On Apr 4, 12:53am, Ken Hornstein wrote:
> Subject: Re: easy ways to crash your NetBSD system
 |>> >Two easy ways to crash a NetBSD system - is it only my system or...?

 |>> >1. main(void) { while(1) fork(); }
 ...
 |>> I think a lot of Unixes lose big on this one.

 |>I knew someone would say this.  Does this mean NetBSD shouldn't be any
    better?
 |>;-)  It doesn't feel too good that any user can bring the system down any
    time
 |>they want. :-(

 |You could try lowering the per-user process limits to fix this.  I assume
 |that eventually you run out of VM?

The problem is that even if you limit the # of processes it will kill a machine
with system call overhead unless you make the limit so severe that it is not
reasonable for "power" X users.  In a previous position as "javelin catcher"
(SA) for a Univ.  CS Dept.  (where this happens so frequently on servers that
you have to do something) I wrote a daemon that monitored process counts (among
other things) and when a particular count was reached a message was logged and
the users processes were "cleaned" from the system.  (all processes were
"SIGSTOP'd then summarily SIGKILL'd.) This was reasonably effective under
SunOS4...  I never ported it to Sloaris as it didn't seem to be as noticeable
there, probably due to higher overhead per syscall?  Its possible just the
first part would be enough.


...
 |--Ken
>-- End of excerpt from Ken Hornstein




-- 

Chris Phillips (chris@platform.com)	
	Platform Computing Corp.,	     (416) 512-9587 x310
	5001 Yonge Street, Suite 1401,       FAX   512-8976
	North York, Ontario, Canada M2N 6P6