[hipl-users] Re: questions on mobility
- From: Mika Kousa <mkousa@xxxxxxxxx>
- To: hipl-users@xxxxxxxxxxxxx
- Date: Mon, 26 Apr 2004 18:54:53 +0300 (EEST)
On Mon, 26 Apr 2004, Simon Schuetz wrote:
-> Some other questions:
-> On my test setup it takes about 2-3 seconds from sending AC to receiving
-> ACR. Where does this delay come from (my RTT is about 100ms)?
This is a little bit weird, because I have not seen that long a delay so
far. Usually the delay during my testings have been only some < 100-300
ms.
-> Is this to prevend DoS attacks?
We have not implemented this kind of a feature, so what causes this is a
little mystery. Our code should produce some debugging information which
might be useful. See the log files for text "calculated RTT", if you
compiled HIP debugging, on that line there should be a RTT value (in
jiffies) which tells the time between the sending of AC and receiving of
corresponding ACR. To get the time in seconds divide that value with the
HZ value found from linux/include/asm/param.h. On x86 this value is
typically 100.
Also see the system log files and the timestamps before the debugging
information to verify if there really are some 2-3 second delays. Look for
texts "hip_inbound" and "Received HIP packet" to see when the packets were
received by the kernel.
Name lookups might cause some delays, so if you are using some packet
capture program as the only source for measuring the delays then disable
the name lookups while capturing. You could try to run tcpdump with option
-n to prevent it from trying name lookups. A simple "tcpdump -n proto 99"
might be enough.
Other related posts:
