What are the exact CPU settings for your Whonix-15 Workstation settings? In virt-manager, under “CPU” tab, you should have only 1 CPU and CPU model set as host-passthrough (see screenshot below)
If it’s not the case, modify accordingly and try it out. Does it still freeze?
Please copy your Workstation .xml file content (on your host do sudo cat /etc/libvirt/qemu/Whonix-Workstation.xml and copy everything here).
Does the Whonix-Gateway work correctly? Did you also update it to Whonix 15 ?
If you create a new VM from scratch using the Workstation disk, does it also freeze (you can make a temporary copy of the disk, and do not forget to remove any virtual network hardware from your newly created test machine before booting it to avoid any leak!)?
Just posting an update (It is my thread after all. :P).
I don’t really have random freezes anymore on the VM but I still do have boot time freezes.
Those were gone when I wrote my last messages but came back not too long after.
I’m just posting now because I noticed something interesting in my case:
Part of that bug is the VM not resizing to the Window VM (Spice usually allows you to set custom resolution that fits virt-manager). I managed to run xrandr before the whole VM froze (Though my mouse was already invisible by that point) and noticed that it said all of my screens were disconnected.
I found this odd as whenever the VM works this is never an issue, but this time the terminal said “Virtual-0 disconnected primary” and the VM stopped responding shortly after.
I’ve always been convinced, at least I was when I originally posted the thread, that the VM doesn’t actually die (CPU usage staying normal and some other stuff) so maybe there’s just some problem with the whole “disconnected output”.
Anything I should look for, should this happen again?
I’ve always been convinced, at least I was when I originally posted the thread, that the VM doesn’t actually die (CPU usage staying normal and some other stuff) so maybe there’s just some problem with the whole “disconnected output”.
Anything I should look for, should this happen again?
Check logs or run some program that continue produces output to confirm
that after no more visible graphics, that system is still running normally?
Login via serial console as early as possible.
To interact with the Whonix-Workstation ™ via console, run.
I wasn’t able to connect to the console due to having the following error:
Connected to domain whonix_workstation
Escape character is ^]
error: internal error: cannot find character device
That being said, I setup a script that’d run on startup and I can confirm that it was running even after the screen froze.
EDIT: Edited the startup script and ran it again and I noticed something kinda funny.
When the screen is “usable” but doesn’t resize, xrandr reports the following thing:
Virtual-0 disconnected primary 1024x768+0+0 0mm x 0mm
That in itself is pretty funny because… well, screen disconnected?
However, when I run the same command and the screen dies, xrandr seems to output the following bit
Virtual-0 disconnected primary
As you can see half of the previous output is just gone. I have no idea what’s up, just thought I’d report.
If both Debian 10 and Whonix are effected this is not an Whonix issue. This is a Debian or hypervisor or hardware(?) issue. Whonix 15 is based on Debian 10. Does this happen with other guests (other thatn Debian ) in KVM? You might want to also look around for bug reports of this nature for Debian. Not sure if I already asked but are you having this issue with a Debian 10 host.
I would avoid changing screen resolution or selecting the autoresize with window resolution setting as this will make the GUI not respond to input. The only recommended way is documented on the wiki. Select the resoltuion from within the VM that best fits your screen then take a snapshot to preserve it
Just ran some tests again, doesn’t seem to matter much.
Even without the autoresize, it still randomly says “Virtual-0 disconnected primary” on boot and the VM freezes.
It would be helpful if you can answer Patrick and 0brand’s question about what system host and package versions you are using. Rught now I can’t see this on Debian stable.
@HulaHoop my apologies, for some reason I thought the question was for Ibr0z, not myself (I didn’t read every answer that’s been posted since my update a good while ago).
I’m using Arch Linux, stable (and updated) Kernel.
As for the packages I use, I download the following (and their dependencies, but I’m unsure how I could list said dependencies… I’m sorry.) to make libvirt/qemu/virt-manager work:
Everything is updated multiple times a day so I’m always running the latest version that’s available on Arch Linux package mirrors.
@0brand I’ve tried installing other OSes to test all of this out but for some reason I haven’t been able to do so, my installs always error’d out at one point or another. I’ll try again soon.
I’ll see what I can do, might not be able to do much before this weekend though (Especially since I’d have to setup things the right way, I don’t just want to redownload Whonix, I’d much prefer copying my actual files to see if i can reproduce the bug in a “similar” setup).
But to be honest it pretty much has to be a problem specific to the workstation and my host, if it was just about whonix_workstation, someone else would’ve noticed by now.
Is there anywhere, anything I could look at to get a bit further? Even if the problem affects few people I’d rather figure out a way to fix it, but I’m not sure where to start.
Alright let’s see fi this helps. It is years old and probbaly not your problem, but try adding a USB tablet device to your VM virtual hardware and see if anything improves.
Updating the thread after some time (Sorry about that…).
I can’t be sure about my findings long-term since I don’t know why exactly all of this happens but after another recent update (Unsure if host or VM package) broke the VM again (xrandr saying “disconnected” for all inputs) I did something simple and changed the “Video” component in virt-manager from QXL to Virtio and it seems to work so far…
VGA works too but it’s nowhere near as smooth for some reason so yeah.
I tested that before and it didn’t work so I’m not sure what’s changed exactly.
Update of hypervisor pacages or do you mean Whonix version?
Haven’t figured it out is what I meant, sorry for not being clear. I have no idea which of an hypervisor or a Whonix update broke it again, and by broke I mean that the problem’s apparently caused by the VM not having proper video input (Which was solved by getting rid of QXL)
Good because this is now default in the latest test release.
