[hipl-dev] Re: Strange behaviour of NC testing
- From: Miika Komu <mkomu@xxxxxxxxx>
- To: hipl-dev@xxxxxxxxxxxxx
- Date: Fri, 11 Apr 2014 00:16:58 +0300
Hi,
On 04/10/2014 07:15 PM, Robert Moskowitz wrote:
I am using "nc -6" for initial testing.
Over my local subnet, the HIP tcpdump showed the HIP/ESP traffic using
IPv6.
Then I moved my client to my Verizon MiFi wireless off my cellphone. It
has IPv6 and natted IPv4. The HIP/ESP traffic was all over the natted
IPv4, even though I used the "nc -6" switch. I ASSuME that this is a
case of IPv6 app over an IPv4 HIP connection. But WHY did HIP go to
IPv4? And NATed at that when there is a perfectly available IPv6
address and I tested that out before starting HIP?
tcpdump shows the following on Linux:
* input traffic twice: once with ESP and another time with HITs
* output traffic only once: just with the ESP header
So, for input direction it can show both IPv4 and IPv6 traffic if the
locator is IPv4 (and the identifier is IPv6-based HIT). I don't remember
how this works with LSIs from the top of my head, but this should be
easy to check out.
The mobility mechanism should switch the interface only when the active
interface does not work anymore.
Other related posts:
