Re: Begs a question: AV in Linux

"Aaron" <[email protected]>
Newsgroups gmane.comp.security.linux
Message-ID <[email protected]>
I could not have said it better.  Each server in a farm 
has a purpose.  AV may be applied as needed.

If there is no chance of spreading a virus with that 
server, then running AV only wastes CPU, memory and 
creates higher IO Wait times.  In addition to that, it 
adds one more potential vulnerability, since some virii 
target known functions in AV and not the OS itself.

On the other hand, as mentioned, if the server is relaying 
files that could potentially carry a virii payload, then 
whether for legal or ethical reasons, it is best to stop 
the payload before it gets to hosts that are potentially 
vulnerable.  Mail servers, file servers, chat servers 
(that allow txfer via the server) should run AV.  ClamAV 
is even adding malware detection in the .90 release as far 
as I know.

Unix web servers by themselves have no need for AV 
_unless_ they allow for users to upload/share files.  One 
such example would be the attachment modules for phpbb or 
various web portal php sites.  Web servers that serve up 
static content have no need for AV, as the webmaster 
should ensure that nothing on the site is malicious.

AV should also always be configured to exclude things that 
would cripple a machine if they attempted to scan or lock 
the files.  One such example would be MySQL/Oracle 
database files.  Another example would be swap "files" if 
swap were extended using non partitions.  This isnt a real 
problem if the AV user is non-root, however being 
non-root, one must trust all of the applications handling 
the payload to make use of the AV software properly.

In terms of protecting the server itself, _if configured 
properly_ an alternate direction to go would be to make 
use of SELinux when using linux as the server.  This is no 
small task in a highly customized site, however does not 
consume the resources of a typical AV solution and 
prevents far more serious issues than virii.

http://www.nsa.gov/selinux/   Included in Redhat and 
Gentoo now, along with a few others.

--Aaron












On Fri, 27 Jan 2006 10:18:03 -0600
  Eric Rostetter <[email protected]> wrote:
> Quoting Moderator <[email protected]>:
> 
>> The following message was submitted to the list by 
>>Alexander Klimov.
> [...]
>> Since there are quite a few replies let me elaborate. 
>>There are two
>> types of viruses: those that exploit software 
>>vulnerabilities and
>> those that exploit wetware (that is a PEBKAC).
> 
> And there are _many_ kinds of linux systems and users.
> 
>> the virus is released. Unlike some other OSes, with any 
>>good Linux
>> distribution it is quite easy to live most of the time 
>>without known
>> vulnerabilities in your system.
> 
> If you run wine, zen, mach, vmware, or anything that 
>runs or can run
> windows (or another vulnerable OS), than you should run 
>AV in at least
> the virtual machine, and preferably in both linux and 
>virtual machine.
> 
> If you run openoffice, you are open to macro viruses and 
>all the same
> things that hit MS Office apps, and you should run an AV 
>if you don't
> want to be a hit by them, or spread them to others.
> 
>> Now if you have a system with no
>> vulnerabilities exploitable by known viruses none of 
>>them can
>> compromise your system -- you cannot get better results 
>>from an AV
>> (AFAIK `unknown virus detection' is more marketing than 
>>reality).
> 
> True.  But you can help spread them.  Of course there is 
>the obvious
> examples of linux machines which are file servers and 
>mail servers and
> the like.  Why would you want these spreading viruses? 
> But even regular
> office user linux machines can spread around viruses via 
>file transfers
> (forwarding e-mail, swaping floppies or usb devices, 
>burning cd-roms, etc).
> Maybe not a big deal if you only deal with other linux 
>machines, but if
> you interact with people using other OS's do you really 
>want to be the
> one who passed a virus on to them?
> 
>> root to solve it: wget ...'. I am not sure I understand 
>>how sharing
>> files with Windows can be dangerous but probably it is 
>>in this
>> category as well
> 
> It is dangerous for other windows users you give the 
>file to, or dangerous
> to you if you run windows in a VM environment in linux, 
>or run OpenOffice
> or other windows-software emulation software.
> 
>> BTW do not get me wrong: if I say that AV is useless 
>>(or, worse, it
>> can have its own vulnerabilities) it does not mean that 
>>you should not
>> use a firewall in both directions or check integrity of 
>>system files.
> 
> AV software _may_ be useless depending on your 
>environment.  I run it
> on my linux mail server, and it is not worthless to me 
>or my users, since
> half my users run Windows and Mac machines.  They thank 
>me for not exposing
> them to the viruses via their e-mail.  You could make 
>the same type of
> arguments for file servers, etc.
> 
> Yes, you _may_ not need a AV product on your linux 
>machine.  Then again,
> you _may_ need one.  It depends on how you use the 
>machine, what you run
> on the machine, and how you and that machine interact 
>with others.
> 
> The real-world example is how it is illegal most places 
>to knowingly infect
> other people with a human virus that you know you carry. 
> It does not matter
> if you are immune to it or not, the law reflects the 
>fact that others are not
> and that you should not knowingly spread it to them as 
>you know it can cause
> them harm.
> 
> Use a similar principle in computers and networks.  If 
>you know your
> computer has or is likely to spread viruses to others 
>and could cause harm
> to them, then the _responsible_ thing to do is to run AV 
>software on your
> machine to try to prevent that.   If you know your 
>computer is
> _highly unlikely_ to spread viruses to others, and 
>should not pose any
> virus risk to others, then there is no need to run AV 
>software if you
> don't want to (and may be very good reasons not to, in 
>fact).
> 
>> --
>> Regards,
>> ASK
> 
> -- 
> Eric Rostetter
> The Department of Physics
> The University of Texas at Austin
> 
> Go Longhorns!
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.