On 04/04/11 11:18, Rainer Traut wrote:
> to prevent scripted dictionary attacks to sshd
> I applied those iptables rules:
>
> -A INPUT -p tcp -m state --state NEW -m tcp --dport 22 -m recent
> --update --seconds 60 --hitcount 4 --name SSH --rsource -j DROP
> -A INPUT -p tcp -m state --state NEW -m tcp --dport 22 -m recent --set
> --name SSH --rsource
>
> And this is part of logwatch:
>
> sshd:
> Authentication Failures:
> unknown (www.telkom.co.ke): 137 Time(s)
> unknown (mkongwe.jambo.co.ke): 130 Time(s)
> unknown (212.49.70.24): 107 Time(s)
> root (195.191.250.101): 8 Time(s)
>
> How is it possible for an attacker to try to logon more then 4 times?
> Can the attacker do this with only one TCP/IP connection without
> establishing a new one? Or have the scripts been adapted to this?
i see similar results on some of my servers, eg:
% grep 'a\.bad\.ip\.address' authpriv|grep 'authentication failure'|awk '{print $3}'|less
15:47:44
15:49:34
15:49:46
15:51:32
15:53:17
15:53:30
15:55:14
15:56:59
15:58:44
16:00:34
16:02:19
16:02:31
16:04:17
[...]
so i can see that yes, at least some automated scripts have been adapted
to back off in an attempt not to trip my iptables rules. you can do a
similar grep to see the times of your attempts, and that will tell you if
they're running a softly-softly script, or if instead they have found a
way to test many passwords without tripping the iptables rule.
On Mon, 4 Apr 2011, David Sommerseth wrote:
> This is just a hunch, but --seconds 60 indicates that it will only look
> back one minute to check if it could find a hit. So if the attacker tries
> to connect again after 2 minutes or even 61 seconds, it won't trigger this
> rule. Try increasing this value to 3600 (1 hour). Maybe you want even longer.
i occasionally trip my iptables rule myself, for example if i scp a couple
of files off a server and then go back for a third; i feel it would be a
shame to lock myself out for an hour, by doing that.
the way i see it is that, even in the limiting case where an attacker can
try two passwords every minute, she will be limited to just under 3,000
attempts a day, and that's not very many when you're trying to brute-force
decent passwords. given that most of those are attempts to guess root's
password, and i have "PermitRootLogin no" in sshd_config, the tiny
additional load caused by an attempt every 30 seconds is something i can
live with in exchange for not locking myself out for too long.
how long you set your lockout for is a call you must make for your
server(s); i just wanted you to have more points of view about what people
are doing out there in the wild.
--
Tom Yates - http://www.teaparty.net
_______________________________________________
CentOS mailing list
CentOS@xxxxxxxxxx
http://lists.centos.org/mailman/listinfo/centos
