14:32:38 #startmeeting Fedora CI SIG 14:32:38 Meeting started Wed Aug 12 14:32:38 2020 UTC. 14:32:38 This meeting is logged and archived in a public location. 14:32:38 The chair is jbair. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:32:38 Useful Commands: #action #agreed #halp #info #idea #link #topic. 14:32:38 The meeting name has been set to 'fedora_ci_sig' 14:32:44 .hello jimbair 14:32:45 jbair: jimbair 'Jim Bair' 14:33:09 .hello siddharthvipul1 14:33:10 siddharthvipul: siddharthvipul1 'Vipul Siddharth' 14:33:21 #topic centOS Infra ticket 35 14:33:33 #link https://pagure.io/centos-infra/issue/35#comment-669695 14:33:39 jbair: right! so just to reiterate 14:33:58 Saffronique: and I were discussing potential dangers from using the current method (in the comment) 14:34:27 bgoncalv: ^ 14:34:28 in the future we would like to see if we can move to a more sustained way of it (atleast start a conversation and see its feasibility) 14:34:45 .hello2 14:34:47 bgoncalv: bgoncalv 'Bruno Goncalves' 14:35:26 but as I said, we think we will come back in the next meeting with a proper ticket and points and then discuss more on it to see what you all think 14:35:55 .hello2 14:35:57 Saffronique: Sorry, but you don't exist 14:36:02 :< 14:36:03 I appreciate looking into a solution as well :) 14:36:07 Saffronique: do you have a FAS ID? 14:36:15 yep, dkirwan 14:36:17 if so you can use .hello *insertfasnamehere* instead 14:36:22 .hello dkirwan 14:36:23 Saffronique: dkirwan 'David Kirwan' 14:36:27 we should need need to use privileged containers once we start using tmt to run the tests 14:36:40 the problem is when don't know when this move will be possible 14:38:19 bgoncalv: did you mean you shouldn't need to use or you are not using privileged containers now 14:38:21 ? 14:39:00 I'm using privileged at moment, but in the future when we use tmt to run the tests it won't be needed 14:39:06 ah, I see 14:39:57 Saffronique: do you have something to add or you want to wait for next meeting? 14:40:27 bgoncalv: thanks a lot for that info.. that's a relief (also, we are not really against privileged containers, but mainly a better way to use those) 14:40:43 maybe leaving things as-is (so long as we make sure to revisit later) is okay, then we simply remove when tmt goes live 14:40:44 :) 14:40:58 as a re-worked process is already in the works 14:41:20 Well ideally, would need a timeline as to when this move to tmt will be complete, before understanding if its worth changing the tests to use Kubevirt operator rather than privileged containers 14:41:30 I'm working on the re-work, but the re-work doesn't rely on tmt, so we need privileged containers... 14:41:38 But if tmt move is months or years, we should talk further 14:42:29 bgoncalv: I think tmt is months but exact timeline I don't know; do you? 14:42:55 yeah, I don't know when the move will be done, but I guess it is not too soon as I was asked to re-work the pipeline and not count on tmt for it 14:44:12 okay, well we should probably set some timelines for go/no-go type of actions 14:44:35 Saffronique: did you want to talk about the work-around in the next meeting in 2 weeks? 14:44:45 sure I can do :) 14:44:54 :D 14:45:33 we've experimented with it internally here, so we have a rough idea how to control VMs via this operator. It shouldn't be massively different to what you are currently doing 14:45:57 okay, I'll add it for the next meeting in 2 weeks to discuss =) 14:46:06 cheers 14:46:17 Thank you jbair and bgoncalv 14:46:25 Saffronique: thanks a lot of discussing it 14:46:43 that's all from out side :) jbair unless you are stuck on something because of us 14:47:18 Saffronique: on jenkins we create a qcow2 dynamically and then use it to boot, after the pipeline is over the qcow2 is not archived 14:48:02 ok, so you have persistent storage attached to the jenkins, probably what 50-100gb ? 14:48:14 you can bring up VMs via Kubevirt via an image 14:48:56 Lets open a ticket with the ask anyway, it would be good to capture your workflow so we can try replicate it using kubevirt 14:49:03 I'm not sure about the jenkins configuration. jbair ^? 14:49:57 we have 100GB attached to Jenkins for it's homedir 14:50:26 * siddharthvipul steps away for 2 minutes, brb 14:50:29 we create a qcow2 on a directory under WORKSPACE 14:51:07 then use standard test roles to provision it for us, it uses kvm 14:51:22 hmm i think it should all continue to work so far anyway 14:51:26 and finally use ansible to run the tests 14:51:35 teh biggest change is you would make an api call to kubernetes rather than to kvm 14:51:57 the kubevirt operator would then make the calls to kvm on your behalf 14:52:09 we don't call kvm directly, we should use https://pagure.io/standard-test-roles/ to provisin qcow2 14:56:50 https://kubevirt.io/user-guide/#/creation/creating-virtual-machines ok so the change there then would be to instead make a call to the Kubernetes api like so 14:58:33 that is the thing, the provisioner must be standard-test-roles 15:02:33 do we need to have any action items for the next time we talk about it, or did we just want to mull it over and reconvene in 2 weeks to discuss again? 15:04:59 I'll work with siddharthvipul and try get a ticket opened first I guess, we can collaborate on there> Yep we can reconvene at the next meeting! 15:05:01 I know it's 5pm in Brno so I don't want to hold bgoncalv hostage :) 15:05:10 okay perfect 15:05:18 ok 15:05:19 I'll make a note in the etherpad for next meeting to check in 15:05:21 thank you again y'all 15:05:25 if you want to add the ticket to the page that would be great 15:05:36 https://etherpad.gnome.org/p/fedora-ci 15:06:37 #endmeeting