17:01:01 #startmeeting networking_l2gw 17:01:08 Meeting started Mon Jul 6 17:01:01 2015 UTC and is due to finish in 60 minutes. The chair is Sukhdev. Information about MeetBot at http://wiki.debian.org/MeetBot. 17:01:09 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 17:01:11 The meeting name has been set to 'networking_l2gw' 17:01:39 who is present here - can you shout, please? 17:02:06 armax: are you here? 17:02:11 yes 17:02:19 * regXboi is lurking/auditing 17:02:37 waiting for India team to join in 17:02:58 #topic: Agenda 17:03:04 #link: https://wiki.openstack.org/wiki/Meetings/L2Gateway 17:03:59 It is late for folks in India - anybody from India here? please shout... 17:04:10 Hi Sukhdev, I am here 17:04:23 marutikamat: cool - 17:04:25 How do you do? 17:05:05 marutikamat: coming out of long weekend - still sleepy :-) 17:05:16 :) 17:05:18 marutikamat: too much party :-) 17:05:27 Cool 17:05:35 lets get going with the meeting 17:05:40 sure 17:05:44 #topic: Announcements 17:05:52 I have no announcement today - 17:05:56 anybody has any? 17:06:05 marutikamat armax : anything? 17:06:10 sukhdev: yes 17:06:26 armax: please do 17:06:33 I personally applied a few changes to the repo 17:06:48 to simplify the integration with Neutorn 17:07:07 and fix a couple of issues that were caused by changes in openstack/common 17:07:22 last time I checked the repo was sane and external CI passed 17:07:30 one more thijng 17:07:51 there’s a change from mestery do relinquish control for tagging releases to the library release team 17:08:09 https://review.openstack.org/#/c/198750/ 17:08:11 armax: yes - I saw that one 17:08:15 https://review.openstack.org/#/c/198749/ 17:09:05 sukhdev: that’s it from me 17:09:05 Thanks armax for the info 17:09:32 armax: actually I was refereeing to https://review.openstack.org/#/c/193127/ 17:10:10 armax: I was going to ping mestery to understand the process to figure out - how the pypi versioning is going to work 17:10:24 Sukhdev: You have to use semantic versioning 17:10:33 And we'll keep all of this under the library-release team 17:10:33 Or it's gonna be nuts 17:10:37 And error prone, etc. 17:10:42 Part of being in the neutron stadium 17:11:05 mestery: any write up? 17:11:24 mestery: that we can refer to - to understand it a bit better? 17:12:17 sukhdev: there’s a ML thread about switching to semver 17:12:28 sukhdev: there was also a session during the summit 17:12:42 sukhdev: and related etherpad 17:12:54 sukhdev: not sure if both would be in the shape of your liking 17:12:59 armax: do you happen to have a link to any of these? 17:13:28 Sukhdev: I have patches to all the networking-foo which convert them to semvar 17:14:01 mestery: ah I see armax posted two of them here - 17:14:03 #link https://review.openstack.org/#/q/status:open+branch:master+topic:semver-releases,n,z 17:14:49 mestery: cool - thanks - let me spend some time reviewing/understanding them - will ping you if have any questions 17:14:58 Sukhdev: Ack 17:15:14 armax: thanks for the update 17:15:34 armax: as to the CI being happy - I have seen two patches sitting with the CI failing - 17:16:04 I have not approved them simply waiting for CI to pass on these - armax are you aware of these two? 17:16:06 sukhdev: which ones? 17:16:24 armax: hang on looking 17:16:51 #link - https://review.openstack.org/#/c/194921/ 17:17:13 #link: https://review.openstack.org/#/c/194919/ 17:17:56 sukhdev: a recheck should do 17:17:58 I’ll watch them 17:18:07 armax: cool - thanks 17:18:56 We kind of went out of order - but, was good :-) 17:19:02 lets get back to the agenda - 17:19:10 #topic: Action Items 17:19:13 sukhdev: sorry 17:19:26 armax: no worries - it was actually good 17:19:50 I had an action to work with armax to come up with an earlier slot to accomodate India team 17:20:19 I completely forgot about it - even though I met armax twice in last two weeks :-):-) 17:20:20 sukhdev: how early would it suffice? 