> Theoretically two nodes in "two-hop" range might get the same linklocal IP,
> because they don't share a link, but the node between them would be in
> trouble.
Yep. A possible workaround would be to assign random IPv6 interface-ids
to the interfaces. (There's 62 bits of randomness in a locally-assigned
interface-id, which should be unique enough unless /dev/random is very
badly broken.)
The downside is that you'll end up with interface-ids that don't stick
across a reboot, but that's the price you pay for dodgy hardware.
Have you actually run into that issue (truly curious)? In case you
have, running this on the guilty interfaces just after they've been
ifupped should fix it (untested):
#!/bin/sh
ip -6 addr flush dev "$1" to fe80::/10
ip -6 addr add "$(ahcp-generate-address fe80::)" dev "$1"
You'll find ahcp-generate-address in the ahcpd package.
Juliusz
_______________________________________________
Babel-users mailing list
[email protected]
http://lists.alioth.debian.org/mailman/listinfo/babel-users
