09:00:37 <flwang1> #startmeeting magnum 09:00:38 <openstack> Meeting started Wed Jun 17 09:00:37 2020 UTC and is due to finish in 60 minutes. The chair is flwang1. Information about MeetBot at http://wiki.debian.org/MeetBot. 09:00:40 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 09:00:42 <openstack> The meeting name has been set to 'magnum' 09:00:42 <flwang1> #topic roll call 09:00:50 <flwang1> o/ 09:00:57 <flwang1> seems only you and me, brtknr 09:01:05 <flwang1> strigazi: jakeyip: 09:01:12 <brtknr> o. 09:01:13 <brtknr> o/ 09:01:45 <flwang1> brtknr: i don't have much topics for today 09:02:16 <jakeyip> o/ 09:02:17 <brtknr> nor me really, just wanted to bring to attention the releases 09:02:28 <flwang1> as long as I finish my internal k8s work, i will get the new api for get nodes of NG and master_lb_enabled for cluster creation done 09:02:41 <flwang1> brtknr: i think the train 9.4.0 is good for me 09:02:57 <flwang1> i need take a look the ussuri 10.1.0 09:03:11 <flwang1> i'd like to make sure all the good patches have been backported 09:03:29 <brtknr> flwang1: yeah i am in no rush for 10.1.0 but 9.4.0 is long overdue 09:03:39 <flwang1> brtknr: let's get it done 09:03:40 <brtknr> but would like the last couple of patches to merge 09:03:53 <flwang1> to avoid release team tired to look after it 09:04:01 <flwang1> sure 09:05:44 <flwang1> i like your shebang patch 09:06:05 <brtknr> flwang1: thanks 09:06:09 <flwang1> just one comment, is it only targeting remove the /usr/bin/bash line? 09:06:30 <flwang1> do you think it's a good idea to print the step for each script? 09:06:30 <brtknr> both /bin/sh and /bin/bash 09:06:50 <brtknr> flwang1: i have proposed a second patch to print step 09:07:02 <flwang1> fantastic 09:07:06 <flwang1> no problem 09:07:10 <flwang1> i will test it tomorrow 09:07:12 <brtknr> https://review.opendev.org/#/c/735862/ 09:07:40 <brtknr> the second patch also unsets options set in the beginning of the script 09:07:55 <flwang1> good 09:07:55 <brtknr> we dont want strange unpredictable interaction between scripts 09:08:04 <flwang1> agree 09:09:53 <flwang1> btw, we have built a pipeline for hyperkube image 09:10:12 <flwang1> but i don't know how to implement it on openstack infra 09:10:39 <flwang1> i think that's one of the high priorities for our team 09:11:47 <brtknr> flwang1: do you want to share it? 09:11:55 <brtknr> Is it docker build? 09:12:53 <flwang1> brtknr: we still need sometime to polish it, but it can be shared for sure 09:13:01 <brtknr> Ok 09:13:23 <brtknr> flwang1: btw did you get a chance to look at this: https://review.opendev.org/#/c/735100/ 09:13:29 <flwang1> now the pipeline will create a VM and sync code to it, then run quick release to generate the hyperkbe image 09:13:46 <flwang1> then run kubeadm to start an allinone cluster and run sonobuoy to test it 09:14:36 <flwang1> sure, will review it 09:14:40 <flwang1> thanks for working on that 09:15:22 <brtknr> flwang1: sorry what is the role of kubeadm here? 09:15:28 <brtknr> to test the hyperkube image? 09:15:34 <brtknr> are you building 1.19.x image? 09:15:46 <flwang1> yep, to test the hyperkube image 09:16:11 <brtknr> ah ok 09:16:14 <flwang1> no, we're building v1.16.9 for testing, and will build v1.19.x in the near future 09:17:32 <brtknr> Oh ok do you think the procedure for v1.19.x image will be the same? 09:18:09 <flwang1> it should be 09:18:16 <flwang1> i can't see why it doesn't work 09:18:40 <brtknr> ok 09:18:57 <flwang1> i think the community just remove the hyperkube docker file building, they shouldn't change all the scripts to break things 09:19:02 <brtknr> The alpha version of 1.19.x is already out 09:19:09 <flwang1> otherwise, they will get much more complains 09:19:19 <brtknr> would be good if you could check quickly if the current docker file is compatible 09:19:27 <flwang1> good point, will do 09:19:58 <brtknr> flwang1: thanks 09:20:20 <brtknr> this is a rather annoying change 09:20:35 <brtknr> i dont completely understand why they did it 09:20:43 <brtknr> do you? 09:21:11 <flwang1> based on the issues, they think there are too many CVEs and the team don't want to maintain it anymore 09:23:08 <flwang1> but as strigazi said, k8s is such a big community 09:23:23 <flwang1> it's not responsible to leave this work to each org 09:23:29 <flwang1> to duplicate the effort 09:24:41 <flwang1> anything else you would like to discuss? 09:25:02 <flwang1> brtknr: can you please revisit https://review.opendev.org/#/c/733027/ ? 09:26:10 <brtknr> lgtm thanks +2 from me 09:26:48 <flwang1> brtknr: thanks, i'm going to give it a ninjia approve based on the +2 from strigazi :) 09:27:45 <brtknr> if you're sure... 09:27:59 <flwang1> btw, as for https://review.opendev.org/#/c/726017/ i'd like to understand how do you guys maintain templates 09:28:30 <flwang1> i'm working on k8s versions support policies, i'd like get some input if you don't mind 09:29:32 <brtknr> flwang1: based on the discussion at PTG? 09:29:45 <brtknr> where you can define upgrade path in templates? 09:30:06 <flwang1> yep, for example, how many versions you're supporting? 09:30:31 <flwang1> what happened if user's cluster are still on a version which has been deprecated? 09:32:10 <brtknr> flwang1: we dont normally worry about this as most clusters are short-lived 09:32:39 <brtknr> users create the cluster templates and clusters on demand 09:32:51 <flwang1> ah, right 09:32:52 <flwang1> ok 09:34:32 <brtknr> flwang1: have you even thought about centos driver btw? 09:34:59 <flwang1> using centos as os? 09:35:05 <brtknr> flwang1: yes 09:35:15 <flwang1> i think blizzard did that 09:35:25 <flwang1> but they didn't contribute it back 09:35:33 <brtknr> flwang1: they never upstreamed the driver right? 09:35:41 <flwang1> yes 09:36:03 <brtknr> i am working on a use case which requires centos driver 09:36:07 <flwang1> they shared some personal code piece IIRC 09:36:16 <flwang1> interesting 09:36:31 <flwang1> you can reach out them and see if they would like to upsteam 09:36:45 <brtknr> because Mellanox OFED drivers cannot be easily installed on fedora atomic 09:36:48 <brtknr> or fedora coreos 09:37:17 <brtknr> so at the moment, i am using a kubespray cluster instead of Magnum 09:38:15 <flwang1> i see 09:39:05 <flwang1> anything else you would like to discuss? i'm going to offline now 09:39:31 <brtknr> thats all from me 09:39:42 <brtknr> I thought about something earlier but I cant remember it 09:40:16 <jakeyip> I have something... thanks brtknr for pointing out hca issue 09:40:19 <flwang1> :) no worries, you can leave message any time or just drop an email to me 09:40:33 <flwang1> #endmeeting