[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Starting iptables



Advanced Policy Firewall is good for the home user,
http://www.rfxnetworks.com/apf.php

Its a very simple one file configuration with some optional perks.
Define the untrusted IF, its egreess, ingress, tweaks (default values
are already sane) and such stick it in init.d - done. I believe it now
has a semi-guided installer.

Very well documented and well suited for home use. Nowhere near as
capable as shorewall, but I think that's the argument to be made for it
in that setting. The web hosting industry relies on it rather heavily
for shared web servers. It gets clunky after a few hundred rules, but
when would an average user ever have a few hundred rules? :)

HTH
-Tim

On Sun, 2006-10-22 at 09:43 -0700, John L Fjellstad wrote:
> dtutty@porchlight.ca writes:
> 
> > On Thu, Oct 19, 2006 at 05:22:24PM -0700, John L Fjellstad wrote:
> >> dtutty@porchlight.ca writes:
> >> 
> >> > If you look at the number of lines of rules you make, and compare it
> >> > to the number of lines (pages!) of iptables rules it makes, you see
> >> > that shorewall is easier.  Also the syntax is easier.  Changes are
> >> > far easier.  Besides, the shorewall book is the best book I've found
> >> > for understanding iptables.  
> >> 
> >> shorewall creates pages of iptables rules and that is considered a
> >> good thing? What happened to KISS?
> >> 
> > Yes it is a good thing.  The purpose of a firewall is to block anything
> > that you don't explicitly want through.  If you don't want anything
> > don't put any 'allow' stuff.  Then the default rules of deny all is in
> > effect.  The issue is that there are different protocols for the same
> > service (e.g. UDP, UTP, etc).  Each little pinprick you want opened
> > takes a few rules to keep it to a specific pinprick.  If you did it
> > manually with fewer rules you would have a more porus firewall or you
> > wouldn't have the services you want traversing the firewall.  If you
> > used too few rules you would have a screen door.
> 
> Bull.  How does few rules create a screen door as opposed to "pages" of
> rules?  How many services do you have that you need "pages" of rules?
> How does each pinprick you open not create another entry point?  How
> does fewer "pinprick" opened create less security, while more "pinpricks"
> create more security?  How is this keeping it simple?
> 
> > For comparision, go to tldp and get the securing-linux manual (redhat
> > edition).  Its in pdf format.  That author took the same approach you
> > suggest and does everything except the base install by hand.  Read the
> > section on firewall.  See the pages of rules he has in his firewall
> > script.  He explains it all too.
> 
> I couldn't find the article you were talking about, but I did find a
> Securing-Optimizing-Linux-The-Ultimate-Solution-v2.0.  And the number of
> rules are insane.  Why would you have an explicit DROP rule when you
> have a DROP policy?  Where is the logging? (Yes, he has a comment about
> how he logs selected denied packages, but no logging actually occur) Of
> course, if you want to be the "ultimate-solution", why would you want to
> keep it simple?
> 
> Sigh...
> 
> > The only ways I know of to KISS a firewall are ipmasq and shorewall.
> > Shorewall makes a better firewall so it makes more rules.
> 
> KISS.  Keep It Simple.  As in as few rules as possible.
> What do you need?
> 
> Take a home user. What does he need?
> 
> Well, he needs to open the loopback. Rule 1.
> He wants any packages that he started to be let through (RELATED,
> ESTABLISHED). Rule 2.
> Maybe he wants to use p2p. That's a range. If you use bittorrent, you
> might have to open an additional port for the control package. That's 4 rules.
> End it with a LOG rule with rate limit.
> 
> That's _five rules_.  Use DROP as a policy.  How is this _less_ secure than
> having "pages" of rules?  How is having _fewer_ rules create more
> insecurity? 
> 
> -- 
> John L. Fjellstad
> web: http://www.fjellstad.org/          Quis custodiet ipsos custodes
> 
> 



Reply to: