he’d said he’d used it before in the OP, otherwise I wouldn’t have recommended it.
Honestly Arch doesn’t really give you trouble if you set it up correctly the first time, but i get why it’s not for everyone for sure. I’m not currently runing it on my daily systems.
Ah, must have missed it.
I’m just starting to see that PCIe passthrough is getting so commonplace now that even “mainstream” distros have good support for it. I mean, if you can do it on Ubuntu with minimal obstacles, that’s a good measurement of “penetration”
With that said, it makes me wonder why anyone would choose the path of more resistance to achieve a nearly identical outcome.
doesn’t apparmor cause a bunch of issues with less basic setups though?
Yeah, I realize mainstream distros can do it (fedora is basically the upstream for virt-manager anyway)
I’d say at least for me, the best trait for a KVM hypervisor host is having access to all the prepatched stuff available in a repo from the get-go, which, in the past, was arch only. If there are up to date coprs for fedora now, it’s probably just as good an option.
Uh, I don’t really know. I know there are rules you need to add if you want to pass through block devices (/dev/sdc, for example) but I think VFIO and PCIe passthrough works fine without apparmor mods.
Yeah, I used to host a repo that had ACS, NPT and all the QEMU patches, but most of it’s been mainlined now, so I just fell back to the other dudes repo.
I’ve had several reports of issues and workarounds during my work at TPP, we generally tell people to avoid distros with apparmor on by default.
That and debian changes its kernel build/patch procedure like every major version.
what i’d really love is a version of fedora with Solus’ update schedule, or a good way to update centos to mainline + without manually doing all the nvidia drivers and cuda/cudnn