dcerouter kernel: [76470.546007] Neighbour table overflow. messages in log

[jmcrtp]

I have been having a problem where every few days I turn on my TV and the server is non-responsive.  Ie I can't connect my fiire chief, I get no video, etc.  I now have a laptop where I can ssh into the server.  I logged in and cpu is very low.  I checked the log and I see a LOT of the following messages:

Jun 16 22:50:24 dcerouter kernel: [76470.546007] Neighbour table overflow.
Jun 16 22:50:24 dcerouter kernel: [76470.546144] Neighbour table overflow.

What do these mean?  Is this what is causing my problem or a result of whatever it is?

[jmcrtp]

Anyone have any idea on this message?

[gumis]

I had that problem on my LinuxMCE 7.10 as well.

This message indicates that your ARP cache table if full basically. You could either change the size of it (increase it) or find out what is causing this problem.
To change default size of ARP table you need to do the following:

Code: [Select]

echo 1024 > /proc/sys/net/ipv4/neigh/default/gc_thresh1
echo 2048 > /proc/sys/net/ipv4/neigh/default/gc_thresh2
echo 4096 > /proc/sys/net/ipv4/neigh/default/gc_thresh3

On my system it was caused by nbtscan used by StorageDevices_SambaRadar.sh script.
Basically nbtscan scans the whole local network for active IP addresses on the network, it tries to communicate with each IP and ARP is catching it and storing in cache table. Because I've changed my local network from 192.168.80.0/24 to 10.10.0.0/16 it was scanning the whole /16 network, thats 65536 IP addresses...

What I did (which is not a great solution!), I manually edited /usr/pluto/StorageDevices_SambaRadar.sh script and changed the following line:

Code: [Select]

IntNetwork="${IntIP}/$(netmaskLen "${IntNetmask}")"
to

Code: [Select]

IntNetwork="${IntIP}/24"

which will force nbtscan to scan only /24 network