In article <chvg80lkgv5ru1jlq0chknsght6l1oumbp@4ax.com>,
Jim <ttiot@miba51.com> wrote:

                                               It's my understanding
that if a WU isn't returned in a certain time frame, it's
redistributed.

Although stopping is probably the most responsible thing in your case,
the above statement is likely to mislead other people.

There is no fixed length of time and excess numbers of copies of a WU
are always transmitted.

Whilst the details have changed, and might have changed again, a newly
split work unit is immediately available for transmission for more times
than the number of results needed to confirm a valid result.  After that,
it becomes available for retransmission if it is WU in the WU pool that
has not been transmitted for the longest.  It remains in the pool until
enough results have been returned to confirm it, and the free space in
the pool has become depleted because of newer work units, and the next
scheduled purge is run.

The time from the first burst of transmissions to the next one or between
subsequent ones depends on the size of the pool, which is fixed, and the
rates at which new units are being split and old ones are being successfully
confirmed.  The interval is of the order of a day or two, when operating
normally.

Sending a number approximately equal to the normal state ratio of 
available processing capacity to splitting rate means the maximum
possible amount of the pool is available for when things go wrong.