I’m pretty sure this is a tailscale related issue mainly because I can move one of the machines so they are in the same intranet. In this case my replication task completes normally.
When I put them in different places and connect them using tailscale’s subnet, they get stuck.
When using tailscale’s subnet, the two machines can ping each other. If I connect directly using ssh in the shell, the two machines can also connect to each other.