17:02:14 <Sukhdev> #startmeeting networking_l2gw
17:02:14 <openstack> Meeting started Mon Jun  6 17:02:14 2016 UTC and is due to finish in 60 minutes.  The chair is Sukhdev. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:02:15 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
17:02:17 <openstack> The meeting name has been set to 'networking_l2gw'
17:02:48 <Sukhdev> #topic: agenda
17:02:54 <Sukhdev> #link: https://wiki.openstack.org/wiki/Meetings/L2Gateway#Meeting_June_6.2C_2016
17:03:51 <Sukhdev> #topic: Announcements
17:04:09 <Sukhdev> I do not have any general announcemnt
17:05:30 <Sukhdev> armax and I have been discussing about L2GW project
17:05:51 <Sukhdev> we need additional contributors for this project
17:06:21 <Sukhdev> I have spoken to few additional people who have expressed interest
17:07:02 <Sukhdev> oferby_ : if you have folks who would want to participate as well, please encourage them
17:07:59 <oferby_> I spoke with armax today. I told him that I have another guy with me that can join and Midekura people also can but
17:08:26 <oferby_> I must say that I'm getting to be frustrated. I see no progress with what I try to put in
17:08:51 <oferby_> even the API which is blue print and not code is not being committed
17:09:33 <oferby_> I'm starting to think that maybe we cannot add what we want (inter cloud connection) as we thought we could.
17:09:39 <Sukhdev> oferby_ : understood and justified as well
17:10:21 <Sukhdev> oferby_ : we have two part issue - I am sure armax must have explained to you
17:10:36 <Sukhdev> 1) adding new features - that is what you want
17:10:55 <Sukhdev> 2) making the foundation of L2GW stable and sound
17:11:24 <Sukhdev> The problem which we are faced with is 2)
17:12:10 <Sukhdev> We believe we need to make the code base solid so that new features can be added
17:13:15 <Sukhdev> hence, we need some additional volunteers who will address 2) while you are working on 1)
17:13:31 <Sukhdev> oferby_ : does this make sense to you?
17:13:57 <oferby_> I see your point but there is no open bugs to reflect unstable code
17:14:02 <Sukhdev> And, we also want to add additional cores to this project - who can participate in the review
17:14:16 <Sukhdev> there are -
17:14:26 <oferby_> ???
17:14:31 <Sukhdev> for instance, we do not have any automated tests -
17:14:42 <oferby_> this is not a bug
17:14:52 <Sukhdev> we have not switched to using standard lib
17:15:05 <oferby_> again, not a bug
17:15:21 <Sukhdev> there are hacks in the code - it works, but, can be better
17:15:32 <Sukhdev> oferby_ : there are some bugs, I believe
17:15:48 <oferby_> I personally worked in the bug that we found
17:15:49 <Sukhdev> https://bugs.launchpad.net/networking-l2gw
17:15:56 <oferby_> that DVR is not working
17:16:19 <itamarofek> Hi, Hope I'm not too late
17:16:34 <Sukhdev> itamarofek : welcome
17:16:55 <oferby_> but armax do not want to put it in because it is not done in a way that the routing will be done by hardware router
17:17:11 <oferby_> I tried to tell him that the code now does not work at all
17:17:43 <oferby_> and it is better to have code that is not the *BEST* but fix the issue
17:18:17 <oferby_> but he does not agree and we have a critical issue that is unresolved.
17:18:30 <Sukhdev> oferby_ : swami filed an RFE which overlaps with your work - perhaps he wanted you work with swami to make sure you both are addressing it in a correct way
17:18:48 <Sukhdev> oferby_: do you have link to his RFE?
17:19:10 <Swami> Sukhdev: oferby_: yes
17:19:16 <oferby_> can you send a link
17:19:22 <Swami> oferby_: i wanted to have a chat with you regarding the patch.
17:19:36 <Sukhdev> Swami : glad you are here
17:19:47 <itamarofek> Actually, I think the code is the right fix as it do allow implement l3 on the vtep device later on. As I understand I simply allows any routing instance to work
17:19:54 <Swami> oferby_: The RFE currently does not have any blueprint associated.
17:19:54 <Sukhdev> Swami : do you have link to oferby's patch?
