Your best course of action is to inspect the TSM server Activity Log
around the time of the scheduled backup, looking for ANR problem
condition messages.

Your action to rerun the backup manually is probably at a time of day
other than when the schedule runs, where the latter may be
encountering resource contention or shortage problems.  But assure
that the two invocations are using exactly the same environmental
parameters: if the scheduled one happened to latch onto an options
file which specifies a nodename having a MAXNUMMP of 0, then that's
one cause.

See ADSM QuickFacts for causes we've all encountered for the ANS1312E
message.

   Richard Sims     http://people.bu.edu/rbs