09:00:37 <oanson> #startmeeting Dragonflow 09:00:38 <openstack> Meeting started Mon Feb 27 09:00:37 2017 UTC and is due to finish in 60 minutes. The chair is oanson. Information about MeetBot at http://wiki.debian.org/MeetBot. 09:00:39 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 09:00:42 <openstack> The meeting name has been set to 'dragonflow' 09:00:46 <lihi> Hi 09:01:19 <oanson> I guess we'll wait a minute :) 09:02:25 <dimak> Hey 09:02:30 <oanson> Hi 09:03:19 <oanson> If it's just the three of us, I suggest we waive the meeting. I don't see a reason to sync three people who know what they're doing 09:04:03 <yamamoto_> hi, you all folks attended ptg had a safe return trip? 09:04:16 <oanson> yamamoto_, yes. Thanks. You got back ok? 09:04:30 <lihi> Yeah, we are back home :) 09:05:19 <yamamoto_> i'm ok thank you 09:06:41 <oanson> yamamoto_, are you interested in the Dragonflow meeting, or did you just drop by to say hello? 09:07:32 <yamamoto_> both 09:07:41 <oanson> All right. Then let's begin. 09:07:49 <oanson> #info lihi dimak yamamoto_ in meeting 09:08:09 <oanson> I'll do a short round-up of what we did in the PTG: 09:09:04 <oanson> We met with the Ironic folks. It seems like if we support Neutron interoperability, we should support them out-of-the-box 09:09:17 <oanson> That's already planned as part of itamaro's work, so that's good news for us 09:09:54 <oanson> We met with the infra guys regarding multinode testing. They have everything in place, and I got a long and thorough explanation on what happens. So it shouldn't be a problem. 09:10:17 <oanson> Multinode can be used to test cross-node communications using encap/decap, provider networks, etc. 09:10:30 <irenab> sorry for being late 09:11:01 <oanson> Spoke with the TripleO guys. Shouldn't be a problem to get Dragonflow on it, but it will need a carrier since we need puppet deployment for it 09:11:08 <oanson> irenab, no worries 09:11:12 <oanson> #info irenab also in meeting 09:11:44 <oanson> Spoke with the packaging people. RPM packaging is easy. dimak agreed to take it once SFC reaches a stable point. It doesn't look like a lot of work 09:12:09 <oanson> debian may take a bit more effort, since they're doing a bit of a revamp there now. But no rush :) 09:12:42 <oanson> We spoke with the OSA and infra guys regarding getting OSA/dragonflow gate to run on Dragonflow. 09:13:10 <oanson> We spoke with the doc guys regarding getting a DF manual installation guide up there. 09:14:26 <oanson> We had a very interesting talk with Kuryr and Octavia regarding dragfonlow support for Octavia API, and LBaaS (Both layer 4 and layer 7). I plan to continue discussion with Octavia PTL (johnson) regarding action items and possible collaboration 09:15:18 <oanson> Discussed Dragonflow as a back-end for the interop challenge in Boston. The plan is container workloads for which they need VLAN aware VMs. I am now working on that. There is already a spec available, and the groundwork is done. 09:15:32 <oanson> #link VLAN aware VMs https://review.openstack.org/#/c/437486/ 09:16:13 <irenab> oanson, do you ahve some references for the interop challenge? 09:16:36 <oanson> irenab, yes, but nothing organised. 09:16:47 <irenab> oanson, no rush 09:17:10 <oanson> Then I'll organise an etherpad and post it in the channel today/tomorrow 09:17:45 <oanson> Regarding gates above, we also discussed with the Neutron PTL regarding running our gate jobs (unit and fullstack) as post-merge jobs + notification to the Dragonflow people upon new failures. 09:17:58 <oanson> Hopefully this will be an early warning system to future gate issues 09:18:23 <irenab> oanson, failures on neutron? 09:18:39 <oanson> Failures on Dragonflow, after a Neutron merge 09:19:02 <oanson> e.g. Neutron changed their APIs or some underlying implementation detailed we relied on. 09:19:11 <irenab> cool 09:19:42 <oanson> I think I covered everything. 09:20:05 <oanson> Anything anyone wants to volunteer for? 09:20:37 <irenab> oanson, loop me in for the interop challenge 09:20:55 <oanson> irenab, will do, but let's take this offline. As I said, I don't have all the info on hand. I need to look it up 09:21:06 <irenab> oanson, sure 09:21:15 <oanson> All right. Roadmap updates 09:21:17 <oanson> #topic Roadmap 09:21:40 <oanson> lihi uploaded an IPv6 on SG patch. 09:22:14 <oanson> #link IPv6 for SG https://review.openstack.org/#/c/431566/ 09:22:21 <lihi> Yeah. The UT still needd an update, and the fullstack tests are missing 09:22:40 <oanson> lihi, so is this patch still work in progress? Or is it ready for review? 09:23:06 <lihi> But other that that, it is ready for reviews. Hit me with your -1 :) 09:23:31 <oanson> Cool! 09:23:32 <oanson> I'll also mention that we also need IPv6 support in port security. I think it is currently bypassed for ipv6. 09:23:58 <lihi> I saw your bug. OK, will do 09:24:04 <oanson> Great 09:24:48 <oanson> NB refactor is almost ready. I think there are 3 patches left before we can start migration 09:24:49 <dimak> lihi, I'd love some help with fullstack tests in IPv6 SFC parts 09:25:02 <lihi> Sure, np 09:25:13 <oanson> dimak, what's the SFC status? 09:25:34 <dimak> oanson, a lot of the refactor patches are in, but I decided to post some more :) 09:25:48 <dimak> I've added some DbStore functionality I needed for SFC 09:25:49 <oanson> Yes, but what's the status? :) 09:26:08 <dimak> SFC is still not fully functional 09:26:25 <dimak> The flow classification was working quite well 09:26:49 <dimak> And I got most of the stuff to work on a single node 09:26:57 <oanson> Cool 09:27:04 <dimak> I'll set up multinode env sometime this week to start testing that 09:27:31 <oanson> Please write down some multinode test scenarios. We'll add them once multinode gate tests are up and running 09:27:41 <dimak> Of course :) 09:28:13 <dimak> Nothing else on my part I think 09:28:19 <oanson> Cool 09:29:04 <oanson> Chassis/Service health was updated, but still has some work to be done. I don't know how available rajivk will be, so maybe someone should help out. I don't think there's much work left 09:29:23 <oanson> We want it both for the feature itself, and the Service parent class which should help us a lot 09:29:54 <oanson> sNAT needs reviews and testing. I tested it and it works, but more eyes would be a great help 09:30:25 <irenab> oanson, I will check as well 09:30:28 <oanson> The API spec is in. I guess the next step is migrating the models, and then getting OpanAPI to produce something and see what it does and how well 09:30:32 <oanson> irenab, thanks 09:31:05 <oanson> That's it for roadmap. 09:31:13 <oanson> Anything else here, or can we move on? 09:31:32 <oanson> #topic bugs 09:32:31 <oanson> Not much to say here actually. 09:32:50 <oanson> Any bugs peiple want to bring up? 09:33:27 <oanson> #topic Open Discussion 09:33:31 <oanson> Take the floor! 09:34:20 <oanson> Or... don't. 09:34:24 <oanson> Can we wrap this up? 09:34:39 <irenab> +! 09:34:57 <oanson> irenab, I think you have a syntax error :) 09:35:01 <irenab> :-) 09:35:04 <oanson> Well, thanks everyone. 09:35:08 <oanson> See you next week. 09:35:11 <oanson> #endmeeting