Hi,
Generally, Tor load-balances traffic to make sure that clients get consistently good bandwidth and latency. Adding significant extra load to some relays makes it hard to achieve this goal. In particular, 100% relay usage can significantly increase latency.
Sending 33% of client traffic to the 20% of guards that have IPv6 will cause an increase in usage for IPv6 relays (and a decrease for IPv4 relays). Even if we check the relay's IPv6 address after its IPv4 address, the traffic is still going to that relay. Tor clients on IPv6-only networks will also add extra to IPv6 relays. These clients don't work now, so we don't know how many there are. Once we've tested IPv6 in at least one stable release, we will have a better idea of the load balancing impacts of IPv6. And we can think about changing the address order. Another reason that we want to check IPv4 first is that is preserves tor's current behaviour. Adding an extra attempt at IPv6 once IPv4 has failed, is a smaller change that is unlikely to have any negative impacts. (Our last attempt at adding automatic client IPv6 didn't work, and we had to disable it in Tor Browser alpha.)
You're right, I opened a ticket to add IPv6 ORPorts to the example torrcs:
See my answer above: 100% capacity is not good for the network or clients. T |
_______________________________________________ tor-dev mailing list tor-dev@xxxxxxxxxxxxxxxxxxxx https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-dev