I have an application which does multiple remote copies in simultaneity
from a large proprietary Unix box to a NetBSD 1.4.1 (i386) box, using a
script with lines like:
  for File in $Files ; do
    gzip < $File | rsh NetBSDHost "cat >/data/$File.gz &
    until  [ `jobs 2>/dev/null | wc -l` -lt 8 ] ; do
      :
    done 
  done
  wait
  exit 0

This works OK for a short time, then the receiving machine comes out with
messages like:
   rshd[22090]: connect second port 859: Connection refused

I get a similar result if I use something like:
   echo " .. put $File.Z\nquit" | ftp NetBSDHost &

My quick and dirty solution is an automatic 'kill -HUP' to the inetd process
every second.  I have also tried appending '.999' to 'nowait' in the 'shell'
line of '/etc/inetd.conf'. But there must be a better (more reliable!) way.

Any suggestions ?

G.