Cluster update

[david nicol]

does the kclug cluster have it's own mailing list yet?  Until it does
I guess this list will have to do -- I propose including the word
"cluster"
in the subject of all cluster-related messages.

This kind of worry is a reason to *not* allow cluster users to be root;
to restrict membership to known quantities with acceptable moral and
ethical sensibilities, etc.

If we have a clear TOS and we are vigilant about removing local root
exploits and so on, if anyone ever does something like that we can
disuser them and claim we are a "common carrier."

Additionally, off-the-shelf MOSIX does not allow you to open ports
from remote machines.  For that you would need rsh or srsh or telnet
or something.  You just get access to timeslices and memory of remote
machines; all IO goes through your own box.

On Mon, 2002-12-09 at 09:23, KRFinch at dstsystems.com wrote:
> Just a thought:
> 
> It occurred to me that some industrious hacker could decide that our little
> cluster would be a great place to launch a DOS or other attack from, all in
> the name of "testing his software".  A properly coded multi-threaded
> beastie running on a cluster could be a powerful attack tool indeed, not to
> mention that it could probably get us all in a load of trouble in today's
> litigious environment.
> 
> In that light , what I think would be best would be to have it queue up the
> test, run it, and then just email the results to the person that initiated
> the test, leaving the test to run in a "closed" environment.
> 
> - Kevin
> 
> 
> 
> 
> 
> 
> 
-- 
David Nicol, independent consultant and contractor            312 587 2868
God did not create the world in seven days; he screwed around for six days
                           and then pulled an all-nighter. -- `fortune -s`