Gordon Messmer <[email protected]> writes:

> You're hypothesizing the infection of your Unix server with malware,
> and spam is the part you're worried about?  

Not the only part.  But is it *a* concern?  Absolutely.  A significant
one, I would say.

In the first place, it does happen.  I've seen a compromised Linux
system used for sending spam, and if you make even a passing attempt
to follow the goings and comings in the security community you will
quickly realize that this is neither far-fetched nor particularly
unusual.  (It doesn't make mainstream-media headlines, but neither do
Unix rootkits.)  Granted, with the incident I happened to observe the
system in question was not up-to-date on its security updates, so in
all likelihood good practices would have prevented that infection.  I
believe that's typical.  Nonetheless, my point stands: it happens.
Unethical persons *do* attempt to compromise Linux systems and use
them in said manner.

Further, it's a much more damaging outcome than anything else an
attacker (especially an automated attacker) is likely to do (assuming
you have backups of your mail, which is pretty much necessary anyway
in case of hard drive failure, which is annoyingly common).  

A root-kitted system can be fixed in one shift (once you are aware of
the problem) by wiping the disk and doing a clean install.  (If you
have a spare server just about ready to go, you can get back up even
faster, and still have the infected filesystem available for forensic
analysis.)  

A tarnished reputation is not so easy to repair.  Once people (and
automated systems) start putting you in filters and blacklists, it can
can take much more than one shift, and much more effort than a clean
install, to get things cleared up and back to normal.

In security there is a principle called "defense in depth" (or, more
cynically, "paranoia", but whatever you call it, it's an important
part of the mindset of a good network administrator).  You try to
protect yourself from being broken into in the first place, but there
are no absolute guarantees.  A good secure design assumes that there
WILL be security failures from time to time and establishes additional
measures and barriers to mitigate the damage a successful attack will
do, even once the primary security measures are compromised (and also
to detect the incursion so that a response can be mounted as soon as
possible and the damage repaired).

So yes, I do think it would be useful to be able to restrict what
outgoing source ports the mail server uses to send mail, and tell the
firewall to drop or reject outgoing mail from all other ports (or,
even better, silently redirect them somewhere private for analysis,
which has the dual benefit of potentially allowing the attacker or the
malware to think it is operating successfully while also providing an
extra way for the sysadmin to become aware of the problem).

-- 
Nathan Eady
Galion Public Library

------------------------------------------------------------------------------

_______________________________________________
courier-users mailing list
[email protected]
Unsubscribe: https://lists.sourceforge.net/lists/listinfo/courier-users

Reply via email to