I have a Linux bridge machine with a 2.6.6 kernel. It's a simple setup: two ethernet cards (both using the eepro100 driver) acting as a bridge to my office. Here's the relevant ifconfig output: br0 Link encap:Ethernet HWaddr 00:02:B3:16:D3:16 inet addr:172.23.22.199 Bcast:172.23.255.255 Mask:255.255.0.0 inet6 addr: feed:4d56::202:b3ff:fe16:d316/64 Scope:Site inet6 addr: fe80::202:b3ff:fe16:d316/64 Scope:Link UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 eth0 Link encap:Ethernet HWaddr 00:D0:B7:74:76:CE inet6 addr: fe80::2d0:b7ff:fe74:76ce/64 Scope:Link UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 eth1 Link encap:Ethernet HWaddr 00:02:B3:16:D3:16 inet6 addr: fe80::202:b3ff:fe16:d316/64 Scope:Link UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 I've manually assigned the site address to br0. I am running radvd 0.7.2 on the machine. Its configuration looks like this: interface br0 { AdvSendAdvert on; # UnicastOnly on; prefix feed:4d56::/64 { AdvOnLink on; AdvAutonomous on; }; }; Occasionally the machine will get into a state where this message fills the logs (thank goodness for the ratelimit): hashi kernel: icmpv6_send: no reply to icmp error It looks like this traffic may be related to the error: 08:14:49.389415 fe80::202:b3ff:fe16:d316 > ff02::1:ffb4:fdbb: icmp6: neighbor sol: who has fe80::20c:76ff:feb4:fdbb 08:14:49.389455 fe80::20c:76ff:feb4:fdbb > fe80::202:b3ff:fe16:d316: icmp6: neighbor adv: tgt is fe80::20c:76ff:feb4:fdbb 08:14:49.389486 fe80::202:b3ff:fe16:d316 > ff02::1:ff16:d316: icmp6: neighbor sol: who has fe80::202:b3ff:fe16:d316 08:14:50.389265 fe80::202:b3ff:fe16:d316 > ff02::1:ffb4:fdbb: icmp6: neighbor sol: who has fe80::20c:76ff:feb4:fdbb 08:14:50.389306 fe80::20c:76ff:feb4:fdbb > fe80::202:b3ff:fe16:d316: icmp6: neighbor adv: tgt is fe80::20c:76ff:feb4:fdbb 08:14:50.389341 fe80::202:b3ff:fe16:d316 > ff02::1:ff16:d316: icmp6: neighbor sol: who has fe80::202:b3ff:fe16:d316 Note in the third and sixth lines the bridge is requesting its own address. fe80::20c:76ff:feb4:fdbb belongs to another one of my machines, also running Linux 2.6.6. There's very little IPv6 traffic on this LAN (mostly my own ssh sessions), though there are quite a few machines with various OS's that have IPv6 enabled. I can provide packet dumps if they are needed. I've temporarily worked around this problem with the following command: # ip6tables -A OUTPUT -p icmpv6 -s fe80::/16 -j DROP This works out okay for me because I always use the site local addresses, but obviously it isn't an ideal solution... -- The universe is ruled by letting things take their course. It cannot be ruled by interfering. -- Chinese proverb