20:00:08 #startmeeting Octavia 20:00:08 Meeting started Wed Nov 1 20:00:08 2017 UTC and is due to finish in 60 minutes. The chair is johnsom. Information about MeetBot at http://wiki.debian.org/MeetBot. 20:00:09 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 20:00:11 The meeting name has been set to 'octavia' 20:00:16 Merged openstack/neutron-lbaas master: Updated from global requirements https://review.openstack.org/516955 20:00:24 Hi folks 20:00:29 hi 20:00:29 hi 20:00:46 Time for another installment of "Fun with Octavia" 20:00:57 #topic Announcements 20:01:02 o/ 20:01:10 Queens MS1 release released last week 20:01:38 It was delayed due to release gate issues with zuul v3, but it did finally go out. 20:01:49 OpenStack Summit Sydney is next week, November 6-8th 20:02:10 Some people may be attending and might not be as available on IRC. 20:02:16 (I am not attending) 20:02:32 Any other announcements of note this week? 20:03:08 Ok, moving on 20:03:15 #topic Brief progress reports / bugs needing review 20:03:32 Just a reminder, we have an open spec for review on the provider drivers: 20:03:39 #link https://review.openstack.org/509957 20:04:01 Please contribute and help us review so it can be the spec we all want it to be. 20:04:41 It is on revision 3 and the title is currently incorrect, but we can keep commenting 20:05:46 Otherwise I have been working on zuul v3 issues, fixing gate breakage (cryptography 2.1), and working around the instance failures at OVH for the gate jobs. 20:06:14 I have some back ports to do and some more zuul v3 work on the stable branches. 20:06:58 We also need to get a scenario gate or two setup for our tempest plugin work. I may work on that if rm_work isn't able to start that. 20:07:26 Any other progress reports? 20:07:55 This dashboard patch could use some core attention too: https://review.openstack.org/513086 20:08:25 And this bug with child object provisioning status: 20:08:31 #link https://review.openstack.org/513086 20:08:42 #link https://review.openstack.org/#/c/514867/ 20:09:09 Any other bugs to discuss? 20:09:13 johnsom: any update regarding Fix allowed_address_pairs network driver to be more tolerant to failure 20:09:33 jniesz Just that I haven't got back around to work on that 20:09:52 ok, and was the last one related to the issue I raised or is that separate? 20:10:39 jniesz It was related to the issues you were seeing 20:11:50 So, half the battle 20:11:53 grin 20:11:55 : ) 20:12:33 Ok, if there aren't any other progress reports/bugs to discuss, I will move on to the next topic. 20:12:44 #topic UDP load balancing 20:13:12 We have an open RFE that has some use case data now: 20:13:19 #link https://storyboard.openstack.org/#!/story/1657091 20:13:43 Feel free to comment on this RFE story. 20:13:59 I think the next step is a spec for this. 20:15:37 My expectation is an initial implementation can be done with LVS sideloaded in our existing amphora image, with the amphora agent updated to include support for configuring it when a UDP listener comes in. 20:16:02 Kind of a keep it simple approach. 20:16:09 I think UDP option makes sense, and it is something I would like to implement with L3 distributor 20:16:52 Yeah, that can certainly be a follow on step. 20:17:06 Important thing is that spec allows for different implementations of UDP LB 20:17:21 Initially I'm thinking we should just do standalone, maybe atc/stdby to get the base stuff, API etc. worked out. 20:17:54 Yep. Please help review to make sure we don't make a mistake 20:18:16 yes, I can review spec when it is posted 20:18:41 Any other discussion on this? 20:19:07 we should be careful not to duplicate work with ACTIVE-ACTIVE LVS 20:19:43 also I think we will need keepalived to manage the health of the UDP members… 20:19:46 Yeah, I think it's 100% different, so really should not overlap 20:20:18 I think it overlaps — for instance you would use the same dib element 20:20:23 There isn't even a spec for act/act LVS yet, so not a lot there anyway 20:21:07 I'm not sure we need a DIB element for UDP. The existing keepalived element should cover everything for UDP 20:21:32 we will need ipvsadm on the amp to get sdiagnostic info 20:21:41 and debug 20:22:13 Maybe to get stats 20:22:18 yep 20:22:29 We will see. 20:23:31 Anything else? 20:23:58 Cool, I think we have a path forward and someone willing to work on it. 20:24:33 I guess I should say, able to work on it... So much stuff we could do, just prioritizing... 20:24:45 #topic Open Discussion 20:24:52 So this just happened: 20:24:55 #link https://twitter.com/knelson92/status/925818114813464576 20:25:14 They didn't make stickers for the PTG, but have now. 20:25:23 rm_work I will put trust in you to send me some from Sydney 20:25:31 Kendall will try to find rm_work to give them to. 20:26:06 If you are at the summit you can catch rm_work to get one. Otherwise we will figure out how to get them to Dublin for the next PTG 20:26:27 derp 20:26:31 disconnected from IRC and didn't notice >_> 20:27:17 Well, a bit of scroll back then.... 20:27:41 rm_work Kendall will get you the peacock stickers in Sydney 20:28:21 Any other topics for open discussion today? 20:28:32 yeah just got through it 20:28:59 umm 20:29:21 nothing here :P 20:29:26 Hahaha 20:29:46 Ok. Thanks folks. Please help with reviews! 20:29:52 #endmeeting