Modding the virtual machine parameters

[synologix]

Hi,

 

- I have a DVB-Sat-Card in my System
- Until yet i have used ESXI on my System and 2 VMs 1. XPEnology 2. VDR Server
- With latest DSM we have a QEMU based VM inside the Synology

 

My Idea is Stopping my ESXI Project and Using XPEnology as Baremetal with the Buildin Synology VM

 

The project stops on the point that Synology has not enabled Passthrough.

But based on the knowledge they use QEMU / libvirt it should be possible to mod the VM config files

 

Has Anybody tried to modify the Settings of GUI generated virtual machines with modding the config files on file Base?

 

i have checked for IOMMU working or not (yes it works):
dmesg | grep -e DMAR -e IOMMU

 

Running Service:
/var/packages/Virtualization/target/usr/local/bin/libvirtd -d --listen --config /etc/libvirt/libvirtd.conf
/usr/local/bin/qemu-system-x86_64 -name fabf0876-7580-............................

 

so checked on console :
qemu-system-x86_64 --h 

 

anybody knows where the config parameters are stored and how to modify / expand it?
 

Best Regards
Synologix

 

[haldi]

9 hours ago, synologix said:

The project stops on the point that Synology has not enabled Passthrough.

But based on the knowledge they use QEMU / libvirt it should be possible to mod the VM config files

They do at least support USB Passtrough now.

 

 

But hey, if you manage to get a PCI-E Passtrough working.. i've got a old Quadro lying around here, and my HP Gen8 got a free slot, so that would be nice.

[sramshaw]

Hi, I have tried to do similar things on a Synology DSM v7.2. , and landed here so I thought I might leave some help for others interested in such topics.

 

Pci passthrough is best evaluated by:

- creating and running a VM so that libvirt loads all drivers (as observed in service pkg-libvirtd 's definition )

- connecting via ssh and checking capabilities reported by libvirt with the command virt-host-validate

 

In my case on DS1621+ I see that the system is capable of PCI passthrough, it is only the hypervisor (DSM's VMM) that does not expose the feature.

Instead of using VMM to create the boot time PCI passthrough, I resorted instead to have a hotplug of PCI device approach, which means modifying the VM once it is running.

This is not a perfect solution, but seems to work one time after each NAS (re)boot.

 

More on this at https://github.com/sramshaw/pci_coral_on_synology