One of our boxes (A) was brought down from another box for the same client (B). A. was running httpd and B. was running sendmail. After examining the logs and determining that B. was the culprit, an examination of the logs on B. showed a particular IP with a large number of simultaneous connections. His vector was to do dictionary/name email blast to the domain, then when he got a valid user, he would try to authenticate via another dictionary attack. This was spawning a large number of sendmail processes that ended up slowing down the box considerably.
Performing a whois on the IP address showed us that the hacker was operating from another country that most likely was not a customer of our client.
whois X.X.X.X
In order to keep the box and network alive, we decided to block the IP address on the machine's firewall. We didn't want to add the IP to be there on reboot, in case it was a DHCP address that might be legitimate in the future, but we can insert the IP address inline, which will block it until the iptables is restarted. If this IP comes through again in the future, we would add it to the /etc/sysconfig/iptables and reload the configuration.
This one-liner will keep any traffic from the 'bad' address from doing anything further on this box:
iptables -I INPUT -s X.X.X.X -j DROP
As soon as you hit enter, that attack is over. If you realize you typed in the wrong IP address, and you want to remove the inline rule you just created, change the -I to a -D like this:
iptables -D INPUT -s 208.122.195.120 -j DROPIf you wanted to block this IP permanently, simply add this line to the /etc/sysconfig/iptables:
-A INPUT -s X.X.X.X -j DROP
Then reload your iptables config.