby> sure. we have a test setup that is working. 17:51:57 <oferby> for the compatibility issue - 17:52:11 <Sukhdev> in other words, while you are testing the new functionality, are you testing that the base L2GW functionality is not broken or impacted in any way? 17:52:47 <oferby> if you will not use the new API - all the current commands are not changed therefor everything will still work 17:53:12 <Sukhdev> excellent 17:53:34 <oferby> only if you will use the new APIs, you will need to support the new schema. I guess this is something that you guys will need to ass 17:53:37 <oferby> add 17:53:42 <oferby> right? 17:54:05 <Sukhdev> I guess so - 17:54:27 <Sukhdev> I assume you have a patch for new schema that I can apply to the server? 17:54:56 <oferby> sure. it is merged to the openvswitch github repo 17:55:08 <oferby> sec - I will send you a link 17:55:53 <oferby> https://github.com/openvswitch/ovs/tree/master/vtep 17:56:13 <Sukhdev> #link: https://github.com/openvswitch/ovs/tree/master/vtep 17:56:19 <oferby> that was the repo 17:56:23 <oferby> this is the file: 17:56:25 <oferby> https://github.com/openvswitch/ovs/blob/master/vtep/vtep.ovsschema 17:56:45 <oferby> the repo also have the emulator 17:56:50 <Sukhdev> #link: https://github.com/openvswitch/ovs/blob/master/vtep/vtep.ovsschema 17:57:48 <Sukhdev> what time zone are you in? 17:58:00 <Sukhdev> we have only 3 min left 17:58:23 <oferby> the file name of the emulator is ovs-vtep 17:58:34 <oferby> GMT +2 17:58:39 <Sukhdev> I will look for your email and will ask any questions, if I have, over the email 17:58:48 <oferby> sure 17:59:04 <Sukhdev> thanks oferby - this is great 17:59:21 <Sukhdev> I will sync up with armax as well and fill him in on this 17:59:29 <oferby> great 17:59:36 <Sukhdev> Thanks for the great work - 17:59:57 <Sukhdev> bye 18:00:00 <oferby> bye 18:00:05 <Sukhdev> #endmeeting