Outbound ports to firewall?

[Jason Dixon]

On Sep 24, 2004, at 9:29 AM, Parker Morse wrote:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Like most people, I've put some effort into filtering incoming email 
> and firewalling my network to prevent nasties from getting in. But 
> recent discussion of preventing the spread of Windows worms, viruses, 
> etc. etc. has led me to a question I don't have an answer for.
>
> Let's assume, for a thought experiment, that one of the Windows boxen 
> inside my gateway firewall is infected with *something*, who knows 
> what. To protect the rest of the 'net from this little bundle of 
> pestilence in the time before I track it down and choke it to death, I 
> should probably have some firewall rules to keep the bulk of the 
> nastiness from leaving my network. Outbound rules.
>
> What ports should I consider closing up to keep hypothetical infected 
> inside my network from phoning home and/or spreading the infection?

You don't.  You block all by default, and only allow approved outbound 
traffic (via proxy or directly).  Otherwise, you're constantly 
attempting to play catch-up with mutating (and new) undesired services. 
  Here is an example list of approved outbound traffic from my (OpenBSD 
PF) ruleset:

tcp_out_services="{ whois, ftp, http, https, ssh, pop3, pop3s, imap, 
imaps, smtp
, bootps, 465, 1723, 1863, 3128, 5190, 6667, 55500 }"
# 465 = SMTP/SSL
# 1723 = PPTP
# 1863 = MSN Messenger
# 3128 = Squid
# 5190 = AIM
# 6667 = IRC
# 55500 = PokiPoker
udp_out_services="{ domain, bootps, ntp }"

HTH.

--
Jason Dixon, RHCE
DixonGroup Consulting
http://www.dixongroup.net