1st attemp at VM, having network issues

[Vanse01]

This is my first attempt at VMs under Unraid. I have 2 unraid servers both similar in hardware configuration.

I will list both as I have tried to make the same VM on both with the same issue.

 

Main Server:

MB/CPU - MSI z390 carbon pro/intel I9 99ook

64 GB DDR4 ram

Nvidia RTX 4060 8gb low profile

LSI 9300-16i HBA

Mellanox 10gb SPF+ NIC

 

Backup server:

MSI z97-sli/Intel i-7 4790

32 GB DDR3 ram

Nvidia RTX 3060

LSI 9200-8i

Mellanox 10gb SPF+ NIC

 

so as I said I have tried on both but stopped working on the main server when I had issues so I will just go forward with tests on the backup until I get it right. I am assuming the solution to 1 will be the solution to the other as the problem is the same and the hardware is so close.

 

When I create the VM I leave most things default with the following exceptions

2 cpu cores

2048mb ram

and I change the primary vdisk location to manual and point it at the Home assistant image file.

I leave the network source as br0

and the network modle virtio-net.

 

when I start the VM with console, HA boots up and grabs an IP address from my router but the unraid GUI becomes unresponsive via network. I can still log into the console to shutdown or if i happen to be in the GUI boot mode everything still works but no network access from unraid os.

 

I am thinking that the default VM setting are somehow conflicting with my Mellanox Nic card and are keeping it from working for unraid OS

 

when I go into Tools>system devices, my Mellanox NIC shows up like this

[15b3:6750] 03:00.0 Ethernet controller: Mellanox Technologies MT26448 [ConnectX EN 10GigE, PCIe 2.0 5GT/s] (rev b0)

would that conflict with the default VM PCI device

 

   <controller type='pci' index='3' model='pcie-root-port'>
      <model name='pcie-root-port'/>
      <target chassis='3' port='0x12'/>
      <address type='pci' domain='0x0000' bus='0x00' slot='0x02' function='0x2'/>
    </controller>

 

I am not sure how the address found in the system devices translates to th XML of the VM  but it seems to have 5 entries similar that index 1 thru 5 us incement values in the port (0x10 - 0x14) and function(0x0 - 0x4) fields. the domain bus and slot fields are the same for all 5

 

I would like to have everything work thru the one 10gb NIC but I also have an unused 1gb network port on the MB that could be used for VMs if there is some advantage to that.

 

I currently the onboard NIC unplug but it is configured as bonded listed as active backup.

 

In an ideal configuation the onboard 1gb would be a failover for the 10Gb and both unraid OS and VM would go thru whatever is available but for now if i can just get the VMs working on one or the other I could progress.

 

I did search around a little but did not find a discription of how the PCI device address of physical devices are converted to the XML format of the VM config file. the few posts i did see were about passing a GPU thru to the VM which will be the next project on a 3rd server after HA get setup on the main machine. So hopefully I can learn from my mistakes here for future projects.

 

Thanks in advance for any help the community can provide

Andrew

[Vanse01]

I found the problem/solution.

 

It was a problem with the way I was assigning the VLAN to the server. I was using VLAN override on the client settings rather than changing the VLAN on the port it is connected to. 

 

I changed the port default VLAN and turned off VLAN override and now I can boot the VM, it gets its own IP and can access the network.