On 2007-07-25 10:01 -0500, Jan Schaumann wrote:
 
> I'll come up with a patch for pkill(1), then. :-)

We might consider using the term 'runtime' rather than 'timeout', which
is clearly prone to confusion.  Though, why [1] limit this feature to
just real time, though?

I started looking at this, and it got me thinking that it might be
even more useful to extend pkill to be able to restrict processes on
some other data stored in kinfo_proc2 like:
  - "nice" priority
  - sys time
  - user time
  - CPU usage
  - Memory usage

And even better-- to restrict on min/max values for each of these.

Any thoughts?

  - Oliver

[1] I think I know the answer-- real time is useful to you.  Well, I
    don't mind spending a little time on this, as some added features
    could conceivably be pretty nice at work.