Thomas, Did you capture a DUMP?  If so, you should open a PMR with IBM service
and make the DUMP available to us
for analysis.  Include the entire trace back including the MVS dump header
information along with any messages prior to the
abend.  In other words, the last several hundred records of the joblog should be
provided inthe PMR.  To capture a dump
for us you can either specify SYSMDUMP in your server startup JCL or set a SLIP
for the ADSM job.
SLIP SET,ACTION=SVCD,COMP=0C4,JOBNAME=ADSMJOBNAME,SDATA=(RGN,CSA),END
You should open a PMR for this problem and IBM service should be able to give
you instructions for sending the DUMP via
tape or instruct  you to put the DUMP on testcase.boulder.ibm.com  in the toibm
directory if you have internet access.
Meanwhile, if you could send the last couple hundred records of the joblog to
me, I might be able to determine whether the abend
is caused by a problem already fixed in 3.1.2.40  Some problems manifest
themselves  differently depending on the code path etc.
Thanks, Bob La Brie MVS Server Development