17:05:32 <Sukhdev> #startmeeting networking_l2gw
17:05:32 <openstack> Meeting started Mon May 11 17:05:32 2015 UTC and is due to finish in 60 minutes.  The chair is Sukhdev. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:05:33 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
17:05:36 <openstack> The meeting name has been set to 'networking_l2gw'
17:05:46 <Sukhdev> welcome everybody
17:05:54 <Sukhdev> #topic: Agenda
17:06:06 <Sukhdev> #link: https://wiki.openstack.org/wiki/Meetings/L2Gateway
17:06:20 <Sukhdev> Anybody want to add/modify the agenda?
17:06:44 <Sukhdev> #topic: Announcements
17:07:07 <Sukhdev> Lets cancel next meeting - everybody will be tired after summit and travels
17:07:22 <Sukhdev> We will resume on June 8th
17:07:26 <marutikamat> sure
17:07:30 <Sukhdev> Is everybody OK with that?
17:07:46 <armax> yes, it should be okay
17:07:52 <Sukhdev> cool
17:08:05 <Sukhdev> #topic: Summit Presentation planning
17:08:17 <Sukhdev> We had a bit of confusion with the slide deck
17:08:22 <armax> yes
17:08:33 <Sukhdev> I thought, like armax, that we were going to work with google slides
17:08:52 <Sukhdev> and then I saw one from marutikamat, and I liked it much better and started to work on it :-)
17:08:53 <armax> at this point it’s probably the best course of action
17:09:24 <armax> so do whe know what is the latest and greatest?
17:09:41 <selvakumar> Sukhdev : perhaps myself and Maruti had missed the armax comments in google drive
17:09:41 <marutikamat> The one I sent today evening
17:09:43 <armax> I made a bunch of comments on one slidedeck
17:09:51 <Sukhdev> Latest deck from marutikamat looks pretty good - I saw armax edits as well - those all seem to be covered by marutikamat's version
17:10:25 <Sukhdev> well, may be not everything -but, mostly :-)
17:10:26 <selvakumar> Maruti: is armax comments addressed  .. just wanted to confirm
17:10:26 <marutikamat> Today, I was checking for comments by Armando, but could not see in the google drive.
17:10:28 <armax> Sukhdev: not taht I recall…I made a suggestion to remove the mpls case, but I could still see it in the latest deck
17:10:47 <Sukhdev> armax:Oh I missed that one - sorry
17:10:49 <armax> also the CLI
17:11:03 <armax> no-one is gonna be able to read that much text on the screen
17:11:12 <Sukhdev> I was on the edge on the CLI - since, it is there, I left it
17:11:24 <armax> I am happy to make the edits myself so long as we don’t stomp on each other
17:11:44 <marutikamat> Please go ahead armax
17:12:01 <armax> so if we agree which draft is the latest edit, and that we have one editor at any given time, then we should be good
17:12:07 <armax> marutikamat: ok
17:12:16 <armax> should we go back to the gdocs form?
17:12:18 <Sukhdev> armax: in that case work with marutikamat's latest deck
17:12:25 <armax> so that we avoid passing ppt back and forth?
17:12:39 <Sukhdev> armax: I think google is the best way to go
17:12:43 <armax> Sukhdev: yes, me too
17:13:17 <marutikamat> armax: can you please edit as you like and upload on the google drive. Then, we can edit from there if required further.
17:13:25 <Sukhdev> armax: Once, finalized, we can move the final to ppt - it may be better to present with ppt (if there are animations)
17:14:10 <armax> we can agree on the ppt at the summit itself
17:14:37 <armax> ok, let me have a stub at moving the ppt to gdocs again, I’ll use the same url that we had
17:14:52 <armax> '
17:14:54 <Sukhdev> armax: cool - thanks
17:15:02 <marutikamat> Myself and Sukhdev are to provide demo links.
