Endless print when "remote host did not accept..."

I am getting the message "Remote host did not accept data file (4)" (and 
occasionally control file) periodicaly in a high print volume 
environment. The jobs are being sent raw, and it occurs on multiple 
printers.
When this occurs, the job *is* actualy being printed, but cups does not 
seem to think so. It re-sends the job, over and over until the job is 
canceled. This is causing a lot of wasted paper and jobs getting queued 
up behind the "stuck" job.
This did not occur before switching to cups from lpr. There has been no 
hardware or network changes. The version of cups is from SuSE 8.2 
1.1.18-77.
How can I prevent cups from re-sending the job when this error occurs?

RTFM answers are welcome, if they point to the right section of the FM! 
I havn't found anything elsewhere.

Thanks,
Jeff Stewart