Someone is using too much bandwidth???
tom
tom at t0mb.net
Tue Nov 21 19:28:27 CET 2006
Taylor, Grant wrote:
> lubasi wrote:
>> How can i interprate the #tail -f /var/logs/messages to determin
>> which machine is doing kazaa or any other P2P???consuming the
>> bandwidth.
>
> By default /var/log/messages will not record any thing about traffic
> that is
> passing through the system. You can add IPTables rules that will cause
> matched packets to be logged via Syslog which you can then see in
> /var/log/messages.
>
> However to get a better idea of what traffic is running on your network,
> consider TCPDump or a GUI front end like Etherial. This will give you a
> real time report of what traffic is flowing in to / out of / through your
> system (presuming you sniff the correct interface). You can tell from
> this,
> which computer is consuming more bandwidth than it should based on the
> frequency of the source / destination IP showing up in TCPDump's output.
>
> You could add rules to IPTables that match specific IPs in question and
> watch the hit counters to see which system(s) are incrementing their
> counters at an exceptional rate. One (or more) system(s) should jump
> out at
> you as being the culprit(s).
>
>> And how do i block these popular P2P???
>
> First you need to find out more about the type of P2P traffic that you
> are
> experiencing so that you can more accurately filter it out / rate
> limit it.
> I will say that you may have better luck with rate limiting. If you
> completely block a users access to something they will find a different
> method to get to what they want to get to. If your users switch to
> something else you then have to learn about that too. Where as if you
> let
> your users use one system but control the amount of bandwidth consumed
> and /
> or the priority you may not play the above game nearly as often.
>
> My family has a saying, "Give 20% to get 80% of what you want.".
>
>
>
> Grant. . . .
>
>
iftop will suit your needs for monitoring like that.
http://freshmeat.net/*iftop*
More information about the netfilter
mailing list