My guess is either:
-
A misconfiguration on your end (possibly package sources or third party packages that aren’t compatible with your distro release?)
-
A packaging bug in your distro that should be fixed
This worked, thank you for the help!
I just made a link to the missing libraries inside the executable folder and renamed the link to the versions it was looking for.
Boots up and works no problem.
I’m not doing anything important, just trying to play Blood lol.
Thanks for the other suggestions wertigon and ack, but I have a small brain and I’m going to take the bodge on this one.
1 Like
I find it interesting how some distros remember multi monitor configs better than others, when screen rotation and display profile sticks its a good thing.
On an annoying note, Dell Thunderbolt Docks randomly drop the 3rd monitor on a bunch of distros… good to see their firmware still sucks.
Why does lxc-create create empty rootfs?
I’ve tried creating a container using lxc-create -n mycontainer -B zfs --zfsroot=hdd/vm -t /usr/share/lxc/templates/lxc-download -- --dist debian --release bookworm. It does create it in a dataset mounted to /var/lib/lxc/mycontainer, says that it’s unpacking rootfs, but then rootfs directory is empty.
I can see nonempty filesystem image in /var/cache/lxc/download/debian/bookworm/amd64/default.
I will answer myself.
If you use ZFS as backing storage for LXC it will unpack the image into the dataset, but it will keep it unmounted, even when the container is running. How? Possibly magic.
What matters to me is that I can now easily create containers with lxc-create and use the resulting filesystem tree to run it with libvirtd.
Never used lxc…But are you observing mount namespaces?
Related: How to view the files of your LXD container from the host – Mi blog lah!
(Edit: I should reply about things I’ve never used less…)
1 Like
Thanks, I didn’t know about that.
OK so… I have a small-ish problem, that I don’t know if there’s actually a solution for.
A while back I bought a physical retail copy of Life is Strange Before the Storm (Limited Edition was physical only). Well, Square Enix in their infinite wisdom is apparently not able to just put a Steam Code into the box, it has to be more complicated.
Basically, the included DVD has a Key Generator and some Steam Data-packages on it. You start the key generator, enter a code that’s in the box (which is NOT the Steam Key), turn around 5 times, sacrifice a goat, and there you have a Steam Key. Luckily I had a goat handy.
The DVD is mounted and accessible in WINE via a mapped drive letter. Opening the keygen via WINE works just fine, but here is the problem: When clicking the button that’s supposed to generate the code, the keygen crashes, and the WINE logs aren’t really telling me what the issue is (it does reference some access violation).
I am assuming that the keygen wants to access some sectors on the disc that are outside the mountable filesystem, and of course WINE has no access to that because it only sees the files mounted on the host system rather then the raw disc (which would explain the access violation because it tries to access an address that doesn’t exist).
I zipped up the keygen and accompanying files and sent them to a friend, and it asks for the physical disc, which would confirm that suspicion.
Now, here’s the question:
Does anyone know of a way to pass through an entire drive to WINE instead of merely mapping a mounted path as an imitated drive?
I can’t seem to find any documentation on that and I have the suspicion that’s not going to work because WINE would need its own USB and/or SATA drivers to handle the drive. Any ideas?
And yes I know I could just spin up a VM but I’m curious if there’s a way without doing that.
Some time ago, I had to replace VMware with Proxmox for a friend of a friend, here are the details:
That works fine with SPICE, but it “has too many steps” to connect, and “it looks different than VMware”.
So I’ve been thinking… Is there a way to RDP into host, port forward to guest with interface connected to vmbr99 not bridged to any physical port on the host.
So far, I have:
- Proxmox host connected with 192.168.88.88 on MikroTik router
- vmbr99 with 10.0.100.0/24 address range
- Windows guest with manual IP setup: 10.0.100.10, netmask 255.255.255.0, no gateway or DNS
- Laptops that are used to remote in to VMs on same 192.168.88.0/24 network as Proxmox host
Someone tried to port forward to LXC here, but generally it was met with “you shouldn’t do that”.
I did find some guides, but they may be fairly outdated, so I have no idea what is going on.