17:20:22 shame on me :-) 17:20:31 marutikamat: please chime in 17:21:09 I wish more India folks were here - they asked in the last meeting 17:21:14 If we could reduce by 30 mins, it would be great 17:21:23 I mean advance by 30 mins 17:21:44 All of them are on vacation right now 17:21:57 we can do 9am PDT, but perhaps we’d have to change day 17:22:00 marutikamat: we will have to move to other day - I have Ironic meeting which runs from 9am to 10am 17:22:09 sukhdev: right 17:22:13 so, basically I have these two meetings back-to-back 17:22:35 Moving to some other day would also work. 17:22:38 marutikamat: everybody on vacation - wow!! I am moving to India :-) 17:23:18 Not every one are on vacation. The three developers (manjunath, phani and vikas) are on vacation 17:23:22 We could move it to Tuesday at 9am (PT) or Thursday at 9AM - 17:23:36 armax: any of these slots work for you? 17:23:37 thur 17:23:40 works better for me 17:24:09 marutikamat: you think that would work for India team? 17:24:15 That would be 9.30pm for India folks I think. Is that correct? 17:24:47 marutikamat: do not know about India time - it is one hour earlier - 17:25:01 is it 10:30 in India now? 17:25:20 approaching 11pm 17:25:24 Thursday 9am PST would be 9.30pm Thursday. And that should be fine with us. 17:25:59 Yes, right now 5 mins for 11.00pm 17:26:14 marutikamat: cool - let check the available channels. I hope I can find one - and will send out the announcement for next meeting 17:26:29 Thanks Sukhdev and armax 17:26:44 #topic: Bugs 17:26:56 I see tons of unassigned bugs - 17:27:43 #link: https://bugs.launchpad.net/networking-l2gw 17:28:30 marutikamat: you mentioned last time that you and selva have moved on to another project 17:28:47 Right now, we have only 3 resources assigned for the bugs. I will assign the owners tomorrow. 17:29:05 marutikamat: cool - 17:29:18 Yes, myself and Selva have been moved into some internal projects. 17:29:42 If time permits, I will contribute 17:30:16 marutikamat: cool - will you be joining for the meetings - now that we are changing the time accomodate you :-):-) 17:30:43 Yes, I will attend the meetings and I will request others also to join the meetings. 17:31:04 marutikamat: cool thanks 17:31:35 marutikamat: I had asked Ashwin to join us here this week to address the flood-list issue with L2 GW 17:31:54 this has been lingering issue for a while - 17:32:21 marutikamat: I thought if I put both of you in the meeting - it may be faster to get the issue resolved/understood 17:32:51 I think myself and Ashwin are on the same page. I wanted his help in configuring something on the Mcast_Macs_Remote table to ensure that ping from a bare metal server to a VM works. 17:33:22 he should be joining shortly 17:33:42 before we discuss this - (as he is figuring out how to join in) - 17:33:50 is there anything else we want to cover 17:33:51 That will boost the confidence level for the changes to be made in the code. 17:35:35 armax marutikamat : while Ashwin is trying figure out to join in - do we want to cover anything else? 17:35:47 sukhdev: nothing to add from my side 17:35:55 The biggest one in my mind was the unassigned bugs - 17:36:29 armax: the l2gw CI team has fixed a problem where the run for the older patch was not getting aborted when a new patch is uploaded. 17:36:59 I will assign the owners tomorrow for the unassigned bugs 17:37:21 marutikamat: Oh could that be the reason for CI failures for those two patches I mentioned earlier? 17:37:36 sukhdev: I don’t htink so 17:38:03 armax: ah OK - so something unrelated 17:38:38 sukhdev: yup, before the cahnge in configuration the CI would report on stale changes too 17:38:58 I will have someone to look at the build failures 17:39:06 armax: ah I see - thanks for clarification 17:39:23 Folks, BTW, ashwins is here now 17:39:27 let me introduce him 17:39:29 marutikamat: afaik the failures where py27 related 17:39:40 I might have solved them during the latest changes I pushed 17:39:48 Ashwin is our local expert in Arista for OVSDB related stuff 17:39:57 if there are failures to the tempest job, that’s a diffrerent can of worms 17:40:15 welcome ashwins 17:40:18 armax: Cool - I will watch those patches as well - and let you know - BTW, those are related to stable/kilo 17:40:18 Hey guys 17:40:30 Hi ashwin 17:40:34 sukhdev: ah! 17:40:35 thanks armax 17:40:54 hi marutikamat 17:40:59 armax marutikamat : I had asked ashwins to look into the flood list issue 17:41:05 sukhdek: ok so it must be something different 17:41:11 sukhdev: sweet 17:41:33 ashwins: can you explain your finding so that everybody is on the same