I've replaced netmask 64 with netmask 56 and it works perfectly now. Very odd that a wrong netmask causes this kind of malfunctioning of the entire networking, but glad it has been resolved!
Thanks a bunch. On Fri, Aug 5, 2011 at 4:12 PM, Guus Sliepen <g...@debian.org> wrote: > On Fri, Aug 05, 2011 at 03:44:41PM +0200, Wouter van Eekelen wrote: > > > Here is the output of all networking related commands I could think of. > > Please do not attach this information to the bug report since it contains > > private information. > > Ok. I think the problem is that ifup fails to bring up the interface > completely, a subsequent ifdown then doesn't do anything, and that means it > doesn't properly bring down the bond interface: > > > root@web1:~# ifdown bond0 > > ifdown: interface bond0 not configured > > root@web1:~# ifconfig bond0 down > > root@web1:~# ifup bond0 > > SIOCADDRT: No route to host > > Failed to bring up bond0. > > Notice the error message: "No route to host". I think the problem is in > this > part of your interfaces file (first 48 bits of the addresses are replaced > by > Xs): > > > # The primary network interface (IPv6) > > iface bond0 inet6 static > > address XXXX:XXXX:XXXX:1::1 > > netmask 64 > > gateway XXXX:XXXX:XXXX::1 > > Notice however that the gateway address has ::1 after the last X, while the > interface address has :1::1 after the last X. That means the gateway > address > is not contained in the XXXX:XXXX:XXXX:1::/64 subnet, and therefore there > is no > route to the gateway address. To work around it, you can replace the > gateway > option with: > > up /sbin/ip route add default via XXXX:XXXX:XXXX::1 dev bond0 onlink > > Why it works at boot time, I have no idea... > > -- > Met vriendelijke groet / with kind regards, > Guus Sliepen <g...@debian.org> >
