19:00:52 #startmeeting Dragonflow 19:00:53 Meeting started Mon Sep 25 19:00:52 2017 UTC and is due to finish in 60 minutes. The chair is oanson. Information about MeetBot at http://wiki.debian.org/MeetBot. 19:00:54 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 19:00:56 The meeting name has been set to 'dragonflow' 19:01:06 Hi 19:01:11 Who's here for the Dragonflow weekly? 19:01:17 Hi 19:01:19 Hi 19:01:43 Whew. For a second I thought I was alone. It was all dark and quiet. 19:02:12 Let's wait another minute. irenab and lihi probably won't make it, but maybe dimak will show 19:02:15 No way... we're here for the party. But I can't find the bottle opener. 19:02:26 First draw, on the right. 19:03:08 All right. Let's do this! 19:03:12 #topic Roadmap 19:03:27 Just a reminder, the agenda is here: https://wiki.openstack.org/wiki/Meetings/Dragonflow#Meeting_25th_September.2C_2017_.281900_UTC.29 19:04:07 LBaaS : the spec is growing. I think most of the main stuff is covered. I try to keep it updated, and you comments are welcome. 19:04:35 L3 flavour: No progress. I saw the video from the Neutron PTG. My understanding is that it's every driver for itself? 19:05:24 oanson: I saw that you updated the LBaaS spec but didn't get a chance to review the latest. 19:05:45 pino, sure. No rush. This one will take all cycle. 19:06:10 Anyway, I'll drop by the Neutron crowd and verify, and also verify what's expected when working with the standard l2 agent. 19:06:55 oanson: each driver is responsible for its own wiring 19:07:22 mlavalle, but is there standard wiring? e.g. there's a port/tap device the driver should connect to? 19:07:58 oanson: I don't think so. but we can see what yamamoto is doing 19:08:37 and worst case we can adopt the wiring that makes sense to us 19:08:46 Right now he's probably sleeping. I think it's 4 AM there. But I'll ask him and armax tomorrow. (I understood armax is also on this) 19:09:15 Yes. But an agreed interface would be better. 19:09:52 etcd publisher: etcd3gw was released this morning (or yesterday night). I think lihi plans to update global-requirements.txt and push some fixes that were waiting for it. 19:10:33 I think she also started to work on re-stabalizing the opentack-ansible deployment. 19:10:50 That also covers that next item :) 19:11:25 RPM packaging: I had a lot of fun with this one. I opened the following RFE to redhat to add python[23]-jsonmodels to the fedora and centos repo. 19:11:27 oanson: please see the l3 flavors section here, to see who is doing what: http://lists.openstack.org/pipermail/openstack-dev/2017-September/122583.html 19:11:27 Link in a second 19:11:49 #link python-jsonmodels RedHat RFE https://bugzilla.redhat.com/show_bug.cgi?id=1494979 19:11:50 bugzilla.redhat.com bug 1494979 in Package Review "Review Request: python-jsonmodels - Models to make easier to deal with structures that are converted to, or read from JSON in python" [Medium,New] - Assigned to nobody 19:11:53 mlavalle, thanks. Will do. 19:12:53 mlavalle, according do that, armax is responsible (If I understand correctly) 19:13:12 in compatibility between L3 and ML2 drivers 19:13:19 yes 19:13:37 All right. 19:13:38 As I said, I'll try contacting both tomorrow. 19:13:39 he will also review the code for the imporvement in the callbacks 19:14:07 I'll have to take that up with dimak. He had a better look to know what's missing. 19:15:00 Anything else for l3 flavour? 19:15:27 (Sorry, I was on a roll with the updates (: ) 19:16:54 All right. So back to RPM packaging: Now with python-jsonmodels, I am slowly hacking the spec to get the gate to accept it. I hope it'll be done this week. 19:17:25 If anyone knows someone that can sponsor me / the package, that would be great 19:18:58 Upgrade and Grenade: There is a spec (Link in a second). I'd be happy if you could take a look. dimak and I have a disagreement and I think an outside voice might help us reach an agreement. 19:19:13 #link Upgrade spec https://review.openstack.org/#/c/500647/ 19:19:36 That's all I have for roadmap. 19:20:05 Oh, I forgot the DHCP rework. 19:20:28 The spec looks almost done. 19:20:31 #link DHCP spec https://review.openstack.org/#/c/498464/ 19:21:01 I think it can be merged this week. Most of the latest comments were language fixes rather than actual design issues. 19:21:23 yep, waiting for response from irenab..hope that clear from neutron side .. 