17:15:02 <armax> I should have something by the end of the PST morning
17:15:11 <Sukhdev> marutikamat selvakumar: Do you guys have demo?
17:15:22 <marutikamat> I can send the demo link in an e-mail.
17:15:36 <Sukhdev> marutikamat: How long is the demo?
17:15:43 <marutikamat> I will send by Wednesday as tomorrow is a team offsite
17:16:29 <marutikamat> It may be just 5 minutes. I am not going to show installation. Just creating gateway, connections and ping between VM and BMs
17:16:30 <Sukhdev> I should have it by then as well
17:16:31 <selvakumar> Sukhdev: Ashish and Alok are working on this
17:16:41 <selvakumar> I think they can give the update if they are here
17:17:02 <Sukhdev> marutikamat: mine is 11 minutes - I will look at cutting it short :-)
17:17:21 <armax> it can’t be more than 3 minutes :)
17:17:29 <marutikamat> Sukhdev, as we are just providing the links in the slide, 11 minutes are also OK
17:17:49 <armax> we can provide an extended version for offline browsing
17:17:57 <armax> if we intend to play the demo on stage
17:18:06 <armax> it can’t be more than 2-3 minutes
17:18:15 <armax> unless the intention is to get people to fall asleep
17:18:41 <Sukhdev> armax: we thought we will not play demo - instead take Q&A
17:18:59 <Sukhdev> just point folks to the link and let them watch on their own
17:19:05 <armax> sounds good
17:19:17 <armax> 11 minutes sounds like a lot of time though
17:19:39 <Sukhdev> Next point about presentation - who is presenting what?
17:19:40 <armax> I guess it depends on the video script
17:19:58 <armax> Sukhdev: I think we need to meet at the summit and have a live dry run
17:20:04 <armax> we can decide who does what then
17:20:29 <armax> thoughts?
17:20:42 <Sukhdev> armax: I give background of L2 GW (from the L2 GW team point of view), describe API, and then the HW setup and, then demo - hence long ---many people do not know about L2 GW - hence, I put everything there :-):-)
17:20:42 <marutikamat_> I got disconnected
17:20:47 <selvakumar> I agree with both armax and Sukhdev
17:21:10 <armax> Sukhdev: so are you presenting on your own?
17:21:11 <armax> :)
17:21:18 <marutikamat_> :)
17:21:21 <armax> that’s pretty much everything that needs to be said!
17:21:47 <Sukhdev> armax: no - I was expelling why my demo is long :-)
17:21:56 <marutikamat_> oh ok
17:21:58 <Sukhdev> explaining
17:22:02 <armax> I’d say let’s get to a final good draft of the ppt and then figure out who talks on what
17:22:12 <armax> Sukhdev: oh, I see :)
17:22:16 <armax> Sukhdev: gotcha
17:22:20 <marutikamat_> :)
17:22:50 <Sukhdev> I actually prepared the demo for Arista internal folks (who have no clue about L2GW)
17:22:57 <armax> Sukhdev: makes sense
17:23:16 <armax> Sukhdev: so it’s not just demo per se
17:23:22 <armax> Sukhdev: it’s like the whole deal
17:23:35 <Sukhdev> armax: kinda
17:23:57 <Sukhdev> but, for the demo for Neutron, I will try to bring it down a bit
17:24:27 <Sukhdev> Anyway - when can we meet at the summit to hash out the final details
17:24:37 <marutikamat_> Is there a place/time where we can meet during the summit?
17:24:43 <armax> yes I think it’s best
17:25:05 <marutikamat_> meaning what means shall we use?
17:25:20 <marutikamat_> to communicate
17:25:23 <Sukhdev> marutikamat_: I am not sure if there is a place -
17:25:39 <marutikamat_> would whatsapp make sense?
