(usagi-users 03001) MIPv6 + Tunneling problems .....

["Daniel Tyrode" <dtyrode4@xxxxxxxxxxx>]

Hi,

I've been trying some configuration scenarios with MIPv6 (Using SuSE 9.0+
USAGI's Stable Release 5 2004-01-04) and I have encountered some issues. I
wonder if someone can provide me guidance:

1. I am trying to establish an IPv6-tunnel between the CN and the MN. I will
refer to the "tunnel" addresses as "internal" addresses and the "interface"
addresses as "outer" addresses. While at the home network, almost everything
works fine. For instance, when I try to ping the CN's  "inner" address from
the MN, for some reason the default "outgoing/source" address becomes that
of the MN's MIPv6 ("outer"). I can force the ping6 with the "-I" to use the
MN's "inner" address as source address instead -which in normal
circumstances should be the correct one-, and in this case it works as
expected. Is there any good reasong for this behaviour ? ... Is there a way
to force MIPv6 not to override other addresses as source ("outer" address
replacing "inner" source address )?  This problem occurs with both Route
Optimization and Bidirectional Tunneling.

This problem is not bidirectional: when the CN pings the MN's "inner"
address, everything works as expected.

2. The other issue is when I start moving to the visiting/foreign Network.
With Bidirectional Tunneling the data flow collected from tcpdump when I
ping the MN's "inner" address from the CN, shows that the CN encapsulates
"inner" address within the "outer" one (tunnel) and sends the packet to the
HA. The HA intercepts it and then encapsulates(tunnel) the received packet
and forwards it to the new MN's "outer" address (Assigned by the Foreign
Node Prefix+Token). The MN receives the packet and performs the two
de-encapsulations perfectly ( strips-off the HA->MN tunnel, then the CN->MN
tunnel) as I can examine with tcpdump the "inner/tunnel" interface on the MN
which shows the ICMP "echo request" and "echo reply" pairs:


            CN                                     ->
HA                            ->                                        MN
Encapsulates "inner" in "outer"                         HA, intercepts and
tunnels to MN                            MN de-encapsulates both tunnels
correctly
and sends to MN's home address


Up to this point, everything seems to be working fine and as expected.
However, the MN does NOT reply back even though the correct routes are
available in the routing table. That is, there is no further communication
between the MN and the HA in this case ( no packets from MN -> HA ). Could
this be a problem with the software implementation or is there something I
should be aware of ?

Many thanks in advance for your valuable advice,

Best regards,

Daniel Tyrode