page 17:42:14 ashwin, did you get a chance to create entries in the Mcast_Macs_Remote table using vtep-ctl and whether the BM is able to ping a VM? 17:42:28 marutikamat I haven't had a chance to try out what you mentioned the other day (ie adding the remote flood list in mcast_macs_remote manually on the hardware switch) 17:42:57 ok 17:43:05 will update after I'm done with that, but I'm fairly confident that it'll work 17:43:50 that is how we do our internal testing, but anyways will confirm just to be sure... 17:43:53 Great. I will make changes in the code to insert the records in mcast_macs_remote and physical_locator table then. 17:44:05 cool 17:44:07 marutikamat: assuming ashwins findings work, how big is the change on the L2GW side to fix this? 17:44:25 Just wanted to be sure before doing the code changes. 17:44:45 sure 17:45:08 It is not much. 17:45:30 marutikamat: sigh….good to hear this - 17:45:39 :-) 17:45:50 marutikamat: whenever you have the fix, let me know and we can test it 17:45:59 sure 17:46:14 marutikamat: just shoot me an email - it may be faster 17:46:29 Cool folks - with this we good with bugs 17:46:54 armax: wanted to ask you a question on the neutron api for L2GW 17:47:03 k 17:47:48 now that we are under neutron tent, we should consider using python client to change the 2GW CLI 17:47:57 there is a patch sitting in the list 17:48:22 sukhdev: the cli is extensible and can be plugged cli commands 17:48:30 sukhdev: we’d need someone who can get it sorted 17:49:20 sorry - got disconnected - back now 17:50:03 armax: did you get my question before I got disconnected? 17:50:21 yes 17:50:32 good - 17:50:35 I the cli is extensible and can be plugged cli commands 17:51:00 it’s my understanding that the author of the cli was unable to make use of it 17:51:06 sorry the author of the patch 17:51:18 sukhdev: so that patch would need a nudge by someone capable 17:51:27 to identify where the problem is 17:52:01 armax: I am not familiar with this stuff - otherwise would have jumped in 17:52:06 sukhdev: but to your point, yes, even outside the tent, it’s technically possible to integrate with the cli 17:52:20 sukhdev: me neither, and I would have to find time to look into it 17:52:29 sukhdev: but I have none so far 17:53:04 armax: I will spend some time looking at the patch to get some familiarity and see what I find 17:53:26 sukhdev: it’s only my backlog 17:53:36 if no-one beats me to it, I will eventually get to it 17:54:23 armax: :-) 17:54:52 I have exhausted the agenda - 17:55:01 anybody has anything want to discuss? 17:55:11 * Sukhdev waiting 17:55:37 I am helping ODL to make L2gw work with ODL 17:55:53 I am helping the ODL team to make L2gw work the ODL 17:56:05 marutikamat: interesting - 17:56:32 marutikamat: I thought they are full blown controller by themselves - how will L2GW fit in there? 17:56:35 armax is guiding me for this 17:56:52 am I? Interesting :) 17:56:56 they want our service plugin to be the NB APIs 17:57:27 but the switch and the OVSDB hardware_vtep schema will be managed by the ODL controller 17:57:31 marutikamat: Oh so, ODL will run as ML2 driver and use L2GW as service plugin - interesting!! 17:58:07 L2gw neutron will be a thin layer and it will invoke the odl plugin driver. 17:58:26 yes, sukhdev, something like that. 17:58:40 they don't want to use our l2gw agent. 17:58:53 marutikamat: so, instead of using l2gw agent - se ODL 17:59:01 s/se/use 17:59:07 yes, you are correct. 17:59:20 that makes sense - 18:00:00 In fact, in our presentation at Vancouver, I mentioned that anybody who do not want to use L2GW agent, they can implement their own agent - 18:00:08 so ODL is one such example - 18:00:22 Yes Sukhdev 18:00:52 marutikamat: if you have any document around that integration - it will be helpful for others as well - you may want to put it on our wiki 18:01:23 we are past our time - 18:01:35 Sukhdev: right now, it is not finalized. There was a meeting on Friday. However, it got cancelled due to the holiday in the US, 18:01:57 Once frozen, I shall update the wiki 18:02:19 marutikamat: cool - that would be good - keep it in mind - many others can leverage form it... 18:02:23 Thanks folks for attending - this was a good meeting/discussion 18:02:32 thanks everyone 18:02:34 #endmeeting