17:25:44 <Sukhdev> I am at Hyatt - perhaps we can meet over breakfast
17:25:46 <marutikamat_> to communicate
17:26:00 <armax> I am sure there will be rooms we could use
17:26:09 <armax> if not, we could all meet and get lunch together
17:26:16 <armax> and so sit all at the same table
17:26:26 <marutikamat_> that is good
17:26:38 <Sukhdev> armax: which day do we want to get together?
17:27:02 <Sukhdev> armax: we have a panel on Monday, after lunch - which may be tight
17:27:11 <armax> sooner the better
17:27:49 <Sukhdev> armax: I am willing to skip part of keynote on Monday to work on this - if you all are game for it?
17:27:50 <marutikamat_> Can we meet in the evening on Monday itself?
17:28:16 <armax> another option is to try and meet for Monday’s breakfast
17:28:21 <armax> and then we can devise a plan
17:28:39 <Sukhdev> armax: I like the breakfast idea better -
17:28:49 <csd> Sukhdev if you've written anything on the demo setup that you can share let me know. I tried unsuccessfully to get a test setup running a couple of weeks back I could take that as a trial - I'll do third party validation :)
17:29:24 <Sukhdev> csd: I made one slide showing the setup - rest all is talk
17:29:30 <Sukhdev> I can share the slide with you
17:29:54 <Sukhdev> csd: here is the slide - https://docs.google.com/presentation/d/1AtufuVFBhgCzE6In68IqRq1q4U-6Sf5QVZmTgRKpM4Q/edit#slide=id.g63620438_0_22
17:30:45 <csd> Sukhdev thanks. sent request for access.
17:31:38 <Sukhdev> csd: Oh - this is Arista Internal site - let me copy it over to my personal drive and send it to you -
17:31:54 <Sukhdev> csd: stay after the meeting, I will send it to you
17:32:13 <marutikamat_> Please include me Sukhdev
17:32:17 <Sukhdev> going back to the plan for breakfast -
17:32:19 <csd> also - our marketing partner has office a couple of blocks away from convention center. If there's a need for a meeting room - I can check with them if I can borrow it.
17:32:22 <Sukhdev> marutikamat_: will do
17:32:58 <armax> csd: it’s good to have that option, thank you!
17:33:12 <armax> Sukhdev: btw I do have an announcement to make
17:33:18 <armax> Sukhdev: totally forgot
17:33:27 <Sukhdev> armax: please do
17:33:49 <armax> l2gw is ‘officially’ a neutron project since last night
17:33:50 <armax> https://review.openstack.org/#/c/179454/
17:34:06 <armax> in due course the stackforge repo will transition under the openstack namespace
17:34:15 <marutikamat_> Hurray, this is great news, armax
17:34:27 <Sukhdev> armax: Wow!! that is a good news
17:34:38 <selvakumar> good news armax :)
17:34:46 <armax> perseverance pays off!
17:35:31 <Sukhdev> armax: thanks for sharing this - awesome!!
17:35:53 <armax> Sukhdev: nothing really changes…but it’s a little milestone
17:36:04 <Sukhdev> Did we agree on where and when are we meeting for breakfast?
17:36:22 <Sukhdev> armax: small steps take you to the destination
17:37:43 <armax> Sukhdev: I was thinking at the conference venue
17:37:56 <Sukhdev> armax: we may not breakfast there -
17:38:07 <armax> Sukhdev: it’s likely we’ll be staying in different hotels
17:38:12 <Sukhdev> which hotels are we staying at?
17:38:13 <armax> Sukhdev: I am sure there is going to be!
17:38:33 <Sukhdev> armax: What time?
17:39:13 <armax> 8am?
17:39:16 <Sukhdev> armax: If we are not coordinated about it, we may be running around like chicken without head :-)
17:39:25 <marutikamat_> I am staying at Auberge
17:40:00 <armax> Sukhdev: I’d say let’s meet at 8am close to the registration desk
17:40:19 <armax> Sukhdev: we should be able to find each other…how hard can it be?
