Jesse Pollard <[EMAIL PROTECTED]> wrote:
+> From: Joe <[EMAIL PROTECTED]>
+> > >
+> > > Furthermore, with this little proggie anyone can "freeze" my
+> > > system;
+> > > i.e. it doesn't halt, but all my resources are eaten up (I
+> > > can't even
+> > > run kill being root) -- can I use another way besides PAM,
+> > > running top
+> > > from inittab or the Sysctl hack to solve this gently (maybe
+> > > using some
+> > > root/kernel memory or perhaps a way to `talk' to init
+> > > directly??)
+> > >
+> > I'd write a small program that would check on system procvess to
+> > see how much memory they are using (read the /proc/PID/status
+> > file) and if a program was using more memory then I wanted it to
+> > kill the process.
+>
+> This is also a call for per/user and per/process limits. You can also
+> hang the system by using too many processes/open file ids ... which in
+> turn calls for more per user information (unabashed plug for a user
+> database instead of a password/shadow file combo :)
I'd like to see this as configrable parameters for the kernel, not as a
real hard limit. Several years ago some people agreed that 640 kb RAM would
be more than sufficient - more than anybody would ever need to have. We all
know how crippled the system was, that arose from this dumb assumption, don't we?
The systems keep getting faster and a limit which is today high up in the
sky is tomorrow a bottleneck. Keep the hard limits away!
Best regards,
Herbert
-------------------------------------------------------------------------------
In theory there's no difference between theory and practice, but in practice...
-
Linux SMP list: FIRST see FAQ at http://www.irisa.fr/prive/mentre/smp-faq/
To Unsubscribe: send "unsubscribe linux-smp" to [EMAIL PROTECTED]
