I have two VMs currently configured. I can usually start and run both of them concurrently after a clean boot, but if the system has been running for a while I can often only start one or the other.
TN 24.10.2.2 on bare metal. AMD Ryzen 9 7900 12-Core/24-Thread Processor, 128 GB RAM
VMs;
A) Wubdiws 10: vCPUs 1, Cores 6, Threads 1, Memory Size 8 GiB
B) Linux Mint 22.1: vCPUs 1, Cores 6, Threads 1, Memory Size 32 GiB
VM A has PCI Passthrough of one of the USB controllers; VM B has PCI Passthrough of a GPU. Removing the PCI passthrough devices and/or increasing the vm.nr_hugepages setting did not seem to help.
I do have a few Apps also installed and running, but right now, after a cold boot with both the above VMs running the system Memory shows; Free 74.1 GiB, ZFS Cache
5.1 GiB, Services 45.8 GiB
The error I get when trying to start the second VM (sometimes) is;
CallError
[EFAULT] internal error: qemu unexpectedly closed the monitor: 2025-05-25T14:03:58.998870Z qemu-system-x86_64: -device {“driver”:“vfio-pci”,“host”:“0000:11:00.4”,“id”:“hostdev0”,“bus”:“pci.0”,“addr”:“0x7”}: VFIO_MAP_DMA failed: Bad address 2025-05-25T14:03:59.014366Z qemu-system-x86_64: -device {“driver”:“vfio-pci”,“host”:“0000:11:00.4”,“id”:“hostdev0”,“bus”:“pci.0”,“addr”:“0x7”}: vfio 0000:11:00.4: failed to setup container for group 33: memory listener initialization failed: Region pc.ram: vfio_dma_map(0x564d279e3020, 0x100000000, 0x140000000, 0x7f19f3e00000) = -2 (No such file or directory)
Does this have something to do with contiguous memory? That after the system has been running for days or weeks, although there may be ample room to start the second VM, there is not enough contiguous space to do so? Or something similar. Is there some way I can clean up or defragment the RAM without rebooting?