restrict access

[Steffen Nurpmeso]

Walter Harms wrote in
 <c5f4926e2308461ab1fa7d7be0b3b5c5 at bfs.de>:
 |I did a little experiment and it worked.
 |
 | if (fnmatch("192.168.1.*",remote_host,FNM_PATHNAME) != 0)
 |       goto out;
 |
 |this will allow only connections from 192.168.1.* to the server
 |that shows the change can be very simple. I did not try with more compli\
 |cated situations. The limits of this approach needs to be evaluated. 

Since the begin of this thread this sounds like a 100% firewall
thing to me.  Why would you like to compile this in?

I mean, i can imagine the NetBSD/FreeBSD black(now block)list
approach in which a server software who "knows" what has happened
acts via a hook instead of let some expensive log parser
reevaluate state which is known in the moment the log happens.

But this?  I am not an administrator and thus firewall guru, but
i for example have in my net-qos.sh:fwcore_start() (heavily
vaporised this is)

   change_chain INPUT
   new_chain i_good i_alien i_sshorvpn i_tcp_new

   add_rule -m conntrack --ctstate RELATED,ESTABLISHED -j ACCEPT

   add_rule -j i_good
   add_rule -j i_alien

   add_rule -p tcp --syn -m conntrack --ctstate NEW -j i_tcp_new

   change_chain i_tcp_new

   fwcore_has_i ssh && add_rule -p tcp --dport ${p_ssh} -j i_sshorvpn

   change_chain i_sshorvpn

So and in here you can allow or deny ssh-specific anyway you want
to, add, remove and change, use "-m recent" and hitcounts etc.,
and all without recompilation.  (Having real address and/or CIDR
tables which could be managed separately would be cool though.)

--steffen
|
|Der Kragenbaer,                The moon bear,
|der holt sich munter           he cheerfully and one by one
|einen nach dem anderen runter  wa.ks himself off
|(By Robert Gernhardt)