Hi,
I’m trying to use Zerotier with Talos Linux using the system extension for Zerotier, but I’m having issues with high packet loss.
All my Talos Linux nodes (3 control-plane, 4 workers) are joined and authenticated to my Zerotier network, and have retrieved their respective IPs.
As it isn’t possible to run arbitrary commands on Talos Linux I have been unable to check whether or not the connections between the nodes are direct or relayed.
What I have been able to do is join an RHEL VM and a Fedora laptop running on the same network as the majority of the nodes to the same Zerotier network, to use for debugging.
From the two debug machines all connections are direct. When they ping each other there is a 0% packet loss. When they ping the Talos Linux nodes there is between 10% and 25% packet loss when using the Zerotier IPs and <0,5% packet loss when pinging the nodes “real” IP.
When I ping one of the Talos nodes’ Zerotier IP from a debug container running on another Talos node I get between 16% and 95% packet loss. When I ping the “real” IP I get no significant packet loss, like above.
If it matters I’m also running Cilium on the Talos nodes, which I’m trying to get to use the Zerotier network as the backing connectivity between the nodes.
All nodes are running Zerotier 1.14.2.
I’d love to know if you have any tips for how to proceed with debugging this issue.