General idea is:
enable forwarding:
net.ipv4.ip_forward = 1
add some rules:
iptables -t nat -A PREROUTING -p tcp --dport 50001 -j DNAT --to-destination 10.0.100.10:3389
iptables -t nat -A POSTROUTING -p tcp -d 10.0.100.10 --dport 3389 -j SNAT --to-source 192.168.88.88
And it should work, but it doesn’t. I did turn off firewall on Windows, enabled remote desktop, added authorized users, I even tried the same thing with Linux and web server (but forwarding to port 80, obviously) and it doesn’t work either.
What I basically want is laptop rdp → proxmox port 50001 → Windows guest port 3389.
I recall proxmox has its own firewall configurations (or its own routing between phys and vmbr99), these might be interfering (and what you’re looking at isn’t conceptually different to forwarding ports on an internet router which, all too regrettably works with rdp). but if we’re in the “it looks different to x” territory of friend of a friend user then, to me at least they’d seem to need a wallet to solve their own laziness.
the laziest solution (for you) is a jump host. RDP to that. it has it’s feet in two networks and mstsc icons on the desktop.
You may also be able to configure a user on proxmox to view/use the console (PVEVMUser role), then they just have a bookmarked url.
https://pve.proxmox.com/wiki/User_Management#pveum_users
1 Like
I just jury rigged pfSense CE VM to port forward for testing, and it does work. I’ll try and maybe replace it with Alpine or something but I’m inclined to just leave it like this.
Fedora has a borked kernel/mesa update and broke 2 of my desktops.
My daily driver is stuck at 1080p and was fixed by downgrading the kernel in the grub menu.
My kid’s machine wont boot into the gui and I’m trying to figure out how to fix it because choosing an older kernel from grub doesnt seem to fix it.
Update: I fixed my daily driver. I updated fedora on the borked kernel and it fixed it. Now to figure out how to fix my kid’s machine without nuke and paving it (the lazy way).
Update2: I got a CLI now and was able to update… to conflicting packages…
Update 3: fixed a problem by reinstalling the mesa-dri-drivers… whatever the hell is that package is…
manually updating everything except the broken packages
DRI is the Direct Rendering Infrastructure, it’s a framework for direct (graphics) hardware access under X.
Considering that, it’s also not unimportant to know whether you’re running AMD or Nvidia on that machine.
You might also wanna run a dnf distro-sync to make sure no packages are out of line.
Would also help to know what
are and what error DNF is showing.
You can pipe the DNF output to fpaste for posting.
1 Like
Thanks for this! Somehow the command wanted to remove grub2-tools minimal, systemd and system-udev? wth??? 
Do you think Silverblue would have saved me these kinds of headaches? I don’t understand why I have a frankebuild now. Gaming and YouTube is the only use case for my kid’s machine and maybe I did install xone (that needs dkms).
Two packages are broken (both arguing between i686 and x86_64 versions): expat and libusb1
also fpaste is amazing!
In my experience rolling releases just don’t seem to work very well in general without a lot of attention to avoid random brakeage.
2 Likes
I’ve had breakages everywhere, only fedora and ubuntu had the least issues I’ve had and I’d rather not use ubuntu.
It (probably) won’t anyway because they are protected packages (see ls /etc/dnf/protected.d). It’s very confusing why it still shows up, but from what I have read about it it happens when there are some unresolved dependencies. Why it pretends to uninstall a package is beyond me though because DNF doesn’t normally uninstall a package unless --allowerasing is given. It’s also possible it wanted to replace it with a new package providing the same thing. IDK, we’d have to see the entire DNF output for that.
Never used Silverblue so can’t speak on that. Theoretically it should because you can just go back to the old snapshot.
Both of these packages can have both Architectures installed no problem. Are they the same version? Sometimes it happens that the version differs because the Mirror isn’t in sync yet.
Fedora isn’t Rolling Release.
1 Like
I haven’t tried Silverblue. I might have if Bookworm hadn’t worked out as well as it did, but don’t really have any incentive now.
If you want to call it half-rolling or semi-rolling or whatever that’s fine I’m not particular on the details.
1 Like
Sure, but then you’d have to call Debian that too 
1 Like