19:21:52 There's also a patch. It's still WIP, but it looks promising. 19:22:02 Now I'm done for roadmap :) 19:22:17 Anything else for this topic? 19:22:51 Speak now or forever hold your pieces... 19:23:15 #topic Bugs 19:23:29 Not much to say here, so early in the cycle. 19:23:36 It's good to see we have nothing critical yet :) 19:23:46 Any particular bugs that need discussing? 19:24:30 All right. 19:24:48 #topic Open Discussion 19:24:53 Two main items here: 19:24:56 1. vPTG 19:25:05 I want to set the dates for 18-19th of October. 19:25:18 Due to the holidays here, we can't do it any earlier. 19:25:39 pino, I understood you have some openstack days exactly then. 19:26:18 mlavalle, do these dates work for you? 19:26:52 They do 19:27:01 Great! 19:27:40 oanson: to make things easier, I won't be able to attend an event on the week of November 5th, That is the summit 19:28:01 other than that, please feel free to schedule any other week 19:28:03 mlavalle, I won't either :) 19:28:24 you going to Sydney? 19:28:24 Something about a workshop and a project update that someone scheduled for me :) 19:28:43 ++ 19:28:52 great, see you there 19:28:56 Yep! 19:29:09 I arrive Sunday morning and fly back Thursday morning 19:29:19 I don't have flight info yet. 19:30:12 But once I'll do, I'll update. I was given a strong recommendation to arrive a day early to resuscitate from the flight. 19:31:54 So I'll send the email tomorrow. I'll probably leave the time issue open for debate, since I assume a 9-17 in UTC would be uncomfortable for anyone not near Greenwich. 19:32:19 Second issue I wanted to raise was ZMQ broker. 19:32:33 #link ZMQ broker spec https://review.openstack.org/#/c/507034/ 19:33:11 I left a comment there .. 19:33:22 Basically this is a call for comment. It should be easy enough reading. The gist is that multiproc publish/subscribe (i.e. running a new service and funnelling pub/sub messages through there) is giving us problems 19:33:48 Mostly when we want compute nodes to send events (But I suspect that as new scenarios arise, new issues will arise with them) 19:34:29 leyal, thanks. This is in fact a known issue in distributed communication models. I think they even mention it in the link I added. 19:35:23 So I'd be happy to hear your thoughts. Especially if there is something that already does this well we can integrate with so we won't have to write it ourselves. (I saw a couple of things, but nothing that looked maintained, or an exact match) 19:35:51 That's all I have. 19:35:58 The floor is free. 19:36:09 Sorry had to step away... catching up now. 19:36:46 pino, main issue was the vPTG. 19:37:00 I'm afraid we have to set the date for 18-19th of October. 19:37:03 Right... what is the format for that? 19:37:16 Otherwise, it's after the summit, which is a bit... funny... 19:37:37 Probably the bunch of us sitting in a video conference room with laptops, pretending to work and having loud discussions 19:37:39 I'm trying to figure out whether I should cancel my talks at OpenStack MexicoCity and Guadalajara 19:38:01 I don't know. I have know idea what kind of turnup there'd be 19:38:14 Will you have a fixed agenda? 19:38:30 We can also (probably) post the recordings on youtube like with the real PTG 19:38:46 Semi-fixed. It's being constructed here: (1 sec, I'll find the link) 19:38:59 Perhaps I can influence the agenda so that I can attend the meetings I'm interested in (at times when I'm not on flights. 19:39:07 #link Queens PTG schedule https://etherpad.openstack.org/p/dragonflow-queens 19:39:19 pino, sounds like a plan. 19:39:31 Ok, will do. 19:39:40 It would help coordinate the international time difference 19:40:06 We're doing this all on US time right? :-D 19:40:49 I think I mentioned Greenwich somewhere... But I admit I didn't say which :D 19:41:50 Let's just set the schedule so that all interested parties can make their respective interesting meetings 19:41:54 I think that's the best we can hope fore 19:41:55 for* 19:43:30 I suggest all agenda/schedule stuff should be done over the etherpad. This way it's all organised in a single location. 19:43:48 It's either that, or I open a github project and we do this with pull requests :) 19:44:31 Anyone wants the floor? 19:45:34 All right. 19:45:39 Thanks everyone for coming out! 19:46:05 Glad we managed to sync and agree on things! 19:46:27 #endmeeting