16:00:56 #startmeeting RELENG (2019-03-28) 16:00:56 Meeting started Wed Mar 27 16:00:56 2019 UTC. 16:00:56 This meeting is logged and archived in a public location. 16:00:56 The chair is mboddu. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:00:56 Useful Commands: #action #agreed #halp #info #idea #link #topic. 16:00:56 The meeting name has been set to 'releng_(2019-03-28)' 16:00:56 #meetingname releng 16:00:56 The meeting name has been set to 'releng' 16:00:56 #chair nirik tyll sharkcz masta pbrobinson pingou puiterwijk maxamillion mboddu dustymabe ksinny jednorozec 16:00:56 #topic init process 16:00:56 Current chairs: dustymabe jednorozec ksinny masta maxamillion mboddu nirik pbrobinson pingou puiterwijk sharkcz tyll 16:02:06 * ksinny waves 16:02:11 * nirik is here, but also in another meeting 16:02:41 nirik: Hehe, same here :) 16:02:56 So, lets get started 16:03:14 #topic #8165 new koji tag for Fedora CoreOS continuous builds 16:03:22 #link https://pagure.io/releng/issue/8165 16:03:41 nirik: So, are you okay with the FBR for koji hub policies? 16:03:55 Just want to know what you think 16:04:34 I guess so... 16:05:30 nirik: Okay, I guess what other people will think of the FBR, but we could try and see what happens 16:05:38 +1 16:05:45 s/guess/dont know/ 16:05:50 sure, we should also see perhaps if rc8 is it... if so, less risk 16:07:19 nirik: Sure, but if we find another blocker (I hope not, /me knocks the wood), then they have to wait another week or risk it more for RC 16:07:22 It should be fine to apply changes after GO/NO GO meeting tomorrow. Meanwhile nice to have FBR in place to get +1 16:09:19 yeah. 16:13:57 #info mboddu will work on koji side of things and will send an FBR for koji hub policy change (probably will need some help from Kevin). This will be applied after the Go/No-Go meeting tomorrow. 16:14:34 #topic #7840 Consider ursa-major 16:14:40 #link https://pagure.io/releng/issue/7840 16:15:12 I think we have a plan that many folks like... 16:15:23 I wanted to bring this up to check whats the status on epel8 and should we close this ticket as I understand it, we are doing this in a different way rather than ursa major 16:15:41 we are not going to use ursa-major... instead of that we will: 16:15:47 nirik: +1 16:16:01 * add a new variant thats the modules that we want in the buildroot. 16:16:14 * use that as a external repo with koji for the buildroot. 16:16:26 and more details, but thats the gist of it. 16:16:41 it needs some releng work, some koji work and some mergerepos_c work 16:17:08 nirik: Do we have the plan documented somewhere? 16:17:53 there's a doc. let me look. 16:18:49 I think it was going to be presented to fesco then actioned 16:19:04 so I would say leave this until after fesco approves it 16:19:30 nirik: Okay, I remember seeing it, but forgot where it is :D 16:21:17 #info We are not going to use ursa-major, yay for that. Instead we will add a new variant of modules that we want in the buildroot and use that as an external repo in koji for the buildroot. This requires releng work, infra work, koji work and some mergerepos_c work 16:26:02 #topic #7403 Consider no longer producing the Workstation repo tree in composes 16:26:06 https://pagure.io/releng/issue/7403 16:26:39 So, I wanted to bring this up and see what can be done here, since we are now looking at improving our compose process 16:27:24 I still dont understand why anaconda needs different install trees 16:30:51 well, it does when it makes install media I think it is... 16:31:09 so for workstation we could drop the tree, but not server (it has a dvd) 16:31:21 but that also means workstation boot.iso will be gone. 16:32:30 nirik: Can anaconda not use the Everything install tree somehow? 16:33:10 it can for non install stuff... 16:34:13 but installs have different anaconda options... like server uses xfs by default, etc. 16:34:34 anyhow, I might be misremembering... 