19:00:45 <oanson> #startmeeting Dragonflow
19:00:46 <openstack> Meeting started Mon Jul 17 19:00:45 2017 UTC and is due to finish in 60 minutes.  The chair is oanson. Information about MeetBot at http://wiki.debian.org/MeetBot.
19:00:47 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
19:00:50 <dimak> Hey
19:00:51 <openstack> The meeting name has been set to 'dragonflow'
19:00:57 <mlavalle> o/
19:01:03 <oanson> Hi everyone
19:01:10 <dimak> mlavalle, how you doing? :)
19:01:25 <mlavalle> doing great, how about you guys?
19:01:26 <leyal> Hi
19:01:32 <oanson> No complaints! :)
19:01:41 <dimak> Some complaints
19:01:45 <mlavalle> as of today, i'll start attending this meeting every other week
19:01:46 <oanson> Let's wait another minute. Maybe lihi and irenab will join
19:01:56 <dimak> The meeting is awfully late and delaying my game of thrones premiere
19:02:08 <oanson> mlavalle, we'll be very happy to see you :)
19:02:23 <oanson> dimak, good. This way there'll be no spoilers.
19:02:38 <mlavalle> the other week, the meeting is at 2am my time. so I won't make it that week
19:02:57 <oanson> I think next time we'll just take a smaller planet, with less time difference
19:03:04 <oanson> All right. Let's start.
19:03:15 <oanson> #info dimak mlavalle leyal in meeting
19:03:23 <oanson> #topic Roadmap
19:03:31 <oanson> #link Agenda https://wiki.openstack.org/wiki/Meetings/Dragonflow
19:03:59 <oanson> SFC - dimak, I saw you updated the patches. Any major changes?
19:04:07 <dimak> None, waiting for reviews
19:04:21 <dimak> leyal asked I take some tox scripts out of the drivers patch so I did
19:04:34 <dimak> Other than that, nothing much
19:04:51 <oanson> All right. I should finish my review tomorrow morning.
19:05:12 <oanson> What about L3 flavour?
19:05:13 <dimak> Great, looking forward to getting this merged
19:05:19 <oanson> So am I :)
19:05:27 <dimak> I've revised the spec today with Irena's help
19:05:53 <dimak> I think it is in a better state
19:06:04 <dimak> Take a look, I'll remove the WIP
19:06:09 <oanson> Great.
19:06:15 <oanson> Everyone, please look at the spec and vote
19:06:21 <oanson> #link L3 flavour spec https://review.openstack.org/#/c/475174/
19:06:41 <oanson> mlavalle, I'd be happy if you could take a look too, since you are now our Neutron expert :)_
19:06:47 <oanson> :) *
19:06:48 <mlavalle> is this what discussed during the Neutron meeting a few weeks ago?
19:06:52 <oanson> Yes
19:06:53 <dimak> I'll start working on the l3 service provider this week I think
19:07:09 <mlavalle> that's great, yes, I will definitely take a look
19:07:19 <oanson> Great. Thank you both!
19:07:26 <mlavalle> added to my reviews pile
19:07:45 <oanson> leyal, anything to report about the DHCP stuff?
19:08:20 <oanson> leyal, ?
19:08:27 <leyal> well - for PXE boot all patch are done
19:08:35 <leyal> waiting for reviews ..
19:09:03 <oanson> All right. I'll get to it tomorrow as well.
19:09:09 <oanson> I see dimak already voted.
19:09:29 <oanson> At least on some of the chain
19:09:31 <dimak> Yes, I think a lot of stuff can get merged once you OK them
19:09:43 <oanson> Sure
19:10:03 <leyal> I also started to work about white-paper for Ironic support - but it's not in state that i can show ..
19:10:36 <oanson> Very cool!
19:10:49 <oanson> Looking forward to reading it.
19:11:38 <dimak> Any updates on the LBaaS?
19:11:39 <oanson> I'll report that I sadly made 0 progress on LBaaS and RPM packaging. I can't make any promises for this week due to personal constraints, but I promise to make progress next week!
19:12:04 <oanson> Sorry about that
19:12:11 <dimak> :)
19:13:08 <oanson> In addition to that lihi isn't here to report on the etcd pub/sub, but I guess we'll see some results soon.
19:13:22 <oanson> Anything else for roadmap? Something I forgot?
