Hello,
I face a tricky problem when OpenBSD 4.4 initiates interfaces
with ipv6 addresses. My setting is a router supposed to route
ipv6 traffic __without__ tunneling:
rl0: exterior interface with static stateful ipv6-adress
fxp0: first interior interface with static, stateful ipv6-address
xl0: second interior interface with static, stateful ipv6-address.
They are all assigned a static inet6 alias in their hostname.if files,
the addresses being assign from my ipv6 net of length 64.
The default inet6 route is to pass over rl0.
As we all know, when /etc/netstart is called from /etc/rc, it will
activate the interfaces in alphabetical order. This will, however,
upset the intended network topology, since fxp0 will as the first
activated interface get the routing entry
2001:470:1234:4567::/64 link#2 UC .... fxp0
This prevents the correct funcionality, since
2001:470:1234:4567::/64 link#3 ..... rl0
would have ensure propered routing outwards over the default route.
The thing is, if I remove "hostname.fxp0" alltogether, then inet6
pinging and routing do work as intended.
Of course, there is the possibility to manipulate the startup:
# /etc/rc.conf.local
netstart_flags="rl0 fxp0 xl0"
# /etc/rc
. /etc/netstart $netstart_flags
but I would have expected there to be an already prepared solution.
There are very many howtos on ipv6 routers using OpenBSD to be
found on the internate, but they concern uses of ipv6-over-ipv4-
tunneled routers, so their information is mostly useless for
pure ipv6 routers on the basis of OpenBSD.
Could somebody provide me with a pointer as to background for
pure ipv6 routing that resolves the matter discussed here.
Best regards
Mats Erik Andersson
__________________________________________________________
Lena pengar utan
sdkerhet. Jdmfvr vilkor online hos Kelkoo.
http://www.kelkoo.se/c-100390123-lan-utan-sakerhet.html?partnerId=96915014
