Sir,

I have a CENTOS 5 machine on which I am running squid proxy. The squid runs fine but once in 24 hours it starts giving RTU and does not ping continuously.
On checking the log file it gives the following message " kernel: ip_conntrack: table full, dropping packet".
On restarting the machine everything is restored back to normal but after 24 hours the same message starts appearing and restart is done.

On finding for a solution on the net the reason pointed out was too many sessions are opened I increased the limit by typing the command : sysctl -w net.ipv4.netfilter.ip_conntrack_max=120536 still the problem remains as it is.

Help is required as to why this is happening and how it can be resolved.

Try this:

http://www.maninmanoj.com/2013/02/ip...ng-packet.html

Sir,

I have already tried doubling the ip_conntrack_max default value. But after 24 hrs the problem comes again.

Moreover I am not able to understand why the server experiences RTU when we continuously ping it.

OK, you have to read that whole page I gave you. If you don't read the entire page you won't know what to do. Your use of the term "RTU" may be incorrect. The RTU pings the server to make sure the server is up. The server does not experience "RTU." The server probably reboots each day, and then the /proc filesystem is reset, so your custom setting for ip_conntrack is reset to the default value.