19:13:37 <dimak> I don't think so...
19:13:46 <oanson> All right. Then let's move on
19:13:50 <oanson> #topic Bugs
19:14:15 <oanson> I did manage to make some progress on bug 1690775 last week.
19:14:16 <openstack> bug 1690775 in DragonFlow "Remove special handling for lport/ofport in local controller" [High,In progress] https://launchpad.net/bugs/1690775 - Assigned to Omer Anson (omer-anson)
19:14:31 <oanson> There seem to be a lot of dependencies piling up on it
19:14:35 <oanson> Mostly other bug fixes.
19:14:57 <oanson> And the distributed ports spec uploaded by dimak
19:14:58 <dimak> Yes I am using several of your patches for the ports work
19:15:14 <dimak> It's a pandora box though
19:15:29 <oanson> Yes. But we knew that when we started
19:15:51 <oanson> But at the end of the day we need an organised manner to work with logical ports
19:15:59 <dimak> https://review.openstack.org/#/c/479828/ <- the mentioned spec
19:16:08 <oanson> Thanks - was looking for it :)
19:16:15 <dimak> Yes, the DNAT fix is blocked by it
19:16:27 <oanson> #link Distributed port spec https://review.openstack.org/#/c/479828/
19:16:36 <oanson> Otherwise things like dnat and trunk port will stop working for no reason
19:16:37 <dimak> Also there's a DHCP related but I opened that will need it
19:16:43 <oanson> Which happened already.
19:16:51 <oanson> dimak, sorry?
19:17:15 <dimak> The one about not using DHCP's MAC address in response
19:17:41 <oanson> I thought we wanted to use the DHCP's mac address in response?
19:17:44 <dimak> We're not caching the DHCP port so there's no obvious way to retrieve without get_all on database
19:17:54 <leyal> https://bugs.launchpad.net/dragonflow/+bug/1703168
19:17:56 <openstack> Launchpad bug 1703168 in DragonFlow "DHCP app always replies in broadcast" [Low,New] - Assigned to Eyal Leshem (leyal)
19:18:08 <dimak> oanson, sorry, yes, the bug is about not using it
19:18:19 <dimak> Obviously the fix is to use the port details :)
19:18:31 <oanson> All right. Thought I missed something important :)
19:18:32 <oanson> yes
19:18:50 <dimak> There's also another funny thing I noticed
19:19:10 <dimak> I have a patch up to remove l3 agent from our gate jobs
19:19:21 <oanson> In my previous company they used to threaten people like you to be bumped over to QA. Just saying :)
19:19:30 <mlavalle> LOL
19:19:38 <oanson> dimak, I didn't see it yet.
19:19:53 <dimak> It failed on a test that check we generate icmp reply with host unreachable
19:20:12 <oanson> Hmm. We need to make sure we don't have a hidden requirement there
19:20:20 <dimak> Apparently with l3 agent the reply is generated in the namespace
19:20:38 <dimak> Because there is no way l3 app generates it.
19:21:00 <oanson> Yes. We may need a default rule to pass the packet to controller, so the controller will generate the packet
19:21:16 <dimak> No no, thats not the issue
19:21:32 <oanson> Then what's the issue?
19:21:33 <dimak> The rule exists, it matches on reg7 of the router interface
19:21:45 <dimak> We don't create those flows in l2 table
19:21:52 <dimak> because those ports are never cache locally
19:21:58 <dimak> cached*
19:22:51 <dimak> Once we sort out the distributed ports implementation we can easily add them
19:23:01 <oanson> I don't remember the code of the top of my head - but I am fairly sure the l3 agent creates the namespaces and interfaces, and patches them in. The L3 app detects that the port is 'real', and adds the relevant rule
19:23:08 <dimak> But then we may want to drop router interfaces embedded in router models
19:23:46 <dimak> I'll investigate in depth tomorroe
19:23:49 <oanson> Yes and no. ICMP unreachable isn't generated with a packet *to* the router. The router generates it, and I think the router port's are handles in a special manner
19:23:50 <dimak> tomorrow
19:24:01 <oanson> Sure.
19:24:18 <oanson> But I suspect you found a missing feature in our L3 apps.