17:20:33 <Swami> oferby_: Yes I like your approach as well, since the one the we were proposing will mandate to have an additional agent and specific node to support this feature.
17:20:57 <Swami> oferby_: but before we go in this direction, I wanted to have some clarification on the patch.
17:21:11 <oferby_> please.
17:21:17 <Swami> oferby_: can we have an offline chat after this meeting, so that we can be on the same page.
17:21:31 <oferby_> sure. I will be happy to do that.
17:21:43 <Swami> oferby_: thanks
17:22:11 <oferby_> sukhdev, going back to where we were
17:22:12 <Sukhdev> Swami oferby : feel free to pull me in that discussion - if we can reach a consensus, we can move forward with this patch
17:22:32 <Swami> Sukhdev: thanks will ping you if get you an agreement.
17:22:51 <Sukhdev> Swami : thanks
17:23:11 <Sukhdev> oferby : back to you - please go on
17:23:19 <oferby_> I need to know if everyone's intention is to add what we would like to see in
17:23:42 <oferby_> and I need to understand if there is a way to speed things up
17:24:08 <Sukhdev> oferby : yes and yes
17:24:12 <oferby_> we really would like to use the L2GW for our needs
17:24:39 <Sukhdev> oferby : however, we need to address 2) which I mentioned above
17:25:10 <oferby_> but the current development is not aligned with the speed I expect from the project
17:25:22 <Sukhdev> oferby : do you want your new feature on top of something which may not be rock solid
17:25:39 <oferby_> again. I agree that we MUST fix bugs first
17:26:01 <Sukhdev> oferby : then we are in violent agreement
17:26:08 <oferby_> but we need to fix the critical / Major and do the minor while adding new features
17:26:24 <Sukhdev> oferby : agreed -
17:27:00 <Sukhdev> here is what is needed to make the code base solid -
17:27:28 <itamarofek> oferby: correct me if I'm wrong, currently only legacy is working, the patch allows for the DVR to work as well
17:27:49 <Sukhdev> We need a way to test things so that when we add new patches, we ensure that we do not break the code base
17:29:07 <Sukhdev> oferby : for instance see this patch - https://review.openstack.org/#/c/202495/
17:29:56 <Sukhdev> this is held up because of ovs lib -
17:31:08 <Sukhdev> itamarofek : I think DVR patch is not an issue - once we get it propertly reviewed, I will approve it for merge
17:31:36 <itamarofek> Great !
17:32:29 <Sukhdev> oferby:  so, coming back to what we were discussing. I am trying to get some volunteers who will help solidify the code base
17:33:14 <Sukhdev> oferby : I want to add some more core's to this project so that we can beef it up to address the speed issue that you mentioned earlier
17:33:26 <Sukhdev> hence, we need more contributors
17:34:13 <Sukhdev> oferby : you still here?
17:34:16 <oferby_> yes
17:34:26 <oferby_> I'm not sure how can I help
17:34:43 <Sukhdev> oferby : work with me - we can help each other
17:35:26 <Sukhdev> oferby : I considered making you core for this project - but, your interest is only in adding new features - which is not bad
17:35:47 <oferby_> this is totally not the case
17:36:09 <Sukhdev> but, we need cores who can jump in review the code - to speed things up (as you say)
17:36:09 <oferby_> I'm more than willing to fix bugs if we have them
17:36:21 <Sukhdev> and help solidify the base
17:36:46 <Sukhdev> oferby : so, work with me - I am in discussion with couple of other guys who want to help
17:37:03 <oferby_> I can ask more people to be reviewers
17:37:09 <itamarofek> Sukhdev: I want to give a hand too.
17:37:14 <oferby_> but I'm not sure they will add code
17:37:26 <Sukhdev> once they get on board, we can add new cores to this project and move things much faster - the way we used to
17:37:30 <itamarofek> with oferby acceptence
17:37:36 <Sukhdev> itamarofek : perfect
17:37:40 <oferby_> obviously
17:38:23 <Sukhdev> oferby: I would love to see you a core for this project so that you and I can move things along - but, in order to do so, we have to plan things right
17:39:12 <oferby_> I'm more than happy to do that. but we have to be able to communicate faster
17:39:24 <oferby_> bi-weekly meeting will not do.