16:34:51 but I think we can drop the workstation one if we want to and if they don't care there's no boot.iso for workstation 16:35:02 nirik: hmmm, I wish anaconda can do install stuff as well from Everything install tree 16:36:27 nirik: I dont know, I will talk to lsedlar and see how much it will save in the compose process, if not much, then we can leave it there 16:36:58 he says not much in comments there... 16:37:13 like... 10-20min 16:38:09 nirik: Hmmm, not much then, and we know its stable without issues. So, I guess we could leave it there then 16:38:24 I'm pretty sure that we (the workstation working group) have decided that we don't care about a workstation-specific boot.iso though I'm not immediately finding the meeting discussion on that 16:39:08 yeah 16:39:25 https://pagure.io/fedora-workstation/issue/45 16:39:26 so I think it's fine to drop the workstation one (well, for rawhide/f31 anyhow) 16:40:22 otaylor: Awesome, then I will make the changes in f31, I guess it might need a change request 16:41:01 nirik: I will create a change request and once approved, I will make the necessary changes 16:41:04 might need some openqa changes if they test the workstation boot.iso 16:41:15 and deliverables wiki page 16:41:22 sure, sounds good 16:41:31 nirik: Yup 16:42:32 #info mboddu will file a change request for removing workstation install tree for F31. Once approved, he will make the necessary changes like pungi config, openqa, deliverables wiki page... 16:44:12 #topic Alt Arch Updates 16:44:22 nirik: you wanna talk about power9? 16:44:27 sure. 16:44:30 How much you enjoyed working on it ;) 16:44:46 ha. Actually it's pretty nice hardware. 16:45:05 However, I tried to move image builders to it and they fail to start vm's. 16:45:33 what's the error? 16:45:38 I fear we may have to reinstall them as fedora... rhel7's kernel doesn't support nested virt with the kvm module they use. 16:46:15 kvm-hv: Parent hypervisor does not support nesting (rc=-2) on 16:47:12 yeah, we need latest virt stack to get it working 16:47:24 https://koji.fedoraproject.org/koji/taskinfo?taskID=33784478 16:47:52 well, and kernel it sounds like. 16:48:29 anyhow, we can get it sorted... 16:48:50 in the mean time I put the image builders back on the power8 boxes with all their various downgrades and bandaids 16:50:26 #info nirik tried to move image builders to power9, he faced issues with nested virt with the kvm module. So, he reverted them back to power8. It could be fixed by reinstalling fedora with latest virt and kernel. 16:50:30 nirik++ 16:50:46 #topic Open Floor 16:50:58 I had one quick item: koji updates 16:51:09 nirik: Go ahead 16:51:42 koji 1.17 is out... after freeze we should update to it. I don't think we will have any big problems, it looks like a pretty smooth release... 16:52:10 there's still work needed on oz and imagefactory (both have pending pr's already) to move to python3, then koji can go python3 only also 16:53:23 Also, I have an update on koji 16:54:06 #info koji 1.17 is released and will be updated after beta freeze. We are expecting it will be a smooth update. 16:55:14 nirik: So, I talked to dgregor about the vm's not getting killed, he said he will look into it and he is considering it as a medium high priority task. So, I am hoping we will get it fixed soon 16:56:06 cool. 16:57:56 fyi, looks like the oz python3 work is merged, just needs a new upstream release. 16:58:26 #info Fedora 30 Beta Go/No-Go meeting is tomorrow, there is a proposed blocker based on how it will be considered, we might go for a release next week. 17:00:03 yeah, I don't thinkt that proposed blocker is likely to be a blocker. ;) 17:00:12 * mboddu is not considering it as a blocker 17:00:19 nirik: Yup same here :) 17:00:28 Anyway, thats all I got 17:00:43 And we are right on time 17:01:05 #info oz python3 work is merged, just needs a new upstream release 17:01:14 Thanks everyone for joining 17:01:21 #endmeeting