Your master server is probably taking serious strain and is possibly running out of resources when bpbkar kicks in to backup itself. Have you tried to monitor cpu and memory on the master while the backup is running?
This TechNote explains why false status code 96 can be seen (problem was still present in 5.1):
seer.support.veritas.com/docs/237534.htm
Also check KeepAliveTime on the master:
Extract from NetBackup Release Notes – Additional Operational Notes (TechNote 278130):
While at first it may appear that there is a problem with the NetBackup master server,
this delay is actually a result of a certain TCP/IP configuration parameter that is set to
7,200,000 (two hours, in milliseconds) by default. VERITAS recommends that you set
this parameter to 15 minutes (900,000 milliseconds) or the next lowest, configurable
value that is greater than 15 minutes. (This parameter is called KeepAliveTime on
Microsoft Windows and tcp_keepalive_interval on Solaris systems. You can check the
man page on your system to identify what this parameter is called. You can find more
information about the KeepAliveTime and other associated TCP/IP configuration
parameters on Microsoft Windows in the following Microsoft Knowledge Base
articles: Q140325 and Q120642 )