gabriel rosenkoetter on Thu, 5 Dec 2002 12:20:08 -0500 |
On Thu, Dec 05, 2002 at 11:37:12AM -0500, Naresh wrote: > Well, at my univ (Pennstate), AIX is set up so if a user has a process > running longer then 45 mins the process is automatically killed. I am sure > its possible... Hrm. That's mainframe-y behavior. (System 390's Linux-like area will kill -9 processes if the take more CPU time than is allocated for thair LPAR, for instance.) Sure it wasn't running within some environment like MPI, PVM, or Sun's GridEngine? In any case, AIX behaving in mainframe-y ways wouldn't be totally shocking... ulimit -a in the shells I just tried (sh, csh, ksh, bash2, tcsh, zsh) lists cpu time as an option, and it's easy to force that on some but not all of your users in /etc/profile (and /etc/csh.cshrc, if you've got any C shell-derived shells in /etc/shells) with something like this (Bourne syntax, presuming uid 1000 is the user to be limited to 1024 seconds of CPU time): if [ x"id -u" = x"1000" ]; then ulimit -t 1024 fi -- gabriel rosenkoetter gr@eclipsed.net Attachment:
pgpwudiyS0zik.pgp
|
|