17:04:38 <Sukhdev> #startmeeting: networking_l2gw 17:04:38 <openstack> Meeting started Mon Jan 18 17:04:38 2016 UTC and is due to finish in 60 minutes. The chair is Sukhdev. Information about MeetBot at http://wiki.debian.org/MeetBot. 17:04:39 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 17:04:43 <openstack> The meeting name has been set to '__networking_l2gw' 17:05:20 <Sukhdev> #topic: Agenda 17:05:25 <Sukhdev> #link: https://wiki.openstack.org/wiki/Meetings/L2Gateway#Meeting_January_18.2C_2016 17:06:02 <Sukhdev> oferby : So, I worked with armax to get the gate issue resolved 17:06:34 <Sukhdev> now those issues are resolved and we can get the patches merged 17:06:57 <Sukhdev> We were able to approve and merge several patches 17:07:05 <[1]oferby> cool. I finished the internal work and I'm also ready to merge 17:07:38 <Sukhdev> excellent 17:08:01 <[1]oferby> there are 2 groups that are interested in the inter-cloud connection 17:08:21 <Sukhdev> who are those? 17:08:22 <[1]oferby> one is Midokura and the other is another group in Huawei 17:08:54 <[1]oferby> Irena from Midokura has added a bug (feature) on the new API 17:09:00 <[1]oferby> did you see it? 17:09:12 <Sukhdev> I bet there are more people as well - I was discussing this use case in one of the meetings and those people wanted it as well :-):-) 17:09:26 <[1]oferby> :-) good to hear 17:10:07 <Sukhdev> what is the bug number 17:10:13 <Sukhdev> our list is large 17:10:18 <[1]oferby> I woulld like to start the work of official API proposal, although I completed the code for internal use ... is that OK by you? 17:10:27 <[1]oferby> sec 17:11:45 <oferby> https://bugs.launchpad.net/networking-l2gw/+bug/1529863 17:11:46 <openstack> Launchpad bug 1529863 in networking-l2gw "[RFE] extend l2gw api to support multi-site connectivity" [Undecided,New] 17:12:53 <Sukhdev> #link: https://bugs.launchpad.net/networking-l2gw/+bug/1529863 17:12:54 <openstack> Launchpad bug 1529863 in networking-l2gw "[RFE] extend l2gw api to support multi-site connectivity" [Undecided,New] 17:13:15 <Sukhdev> oferby : Thanks - I will review it and add remarks 17:13:42 <Sukhdev> oferby : going back to your other comment - about official API proposal 17:14:07 <Sukhdev> Yes, if you could write up something, that would help 17:14:28 <oferby> sure, I will do that tomorrow. 17:15:07 <Sukhdev> once you have it written it up, please let me know so that I can have the rest of team review it 17:15:41 <oferby> sure. I also want to bring up the issue of seperating the API from the implementation 17:15:59 <oferby> something similar to MLw 17:16:02 <oferby> ML2 17:16:29 <oferby> this will work for other teams that are looking for the API but would like to have their own impl 17:16:36 <oferby> what do you think? 17:17:12 <Sukhdev> maruti had started work on that - where folks can use the mail L2GW API and have their own back-ends 17:17:19 <Sukhdev> have you looked at that patch? 17:17:40 <oferby> no. wasn't aware of it .... 17:17:48 <oferby> can you send a link? 17:18:58 * Sukhdev looking 17:19:45 <Sukhdev> I think this is the one - https://review.openstack.org/#/c/206638/ 17:20:30 <oferby> ok. I will check it out. 17:21:35 <Sukhdev> He wanted to make sure south bound framework is available so that vendors can have their own back-end implementations 17:22:43 <oferby> this is what we want. I will be happy to join the effort here too. 17:23:19 <Sukhdev> review the patch and see if this helps 17:23:36 <Sukhdev> this will give you a starting point at least 17:23:45 <oferby> sure 17:24:17 <Sukhdev> I met with armax last week and we discussed about the work you are doing - we are looking forward to it 17:24:40 <Sukhdev> as soon as you have written it up, send it over 17:25:07 <oferby> I think I will be able to send out the patch tomorrow 17:25:25 <oferby> if nothing will wait for me in the office .......... 