We have 30 some branch servers rsync'ing their user data folders to a central site and jobs run with about 80-90% success rate (the others usually generate the send_buffer error) except with one location we always get the same error and we get it immediately on the server side (daemon side) and it unloads on the client side.

read_timeout: Error peer socke closed fd = 179
rsync error: error in socket IO (code 10) at io.c(550)
SendBuffer: Error WSASend fd=179 nRet=-1 err=10053
rsync error: error in socket IO (code 10) at io.c(1971)

Here is the NCF file for the client:
rsync.nlm -arzq -z --volume=SYS: --log-file=sys:/system/rsync1.log --timeout=14400 --bwlimit=4 --include="/mor_po" --include="/mor_d" --include="/users" --include="/apps" --include="/data" --exclude="/*" --exclude="/*.*" / --delete
(We don't ever see anything in the log file - the bwlimit is low because this runs over an Internet VPN link)

The message is always exactly the same. Each server is running Netware SP5, using Rsync 2.63, and libc.nlm 9.00.03. Any ideas would be greatly appreciated.