If you want prevent a brute force attack you can use public key authentication and configure the ssh daemon to use the version 2. As you wrote the rule > /usr/sbin/iptables -A INPUT -p tcp -i eth0 -m state --state NEW \ > --dport 22 -m recent --update --seconds 15 -j DROP all packets travers the rule, does not matter from where they come. Hope this helps. Jorge. El mié, 08-03-2006 a las 15:37 +0000, JCA escribió: > I have the following setup: > > Internet ---- ADSL modem/router A ----- Linux firewall F ------ Linux box B > > TCP/IP packets for port 22 (the SSH port) are allowed in by A. In > turn, F just forwards such packets to B without any further > processing. In B I have the following IPTables set of rules: > > /usr/sbin/iptables -A INPUT -p tcp -i eth0 -s 192.168.1.0/24 \ > --dport 22 -j ACCEPT > /usr/sbin/iptables -A INPUT -p tcp -i eth0 -m state --state NEW \ > --dport 22 -m recent --update --seconds 15 -j DROP > /usr/sbin/iptables -A INPUT -p tcp -i eth0 -m state --state NEW \ > --dport 22 -m recent --set -j ACCEPT > > My understanding is that with these rules, whenever an SSH > connection attempt to port 22 from a given IP address fails (because > the wrong credentials were used), no further connection requests from > that IP address will be honored for 15 seconds. The first rule just > specifies that this mechanism is not to be applied to connections from > my internal LAN. > > If I make an SSH connection from some machine I in the Internet to > B, B honours this connection request promptly (modulo network > performance, of course). If, the first connection having been > successful, from I another SSH connection to B is opened immediately > afterwards (that is, within a few seconds), this second connection > request takes a long time to be honored - 45 seconds or more. The only > thing that I can think of interfering here is the set of rules above. > > Can anybody throw some light on this? In particular, if the rules > are causing this behavior, can they be modified so that the behavior > disappears, while preserving the 15 second timeout for failed > connection attempts? > > B is running a 2.4.27 Linux kernel, and iptables 1.2.10. > -- Jorge Isaac Davila Lopez Nicaragua Open Source +505 808 2478 davila@xxxxxxxxxxxxxxxxxxxxxxx