"
I tried reducing the MTU to 1360 but the issue persists. I could be wrong but it seems when I lower MTU the rsync transfer sustains higher speeds for longer. I can transfer around 12G before the speed drops.
While I can't try something with bare metal at the moment, I did experience the same behavior with a Debian LXC container instead of OPNsense on the same Proxmox host with the same site to site configuration so the issue is likely related to the double NAT or the virtualization host.
I did not passthrough the NIC. I am using virtio networking as described in the HOWTO post. I am not sure what you mean by "RSS in OpnSense and on the VM NIC settings." I have tried with and without multiqueue set to the same number of vCPUs which gives the same results.
Here are the VM settings:
Hardware:
VM options:
While I can't try something with bare metal at the moment, I did experience the same behavior with a Debian LXC container instead of OPNsense on the same Proxmox host with the same site to site configuration so the issue is likely related to the double NAT or the virtualization host.
I did not passthrough the NIC. I am using virtio networking as described in the HOWTO post. I am not sure what you mean by "RSS in OpnSense and on the VM NIC settings." I have tried with and without multiqueue set to the same number of vCPUs which gives the same results.
Here are the VM settings:
Hardware:
VM options:
QuoteWireguard uses UDP and does not benefit from TCP buffer algorithms, so you might also try to use traffic shaping to ensure that buffers are not overrun.Do you have any guidance on doing this?
