14:04:20 <dmellado> #startmeeting kuryr 14:04:21 <openstack> Meeting started Mon May 13 14:04:20 2019 UTC and is due to finish in 60 minutes. The chair is dmellado. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:04:22 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:04:24 <openstack> The meeting name has been set to 'kuryr' 14:04:30 <dmellado> hiya foilks, who's here today for the kuryr meeting? 14:04:51 <dulek> o/ 14:05:19 <dulek> Though I'm just back from my PTO, so there won't be a lot of updates from me. ;) 14:05:53 <dmellado> I came back from SnowpenStack teh 2nd 14:06:00 <dmellado> that's just Denver for all of you 14:06:19 <dmellado> I had a nice discussion with sean k mooney and jan from os-vif 14:06:24 <dmellado> did you get to see the etherpad? 14:06:58 <dmellado> #link https://etherpad.openstack.org/p/os-vif-kuryr 14:09:12 <dmellado> #topic kuryr-kubernetes 14:09:37 <dmellado> besides that, I also want to let you folks know that all kuryr cores are now also devstack-plugin-container cores 14:09:58 <dmellado> as hongbin kindly asked us to help on that so we could all get faster reviews 14:10:06 <dmellado> so yahoo new +2 powers 14:11:49 <dmellado> we were also discussing about...who would guess 14:11:52 <dmellado> octavia and amphoraes 14:11:59 <maysams> XD 14:12:36 <dmellado> johnsom: around? 14:12:39 <johnsom> Great things 14:12:45 <dmellado> hiya! 14:13:19 <dmellado> I kinda recall that we noted down the current issues on the non-lxc containerization 14:13:22 <dmellado> wouldn't you have that link around? 14:13:37 <johnsom> Sure, just a second 14:13:56 <johnsom> https://etherpad.openstack.org/p/octavia-train-ptg 14:14:03 <johnsom> Line 204 14:14:19 <johnsom> It's a rough list Carlos captured. 14:14:29 <dmellado> thanks johnsom 14:14:49 <johnsom> Definitely do-able, just some work required. 14:15:29 <johnsom> I think most of that list is about issues running under k8s. Not as focused on Docker 14:15:40 <dmellado> so overall, most of my concerns are also regarding performance 14:15:42 <johnsom> Or other non-lxc flavors 14:15:47 <dmellado> yeah 14:16:37 <dmellado> but in any case, ltomasbo I command thee to set up some fancy meeting to discuss this as and try to set up work items 14:16:39 <dmellado> xD 14:16:45 <johnsom> I think the key part is getting the multiple processes we use communicating 14:17:28 <dmellado> honestly yeah 14:17:40 <dmellado> but also how to handle scheduler if that makes you lose connection 14:18:23 <johnsom> Yeah, I personally feel that k8s is just not a good fit for a low level networking service. At least not last time I looked. 14:18:48 <ltomasbo> o/ reading backlong... 14:18:54 <ltomasbo> backlog 14:21:01 <dmellado> ltomasbo: not that much so far, so no worries 14:21:10 <ltomasbo> yep! 14:21:26 <dmellado> I was just going to say that the same that we'll be having a meeting with the os-vif folks 14:21:40 <dmellado> let's schedule another one with the octavia team 14:21:48 <dmellado> and let's call it vPTG 14:22:04 <dmellado> I'll put up some etherpads/meetings and so 14:22:29 <dmellado> johnsom: you're be our most welcomed guess 14:22:39 <johnsom> Nice 14:23:48 <dmellado> Another topic that got covered around the PTG, besides the trains, which might sound funny but still affected some rooms 14:24:01 <dmellado> was about making some changes at the openstacksdk/client 14:24:17 <dmellado> in order to tackle the missing parts for bulk operations 14:24:59 <dmellado> we had a meeting with dhellmann and agreed to somehow add a new model to the openstackclient 14:25:09 <dmellado> pretty much in the same way that kubernetes is doing with files 14:25:15 <dmellado> so insted of kubectl apply -f 14:25:34 <dmellado> it'll be somehow along the lines of openstack bulk create -f foo.yaml 14:27:39 <dmellado> njohnston will be the one proposing such a model change so maysams, dulek, ltomasbo, please do lend him a hand and review his patches 14:28:16 <dmellado> also, ltomasbo 14:28:40 <ltomasbo> dmellado, it will be real bulk operation? or just a way to call the api just once? but the tasks are done 1 by 1 14:28:47 <dmellado> I'll be getting you in touch with some INRIA folks who we might want to sync in Kuryr edge networking usecases 14:29:07 <dmellado> ltomasbo: njohnston modified neutron so now there's real bulk operation using the api 14:29:27 <ltomasbo> dmellado, bulk operations for what? creating/deleting ports? 14:29:39 <ltomasbo> or for anything? like creating sg rules? 14:29:53 <ltomasbo> the later will be awesome for network policies.... 14:29:56 <dmellado> ltomasbo: seems like everything that the api supports would now be real bulk 14:30:01 <dmellado> so that'd be really cool 14:30:01 <maysams> for sure!! 14:30:11 <dmellado> I'm pretty sure it'll heavily affect performance 14:30:35 <dmellado> and we might even want to deprecate port pools if this goes all well 14:30:38 <ltomasbo> I won't believe it until I see it in action :) 14:30:54 <dmellado> ltomasbo: me neither, so *again* I command thee 14:30:56 <dmellado> xD 14:30:59 <ltomasbo> dmellado, not sure about that either... still better to have them ready than re-create them 14:31:10 <dmellado> well, you know the rest, please do test this if you get the chance 14:31:21 <dmellado> and let's sync with the neutron guys if there's something that is *not* there 14:32:07 <dmellado> ltomasbo: maysams any update from your side? 14:32:31 <dmellado> dulek: how's being back home? hope you like the beers better now :D 14:33:01 <maysams> dmellado: I've been mostly working on adding more tests to upstream k8s, and fixing some bugs regarding NP 14:33:20 <ltomasbo> dmellado, not much, working on the patch to add pools to newly created namespaces: https://review.opendev.org/#/c/656975/ 14:33:27 <dulek> dmellado: First the Internet is waaaay better. ;) 14:33:43 <dmellado> dulek: oh, true! you'll even be able to use video now xD 14:33:47 <dulek> :) 14:34:08 <dulek> dmellado: And beer was here as well, as friends got us a surprise party on Friday when we drove back to our place. :D 14:34:32 <dmellado> btw, recalled about you in Denver as I got to see your Intel friends 'from Berlin' 14:35:04 <dulek> dmellado: Ah right, Roman was there, and the rest of OVH folks. 14:35:21 <dmellado> and in case you're wondering what did the foundation was giving away this time 14:35:29 <dmellado> now I do own some openstack socks 14:35:31 <dmellado> xD 14:37:37 <dmellado> all right, so it looks like this was all for today, we'll be at #openstack-.kuryr in case anyone wants to hang around 14:37:38 <dmellado> thanks all! 14:37:42 <dmellado> #endmeeting kuryr