17:40:32 <Sukhdev> armax: I am good with this - everybody OK with that?
17:40:49 <Sukhdev> I have pre-paid breakfast at Hyatt - but, I can skip it :-)
17:40:53 <marutikamat_> 8.00am at registration desk looks good to me
17:41:01 <selvakumar> even me too
17:41:01 <armax> this is just the first rendez-vous, we might be able to meet again at lunch the same day
17:41:12 <armax> or at the end of the day when things go quiet again
17:41:35 <armax> presentation is on Wed afternoon
17:41:53 <armax> so we’ll have a few opportunities to get together to get together
17:42:17 <Sukhdev> armax: for some reason I thought it was on Tuesday
17:42:33 <armax> http://libertydesignsummit.sched.org/event/dc377389f3aad560b5bb0a7b666457a1#.VVDqAmRViko
17:42:41 <armax> it looks like it’s May 20 4.30pm
17:43:42 <Sukhdev> In that case we do not need to rush for Monday 8am
17:43:58 <armax> well...
17:44:06 <Sukhdev> but, since we decided on 8am - lets stay with this
17:44:08 <armax> it’s better to be conservative….the summit may get busy
17:44:18 <armax> and it’s better to be safe than sorry
17:44:47 <Sukhdev> So, we have a consensus to meet near registration desk on Monday at 8am
17:44:51 <marutikamat_> I agree
17:45:01 <Sukhdev> Please look for each other
17:45:15 <Sukhdev> I have couple of additional items on the agenda - shall we move on?
17:45:25 <armax> yes
17:45:47 <Sukhdev> BTW, I am always on Google hangout  (sukhdev@arista.com) - you can ping me there, if needed
17:46:13 <Sukhdev> #topic: Release of L2GW
17:46:37 <Sukhdev> I have been dragging my feet to cut the release - wanted to get as many bugs fixed as possible
17:46:47 <Sukhdev> I have been busy approving code - so that it merges
17:46:56 <armax> I think you should cut it and then backport as we see fit
17:47:22 <Sukhdev> I plan on tagging the release either today or tomorrow - are we all OK?
17:47:30 <armax> aye
17:47:35 <Sukhdev> this means it will be available on pypi
17:47:49 <Sukhdev> and all bugs need to be back ported
17:48:12 <armax> Sukhdev: if you could make a note of the process to share with the rest of the team that would be good
17:48:27 <Sukhdev> there are two bugs very close to approval - can you please close them
17:48:30 <armax> so that we can rotate the responsability amongst the members of the release team
17:48:43 <armax> and no-one is overloaded
17:48:43 <marutikamat_> selvakumar what is your thought?
17:48:57 <armax> what are the two bugs?
17:49:23 <Sukhdev> armax: good point - shall I document it on the wiki
17:49:37 <armax> Sukhdev: the wiki sounds like a good idea
17:49:59 <selvakumar> maruti : sukhdev: let me know what are those two critical defects so that we can close ASAP
17:50:10 <armax> Sukhdev: it does not need to be polished, even the plain list of commands would do
17:50:15 <Sukhdev> armax: you are holding one, and I am holding one for comments
17:50:24 <Sukhdev> let me look for the links -
17:50:28 <armax> am I?
17:50:37 <marutikamat_> https://review.openstack.org/#/c/181860/ is this the one?
17:50:58 <Sukhdev> https://review.openstack.org/#/c/167648/
17:51:32 <marutikamat_> that is not required
17:51:38 <armax> one is in conflict
17:51:40 <marutikamat_> I am going to abandon it.
17:51:42 <armax> the other looks like it breaks the ci
17:51:50 <Sukhdev> https://review.openstack.org/#/c/181361/
17:52:25 <Sukhdev> these two links that I provided are almost ready to go
17:52:37 <marutikamat_> 0 seg id is not supported on HP switches. As it works with Arista, we thought of allowing it.
