On Sun, 25 Jan 1998, Jukka Marin wrote:

> On Sat, Jan 24, 1998 at 09:50:37PM -0500, Greg A. Woods wrote:
> > I've noticed this problem on various operating systems (most or all
> > with at least their networking based on *BSD).  NetBSD recently seems to
> > be exceptionally prone to this behaviour.
> 
> I saw this problem on AIX some 7 years ago and IBM did nothing, despite
> of our bug reports.  I was happy that NetBSD didn't have this problem
> (I've never seen it on 0.9, 1.0, 1.1, or 1.2) but now I don't think I can
> upgrade all systems to 1.3 because I can _not_ allow this problem on
> public Internet machines.  Many people are using pine and screen on our
> servers...

The problem we've seen w/ AIX & pine is that the process will chew up lots
of CPU. It seems the program is trying to write to the old output, and is
getting stuck. It's like the scheduler pulls it up, then it decides it
can't run, then it pulls it up, then it decides it can't run. We can end
up with such processes taking 70 % of the CPU, when all the other uses of
the CPU add up to about 50%.

Could this be our problem too?

Take care,

Bill