17:25:47 <Sukhdev> oferby : my suggestion will be to make sure it has sufficient documentation with it 17:26:21 <Sukhdev> sometimes it is hard to figure out the design from the code - it is better done through documents 17:26:24 <oferby> Irena from Midokura suggested that she will do the documantation 17:26:49 <oferby> she is in a business trip and will join the effort from next week. 17:27:05 <Sukhdev> you may want to post the document along with the patch 17:27:15 <Sukhdev> next week is fine - no worries 17:27:28 <oferby> cool. 17:27:52 <oferby> I have another issue. 17:28:08 <Sukhdev> sure 17:28:29 <oferby> I would like to put the HARDWARE VTEP schema file in our project instead of in OVS 17:29:19 <Sukhdev> I am not sure if I understand 17:29:21 <oferby> it is for the use of the L2GW anyway and I see no reason why we need to ask OVS people for permission to modify the schema whan we are the one to use it. 17:29:24 <oferby> agree? 17:29:52 <Sukhdev> Oh I see - 17:30:31 <Sukhdev> Well, you mean you want to keep a snapshot of it locally - 17:30:52 <Sukhdev> ultimately, it is owned and driven by the ovsdb 17:31:24 <oferby> yes. also there is a small implemantation of hardware emulator which also need to be hosted in L2GW. 17:32:14 <oferby> no other project is using the hardware vtep schema. it was built for the L2GW. it is just a schema, not the DB itself 17:33:05 <oferby> anyone can have their own schema and I see no reason why we should do cross-project work only to add one column in the db, as I did. 17:33:44 <Sukhdev> You had a patch for the IDL support 17:34:12 <oferby> the IDL is owned by OVS because other project are using it 17:34:35 <oferby> no one other than L2GW is using the HARDWARE VTEP schema 17:35:04 <Sukhdev> you mean in the opentack umbrella, right? 17:35:23 <oferby> yes 17:35:32 <Sukhdev> I know many others using HW VTEP schema :-):) 17:35:40 <Sukhdev> correct 17:36:49 <oferby> you mean other projects, not L2GW? 17:36:52 <Sukhdev> I think this should be alright 17:37:14 <Sukhdev> I meant other vendors 17:37:25 <Sukhdev> who use ovsdb 17:37:32 <oferby> right ... still is it L2GW. 17:37:58 <Sukhdev> I am not sure if OVN project uses it or not - but, they want to use L2GW API for it 17:38:25 <oferby> using the API does not mean you have to use OVSDB 17:38:44 <Sukhdev> I mean they use API as well OVSDB 17:38:52 <oferby> actually, Midokura wants the API but they use their own db 17:38:58 <Sukhdev> OVN 17:39:37 <oferby> in case others will want to use the API and teh DB, even better, we must own our db schema. 17:40:01 <Sukhdev> that is why it is so important to have this documented - so that it is easy to understand 17:40:13 <oferby> sure. 17:40:26 <Sukhdev> once you post the document and patch, then we can further detailed discussion 17:40:27 <oferby> we will doc all the changes. 17:40:34 <oferby> sure. 17:40:39 <Sukhdev> sounds good 17:40:42 <Sukhdev> anything else? 17:40:46 <oferby> nop. 17:41:06 <Sukhdev> cool - will look forward to a note from you. 17:41:14 <Sukhdev> Thanks for your time 17:41:20 <oferby> thank you. 17:41:22 <Sukhdev> we will meet later 17:41:25 <Sukhdev> bye 17:41:28 <oferby> bye. 17:41:30 <Sukhdev> #endmeeting