I am running a Fedora (Workstation 37) on a Framework Laptop (frame.work, 12th Gen).
I am using multiple Boxes for my work, which are ran through qemu by GNOME Boxes.
From time to time the Boxes performance completely crashes to a point where working is impossible.
The next day it will be fine again.
The task manager (it is a Win10 VM) says the “disk” usage is at 100%.
I cannot really figure out why this is happening since I am not very much of an hardware/OS expert.
I would be grateful if you could help me with tracing the problem and hopefully help to solve it.
I know I provide very little information in this post, since I don’t know what information can be helpful. I am happy to provide more of the information needed for solving this problem
Those multiple VMs you are running probably consume all allocated RAM over a period of time and start swapping, meanwhile your disk likely cannot handle the load spike which negatively affects performance.
You should analyze memory usage for both host and guests and reallocate available hardware resources to fit each OS needs, possibly adding extra RAM modules, or reorganizing your workflow.
However, in some cases storage performance can still be the bottleneck, then you need to delegate the task of virtualization to server grade hardware that can utilize RAID to distribute the load and improve performance.
Boxes by default will allocate all CPUs available to every virtual instance created through the application. If you haven’t already, the first thing I would do is look at the VMs you’re running concurrently and make sure the CPU allocation for each reflects the need of the virtual instance.