19:24:30 <dimak> Anyway we have the router interfaces in our database, and we can use references
19:24:36 <dimak> I see no reason to embed
19:25:14 <oanson> I think we tested this a while back, and decided we want to stay with embedding.
19:25:30 <oanson> We used to have a bug on it too. But it
19:25:49 <oanson> But now it should be easy to make the change and retest. So, if you feel like it, go for it.
19:25:50 <dimak> Well the models were different back then
19:26:02 <oanson> Yes. This test was done a while back
19:26:33 <lihi> Hi all
19:26:41 <mlavalle> hey
19:26:44 <lihi> I'm sorry for being late
19:26:45 <oanson> Hi lihi. Thanks for making it!
19:26:51 <oanson> No worries.
19:26:54 <dimak> Hey lihi
19:27:02 <oanson> There are two High bugs left: 1704123 and 1651988. Both have patches that fail the gate.
19:27:04 <dimak> That's all I had on ports BTW
19:27:55 <oanson> The snat patch fails since the kernel is too old in the test VM. I tried to install new kernel modules as part of devstack, but that failed. I don't recall why
19:28:13 <oanson> The MTU patch fails since I mucked up the test code. I should have it fixed tomorrow.
19:28:22 <oanson> dimak, don't worry, after your reviews :)
19:28:30 <oanson> Anything else for bugs?
19:29:08 <oanson> #topic Open Discussion
19:29:15 <oanson> The floor is free.
19:29:22 <mlavalle> I have a question
19:29:36 <mlavalle> does this work well? https://github.com/openstack/dragonflow/tree/master/vagrant
19:30:04 <oanson> I haven't used it in a while.
19:30:28 <oanson> I use https://github.com/omeranson/openstack-vagrant and it worked until 2 weeks ago (when devstack on fedora 25 broke)
19:30:28 <mlavalle> well, i'll give it a try
19:30:43 <oanson> lihi also has a version for xenial which works well (tested last week)
19:30:50 <dimak> +1 on oanson's openstack-vagrant
19:31:05 <mlavalle> can I see the one on xenial?
19:31:24 <lihi> https://github.com/lihiwish/openstack-vagrant/tree/xenial
19:31:33 <mlavalle> lihi: thanks :-)
19:31:38 <mlavalle> I'll give it a go
19:32:18 <oanson> Cool.
19:32:24 <lihi> But the xenial version does not work with Fedora (for now)
19:32:45 <mlavalle> ok, i'll keep that in mind
19:32:48 <lihi> When I'll get it to work, I'll merge it to oanson's repository
19:33:00 <mlavalle> where are you located lihi ?
19:33:17 <lihi> In Israel
19:33:46 <mlavalle> so the meeting this week is targeted at people in Israel and the Americas?
19:34:01 <oanson> The meet
19:34:23 <oanson> Basically, yes.
19:34:37 <oanson> The one next week is Israel and far east.
19:34:38 <mlavalle> in that case, it doesn't have to be this late for you
19:34:50 <mlavalle> it is 2:30pm my time
19:35:00 <mlavalle> I'm in US central time
19:35:22 <oanson> I am only available after 17:00 UTC (18:00 during the winter)
19:35:32 <mlavalle> ah ok
19:35:46 <dimak> Yes, by this time we're all home
19:35:55 <mlavalle> cool
19:36:16 <oanson> We also hope that people from the west coast will join
19:36:16 <mlavalle> it's fine with me. Just tryin to make it easier for the Israel bunch :-)
19:36:42 <oanson> I find the grey background of my IRC client soothing :)
19:36:50 <mlavalle> just fyi, for the West Coast people, it is a little past 12:30pm
19:37:30 <oanson> You got me. My main constraint was that I'm only available this late :)
19:37:41 <mlavalle> cool
19:37:54 <oanson> All right. Anything else for open discussion?
19:38:01 <mlavalle> not from me
19:38:13 <mlavalle> delighted to be here :-)
19:38:29 <mlavalle> dimak: go enjoy GoT
19:38:33 <oanson> Well, thanks everyone! :)
19:38:40 <mlavalle> I watched it last night
19:38:54 <oanson> Please, still no spoilers! I have to wait till tomorrow.
19:38:56 <mlavalle> Winter is here
19:38:58 <dimak> Thanks! :)
19:39:04 <lihi> I've forgot there's a new episode! :D
19:39:13 <oanson> #endmeeting