Despite the name, I gather this is not an excessive page faulting problem. If
it is then perhaps increasing the working set size and/or physical memory is in
order.
I also gather that simply training the users to exit the "pager/paging"
application is not sufficent.
The problem is if you do figure out how to kill a process running the
application after a fixed amount of time, you might be doing the wrong thing,
inasmuchas the user would have to restart.
The application must be fooling HITMAN somehow. The best way to determine if
there's a user still making use of the application is to time input.
Whereas the following loop should not defeat HITMAN;
$ loop:
$ show time
$ wait 0:1
$ goto loop
the following could;
$ loop:
$ set terminal /inquire
$ wait 0:1
$ goto loop
The difference is the set terminal /inquire command causes the terminal (or
emulator) to send a sequence of bytes which look a whole lot like user
keystrokes. If HITMAN is fooled by the first loop then the application might be
ticking a clock on the display.
If the pager program runs under (or can be made to run under) a specific
username, or in a specific queue, one can use existing CPU limits in the
SYSUAF or in the queue definition.
I'd strongly recommend another approach, as blindly deleting a process
has the potential to cause all manner of interesting corruption(s).
If you have access to the source(s) of the application, it would likely
be *much* safer (and easier) to simply recode the application to include
an "exit" sequence when a timer has expired.
