@RaHu

In theory you would want to change the MTU of parent interface in OPNSense to 9000. In my case testing is not possible since I run all vlans off a since interface (2x10gbe NIC) and do not want to change the MTU of systems outside my virtual environment. Anyways, with the change mentioned above you achieve 10gbe speeds between VMs.

@TJ3f47501

This has been resolved, just deployed a new XOA appliance to the correct network and deleted the old one.

@olivierlambert said in How to get 10gbe speeds on guest VMs?:

OpenvSwitch

Found out I was missing setting the MTU of the 10gbe PIFs on my XCP-ng server. "xe pif-list" to find the <network-uuid>, then "xe network-param-set uuid=<network-uuid> MTU=9000" Found out the hard way you need to power off all VMs that have any attachment or simply unplug those NICs for the change to take. After that, configured the MTU of my guest test VM to 9000 and ipef3 testing is now at 10gb.

New to XCP-ng and am having trouble determining why guest VMs cannot complete an iperf3 test at 10gb. Running this same test on physical servers or guests in VMware works properly. On a XCP-ng guest VMs this test maxes out at <3gbps. Are there any obvious settings I may be missing? Currently the VIFs I'm testing with have an MTU of 9000, which reports correctly on the guests. I am questioning the "Pool-wide network associated with ethX" setting which the MTU of 1500, which may be overriding my VIFs settings? Iperf3 command I'm using is: (iperf3 -c x.x.x.x -p xxxx -t 60 -P 10). Any help would be appreciated.

New to XCP-ng and want to change the physical interface initially setup for management. I was able to do this for XCP-ng, but cannot find any documentation on how to do this for XOA CE. Any hep would be appreciated.