Had the same issue! Switched to `--send-retries 2` and bumped up the socket buffer sizes. Not perfect, but way fewer timeouts.
For IPv6, nsock *can* work, but you gotta babysit it. Check your kernel logs for weirdness.
Also, `strace` is your friend—see where nsock is spending all its time.
For IPv6, nsock *can* work, but you gotta babysit it. Check your kernel logs for weirdness.
Also, `strace` is your friend—see where nsock is spending all its time.