17:52:45 <selvakumar> Sukhdev : regarding https://review.openstack.org/#/c/181361/  just wanted to confirm from Arista side
17:52:47 <marutikamat_> It is ready to go
17:53:02 <selvakumar> I agree with marui:
17:53:41 <Sukhdev> marutikamat_: so, HP switches do not allow untagged packets to the hosts?
17:53:56 <marutikamat_> On safe side, if we cut the branch on Wednesday your time, would it be too late?
17:54:31 <marutikamat_> selvakumar can you confirm what Alok mentioned?
17:54:34 <Sukhdev> marutikamat_: there are two use cases, where BM will run by itself and in a second case, BM may run multiple tenants
17:54:35 <armax> I honestly don’t think that cutting a not-so-stable branch is a big deal at teh moment
17:55:02 <armax> we don’t have yet many users of this API and we can cherry pick back to stable easily
17:55:13 <Sukhdev> marutikamat_: Wednesday is not going to be too bad - I can wait until then
17:55:13 <marutikamat_> ok
17:55:17 <selvakumar> Sukhdev: just confirm from your side for the defect by pulling the code locally  as HP switch didn't allow
17:55:43 <selvakumar> I will reconfirm once again on wednesday with Alok and Ashish
17:55:57 <Sukhdev> selvakumar: I will test it and let you know - did you verify the OVSDB dump - does it match with the one I posted on the bug?
17:55:59 <marutikamat_> HP switch does not allow physical port binding with 0 vlan ID
17:56:53 <Sukhdev> marutikamat_: actually, we use 0 to indicate that this is an access port as opposed to trunk port (hence untagged packts)
17:57:24 <marutikamat_> For the time being, let's allow 0 seg Id in the API
17:57:27 <Sukhdev> marutikamat_: this is more of a property of BM server - i.e. what kind of services they are running….
17:57:52 <Sukhdev> Cool - I will approve that patch and then test it….will let you know my findings
17:58:06 <Sukhdev> OK - one last thing -
17:58:11 <marutikamat_> Got it.
17:58:16 <selvakumar> I have tested on non native OVSDB case , in that case dump is same as what Sukhdev provided
17:58:38 <selvakumar> hopefully I belive Arista switch also it will work
17:58:39 <Sukhdev> I notice the flood list is not pushed to HW Vtep - this is a bit of problematic
17:58:53 <Sukhdev> selvakumar: will test and report back
17:59:31 <Sukhdev> Since the flood list is not pushed to the HW veep, I have to add arp entry on BM to make it ping VM
17:59:31 <selvakumar> Thanks Sukjdev:
18:00:01 <Sukhdev> Once GW connection is created, VM can initiate ping, but BM can not unless arp entry is added
18:00:13 <marutikamat_> Sukhdev: Can you explain the flood list?
18:00:29 <Sukhdev> But, once VM pings BM, the entry is populated and then BM can ping
18:00:39 <Sukhdev> so, VM has to always initiate the first ping
18:01:02 <marutikamat_> Sukhdev, let me check with the team tomorrow, and let you know
18:01:17 <selvakumar> you mean ucast_macs_local entry for the BM Sukhdev:
18:01:48 <Sukhdev> marutikamat_: If you look at the OVSDB dump, you will notice, Arista publishes the flood list for SW vtep
18:02:13 <marutikamat_> selvakumar: Sukhdev means ARP table of the Ovs switch
18:02:25 <Sukhdev> marutikamat_ +1
18:03:10 <marutikamat_> This list is populated as part of Mcast_Macs_Remote?
18:03:18 <marutikamat_> by the switch?
18:03:30 <Sukhdev> selvakumar: I can send you the OVSDB dump to show what I mean -
18:03:45 <selvakumar> ok Sukhdev:
18:03:58 <Sukhdev> we are past our time -
18:04:11 <Sukhdev> Thanks folks - this was very productive meeting
18:04:23 <Sukhdev> #endmeeting