ChanServ changed the topic of #freedesktop to: https://www.freedesktop.org infrastructure and online services || for questions about freedesktop.org projects, please see each project's contact || for discussions about specifications, please use https://gitlab.freedesktop.org/xdg or xdg@lists.freedesktop.org
ximion has joined #freedesktop
<DemiMarie> Has the use of Kata containers in CI been considered?
columbarius has joined #freedesktop
<__tim> yes
co1umbarius has quit [Ping timeout: 480 seconds]
<DemiMarie> As a security researcher I am curious how the attackers managed to break out of the sandbox. This would make for an interesting postmortem, but not until the flaw is patched!
co1umbarius has joined #freedesktop
columbarius has quit [Ping timeout: 480 seconds]
agd5f has quit [Remote host closed the connection]
jramsay has quit [Read error: Connection reset by peer]
lack has joined #freedesktop
Lyude has quit [Ping timeout: 480 seconds]
Lyude has joined #freedesktop
ofourdan has quit [Ping timeout: 480 seconds]
<mupuf> DemiMarie: the runners are still running containers as privileged
<mupuf> I'll let you figure out the rest ;)
<mupuf> We have been working towards moving away from privileged runners, but of course it regresses some projects... and admins didn't have the time to get there
<mupuf> At least, the attack made it easier for us to just... not care and ask devs to rework their CI pipelines
lack has quit [Read error: Connection reset by peer]
jarthur has quit [Quit: Textual IRC Client: www.textualapp.com]
lack has joined #freedesktop
Lyude has quit [Quit: Bouncer restarting]
Lyude has joined #freedesktop
Lyude has quit []
Lyude has joined #freedesktop
Lyude has quit [Ping timeout: 480 seconds]
<DemiMarie> Yeah, that will be the first step.
fr0hike has quit [Remote host closed the connection]
Lyude has joined #freedesktop
ximion has quit [Quit: Detached from the Matrix]
danvet has joined #freedesktop
alanc has quit [Remote host closed the connection]
alanc has joined #freedesktop
AbleBacon has quit [Read error: Connection reset by peer]
Lyude has quit [Ping timeout: 480 seconds]
Lyude has joined #freedesktop
i509vcb has quit [Quit: Connection closed for inactivity]
wermos has joined #freedesktop
lack has quit [Read error: Connection reset by peer]
wermos has quit [Quit: Page closed]
lack has joined #freedesktop
<alatiera> mupuf the only project that still needs priv is virgl testsuite iirc
<alatiera> otherwise all the gst runners are configured non-priv already
<alatiera> btw, slightly related https://github.com/containers/buildah/issues/4576
<alatiera> not sure if anyone less here had updated podman/build yet but in case the ci-template image builds stop working that's why
<daniels> alatiera: weston and mesa also use kvm
<alatiera> ah, but isn't that under qemu?
<mupuf> alatiera: thanks for the link, that may explain why valve infra had to reenable privileged mode after a buildah update
<alatiera> mupuf let's say it was a while until I've found out why myself too :P
<alatiera> we can fix that though through the gitlab-runner config
<alatiera> I think the key si "cap_add" in the runner toml
<alatiera> there's also another one to explicitly set the whole seccomp profile if needed
<mupuf> Thanks :)
<mupuf> Will revert to unprivileged when I come back to work
<mupuf> All the test machines are unprivileged
<mupuf> Only the gateway runners aren't... But they have pretty strong restrictions on who can submit jobs there
gchini has joined #freedesktop
lack has quit [Read error: Connection reset by peer]
lack has joined #freedesktop
<DemiMarie> If something really needs privileges (perhaps because it deals with privileged kernel APIs), it can be run in a KVM VM via Firecracker.
Haaninjo has joined #freedesktop
ximion has joined #freedesktop
psychon has quit [Remote host closed the connection]
<mupuf> DemiMarie: true that, it's just making CI pipelines more complex. Luckily, bentiss wrote vm2c (VM to container, a wrapper around my boot2container initramfs) that makes running VMs more manageable
<DemiMarie> mupuf: what I meant is that your container orchestration tool might have Kata container support
<alatiera> it does, but its not plug n play
<alatiera> supposedly you can use kata as an oci runtime, but a bunch of things in our setup need to be adjusted accordingly
* alatiera goes to mumble somewhere about oci
<mupuf> DemiMarie: ha, right. Never tested that
* alatiera spent a couple days yelling at the computer trying to get runner+kata working
<alatiera> didn't get far
<mupuf> alatiera: sounds like a sensible place to plug a VM
<mupuf> Oh boy!
<alatiera> half of my issues wouldn't exist if instead of tarballs+json it was some sensible storage format like DDIs
<alatiera> which would accommodate for the kernel+initiramfs we have to shove into the kata setup
<mupuf> Ddi?
<alatiera> discoverable disk images
<DemiMarie> alatiera: kernel+initramfs should be part of the runtime configuration. The management stack should not need to know about it.
<alatiera> DemiMarie but we also need custom images for kata in order to do things like virtio
<DemiMarie> alatiera: what do you mean? I thought the initramfs would deal with that.
ximion1 has joined #freedesktop
ximion2 has joined #freedesktop
ximion has quit [Ping timeout: 480 seconds]
ximion1 has quit [Ping timeout: 480 seconds]
ximion2 has quit [Quit: Detached from the Matrix]
robobub_ has quit []
ximion has joined #freedesktop
lack has quit [Read error: Connection reset by peer]
lack has joined #freedesktop
<DavidHeidelberg[m]> Something like this seems to be something we (as FDO) want: https://gitlab.com/marge-org/marge-bot/-/merge_requests/349 for Marge, right?
<DavidHeidelberg[m]> alatiera: btw. https://gitlab.com/marge-org/marge-bot
<DavidHeidelberg[m]> currently Marge waits for CI being in place and merging can start
<alatiera> DavidHeidelberg[m] the one I am using is https://github.com/smarkets/marge-bot/pull/270
<alatiera> DemiMarie I will have to check tmr
<DavidHeidelberg[m]> alatiera: you wanted say https://gitlab.com/marge-org/marge-bot/-/merge_requests/270 ? 😉
<alatiera> DavidHeidelberg[m] I don't think I should be reviewing python things other than the simple ones 😆
<alatiera> but I've been using that and its working
<alatiera> DavidHeidelberg[m] this will also help killing our gitlab https://gitlab.com/marge-org/marge-bot/-/merge_requests/362
<alatiera> not killing*
<__tim> wonder if gitlab has stopped sending out mail
danvet has quit [Ping timeout: 480 seconds]
i509vcb has joined #freedesktop
<DavidHeidelberg[m]> alatiera: nice, I also pinging the another company who contribute (but in their company named fork, so we'll see if they join)
systwi_ has quit []
systwi has joined #freedesktop
Haaninjo has quit [Quit: Ex-Chat]
krushia has quit [Ping timeout: 480 seconds]
krushia has joined #freedesktop
lack has quit [Read error: Connection reset by peer]
lack has joined #freedesktop