[support] DAD Delay too high?
Arnaud Ebalard
arno at natisbad.org
Thu Sep 3 20:03:19 JST 2009
Hi,
support at adhari.de writes:
> Hello everybody,
>
> I am performing MIPv6 handover measurements using UMIP-4.0
> on a debian system with a 2.6.29 kerne, and i am recording
> some unusually high values of handover latency (upto 8
> – 10 seconds) and am wondering if any one has an
> idea.
>
>
> I am performing a manual handover from HA to foreign
> network in a WLAN network (using Linksys APs), and I am
> capturing traffic using wireshark and I notice that after
> Router Discovery, the MN will understandably perform DAD
> test on the new CoA by sending Neighbor Solicitation
> messages (NS) I notice that the MN sends out two NS, one
> having the Link Local Address and the other the tentative
> CoA. They are send at times 5.255676 and 5.423619 sec
> respectively.
>
> Now as per my understanding, the DAD test will take 1 sec
> and if it passes, the MN will assign the CoA and send a BU
> to the HA immediately. However I observe that the BU is sent
> almost 3 seconds later (i.e., at time 8.182842 sec. This can
> be seen in the wireshark log below:
Testing it on my Debian (2.6.30.5 kernel), the 3 NS (DAD for link-local
and 2 other unicast global addresses) are sent and a protected BU is
sent exactly 1 second after the emission of the last NS.
> Can any one explain as to why the MN spends so much time
> after the DAD test before sending out a BU message? Is this
> normal as I expect the BU to be sent almost 1 sec after the
> NS is sent out?
No, it is not normal. Can you post your configuration file?
> Where can I check the value of the DADRetransTimer
> parameter? Is there a possibility to manually set the
> DADRetransTimer parameter?
For DAD, there should be no retransmit IIRC. Is that question linked to
your problem?
Cheers,
a+
More information about the Support
mailing list