17:39:44 <Sukhdev> oferby : we can fix the communication issue
17:40:45 <Sukhdev> oferby : I am most of the time on IRC and email works best as well
17:41:21 <Sukhdev> oferby: considering we are in differnet time zones - we can set time to ping each other on set times every other day
17:41:32 <Sukhdev> I do it all the times with others -
17:41:48 <oferby_> sure. let me know what you need to move this forward
17:41:57 <oferby_> from me.
17:42:24 <Sukhdev> oferby : we need a plan to begin with - from the top of my head, here is what we need
17:42:54 <Sukhdev> 1) A way to test things - to make sure anytime we push a new path, we did not break the basic functionality
17:43:27 <Sukhdev> 2) we need to switch to ovs lib - and get away from the hacks used in the code base to make things work
17:44:16 <Sukhdev> 3) and put a plan of action to scrub the bugs and address them so that the code base is usable by everybody
17:44:34 <oferby_> Sukhdev, test coverage for the entire project will take a long time. we can't just put an effort there
17:45:05 <oferby_> we have to have a way to add new features too hand to hand with new testing.
17:45:34 <Sukhdev> oferby_ : agree - I am not looking for moving the mountains - as long as we have some basic tests
17:45:55 <oferby_> I offer that a new feature HAVE to come in with unit test.
17:46:21 <oferby_> eventually we will have full coverage.
17:46:21 <Sukhdev> right -
17:46:39 <fyeh> Hi, I'm new here.  Apologies if this is off-topic.  I need to provide VLAN networking for Ironic Baremetal nodes under Liberty+.  Where can I find patch(es) that can provide communication between Ironic and Neutron?
17:47:22 <Sukhdev> fyeh : you are talking to the right person, but, you are in the wrong meeting -
17:47:36 <Sukhdev> fyeh : ping me after this meeting and I will help you
17:48:15 <fyeh> Sukhdev: Ok will do.  Thanks!
17:48:50 <Sukhdev> oferby_ : So, I am not looking to move the mountains on day one - lets put our heads together come up with a solid plan and start to execute on it
17:49:09 <oferby_> I agree.
17:50:28 <Sukhdev> oferby_ : for instance, I have +2 on your patch, I need more reviewers to review it -
17:50:45 <oferby_> which one?
17:50:57 <oferby_> I can ask people to look at it.
17:51:22 <Sukhdev> the new API that you are proposing
17:52:01 <oferby_> I believe that there are few that added remarks which I responded promptly.
17:52:27 <Sukhdev> I mean I want cores to jump in to review as well -
17:52:51 <oferby_> I see. I can ask other cores to look at it.
17:53:21 <Sukhdev> therefore, we need additional cores for this project - armax is very busy with PTL duties
17:54:34 <oferby_> again, I can have core people to review but not sure that they will add code in.
17:55:32 <Sukhdev> I am not looking at them to add code - I want them to review your proposal so that we can move forward with your patches
17:55:46 <oferby_> I can have tht.
17:55:49 <oferby_> that.
17:56:01 <Sukhdev> good
17:56:34 <Sukhdev> Also, going back to my original point, how will you ensure that everything works?
17:56:47 <Sukhdev> we have been relying on HP's CI for that
17:57:54 <oferby_> personally I run massive tests on every patch I put in
17:58:12 <oferby_> we do not have internal CI, if that's what you mean.
17:58:31 <oferby_> the current one does not test all the patches?
17:58:32 <Sukhdev> are those generic tests?
17:59:35 <Sukhdev> can those tests be leveraged?
18:00:20 <oferby_> I'm not sure I follow ...
18:01:09 <oferby_> The patches I added were system tested, not just unit testing.
18:01:11 <Sukhdev> you said you run massive tests -
18:02:03 <Sukhdev> are those tests that can be added to upstream code ?
18:02:38 <Sukhdev> I did not realize, we are out of time -
18:02:50 <oferby_> ok
18:02:55 <Sukhdev> oferby_ : lets switch to neutron channel - need to clear this channel
18:03:02 <Sukhdev> thanks for attending everybody
18:03:05 <Sukhdev> bye
18:03:08 <Sukhdev> #endmeeting