I chose that number because I figured that was all I needed for now, my plan is to add more as needed and as my DNS already points a /112 of the /64 I didn't want to have to redo it. I could easily use the entire routed /64 and that would be fine if that is easier I just didn't know the performance impact of having that many IP addresses on the machine and didn't really need them so I figured I would leave them available until needed but yes I could easily use the entire /64 and I'm not really that opposed to it.
A /64 is more "normal" as the IPv6 world uses that as the smallest routeable block. I don't think performance will be an issue either.
Here's a traceroute to the address you listed
[carl@mars ~]$ traceroute6 2001:470:8:920:1:1:1:1
traceroute6 to 2001:470:8:920:1:1:1:1 (2001:470:8:920:1:1:1:1) from 2001:470:c27d:e000:20c:29ff:fe8a:1618, 64 hops max, 12 byte packets
1 2001:470:c27d:d000:2e0:81ff:fe79:f4c4 1.465 ms 0.856 ms 0.715 ms
2 servicespring-1.tunnel.tserv9.chi1.ipv6.he.net 48.412 ms 48.183 ms 47.484 ms
3 gige-g3-4.core1.chi1.he.net 81.694 ms 45.169 ms 45.440 ms
4 10gigabitethernet2-4.core1.nyc4.he.net 68.630 ms 74.351 ms 70.995 ms
5 10gigabitethernet2-3.core1.ash1.he.net 73.660 ms 73.243 ms 92.107 ms
6 gige-gbge0.tserv13.ash1.ipv6.he.net 108.317 ms 76.671 ms 78.062 ms
7 * * *
Do you have another address on that /64 that we can test? I would like to rule out a routing issue at HE by testing another.
This is what I get:
debian:~# ping6 2001:470:8:920:1:1:1:2
PING 2001:470:8:920:1:1:1:2(2001:470:8:920:1:1:1:2) 56 data bytes
64 bytes from 2001:470:8:920:1:1:1:2: icmp_seq=1 ttl=62 time=472 ms
64 bytes from 2001:470:8:920:1:1:1:2: icmp_seq=2 ttl=62 time=445 ms
64 bytes from 2001:470:8:920:1:1:1:2: icmp_seq=3 ttl=62 time=469 ms
--- 2001:470:8:920:1:1:1:2 ping statistics ---
3 packets transmitted, 3 received, 0% packet loss, time 2007ms
rtt min/avg/max/mdev = 445.090/462.175/472.413/12.159 ms
debian:~# traceroute 2001:470:8:920:1:1:1:1traceroute to 2001:470:8:920:1:1:1:1 (2001:470:8:920:1:1:1:1), 30 hops max, 40 byte packets
1 mcgurrin-2.tunnel.tserv4.nyc4.ipv6.he.net (2001:470:1f06:cef::1) 289.626 ms 292.568 ms 295.184 ms
2 gige-g3-8.core1.nyc4.he.net (2001:470:0:5d::1) 295.674 ms 295.745 ms 295.811 ms
3 10gigabitethernet2-3.core1.ash1.he.net (2001:470:0:36::1) 301.438 ms 301.595 ms 301.140 ms
4 gige-gbge0.tserv13.ash1.ipv6.he.net (2001:470:0:90::2) 302.013 ms 302.928 ms 295.066 ms
5 1.vpslime.mcgurrin.net (2001:470:8:920:1:1:1:1) 320.683 ms 320.864 ms 321.055 ms
Try changing the last 1 to anything 2-8 as those are the others currently set up if you want a different address, I will be setting up more later today probably. The whole block is from HE though so I don't know if that will help with a routing issue at all. That test is from a different machine on a different HE tunnel where the connection to the internet from the machine is very slow so the long delays are from that, that is normal from there though I need to get it fixed but ipv4 pinging google.com would be similar times.
EDIT: make that 2-1c as I just added a bunch more.