[NTLUG:Discuss] ip_conntrack questions

[JD]

> I have been run the same server (RH 6.0) for more than 6 years,
> Recently I get this message very often -- about every 2-3 weeks.
>
> ip_conntrack: table full, dropping packet.
>
> I even increased the MAX value.
> still happens every 2-3 weeks,
> can any one think about what's going on it?
> any one have quick way to flush the table without reboot?
>
> If I don't load module ip_conntrack, this problem should be go away?
>   
Depending on what else you are doing - it may not be advisable to also 
run ip_conntrack.  I remember that stuff like tarpitting connections 
advises not to run conntrack.

You can probably restart the service after saving the current state of 
the tables/chains.  You don't have to restart the whole box.  Look for 
the /etc/init.d script that starts the process.  Its been awhile since I 
looked at RH6.