[Olsr-dev] Bug? "Unknown type: 0, size 1428, from a.b.c.d", OLSR hangs w/ 90% CPU load

[Jo-Philipp Wich]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi list,

I have the latest OLSRd 0.5.5pre tip revision from HG repository (Sat,
12 Jan 2008 22:35:25 +0100) running in our mesh cloud in Leipzig.

After several hours of normal operation OLSRd stops working.
When I log into my Linksys to track down the problem I see that the
OLSRd process consumes about 90% CPU and doesn't do any routing anymore.

To further investigate this problem I've set up a netcat listener which
redirects any incoming data to a logfile on a computer connected to one
of the node's LAN ports. Afterwards I restarted the OLSRd process on my
node with:

  killall olsrd; olsrd -d 9 | netcat 192.168.1.7 10000


The first few hours everything was normal and no suspicious entries can
be found in the debug dump file. However, suddenly OLSRd starts to only
generate thousands of messages like:

  Unknown type: 0, size 1428, from 104.61.17.11

It's always the same ip address. There is nothing of the other usual
debug output anymore, it just repeats the line mentioned above.

I then rgrep'd the OLSRd source for the corresponding message and had a
match in src/parser.c in the unknown packettype exception code.

I have no real insight in the inner workings of OLSR but for me it looks
like the process is trapped somehow in the package processing loop.

How can I aid to debug this problem?


Greetings from Leipzig,
jow
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFHjP/mdputYINPTPMRAjJzAJ0Vig57gnK9t5lHKDCvthSIVE9ISgCfT/PX
tgjDGL7qsb9QQkinoXGeJA0=
=2wuI
-----END PGP SIGNATURE-----