I recently installed a 3.1.0.6 client on an SCO UNIX system. It is
communicating with a 3.1.2.0 MVS server. I started a scheduling process for
prompt mode scheduling and ran incremental backups under control of the
central scheduler for several days with no apparent problem. Yesterday morning
I found that a scheduled backup had failed. The server log showed the backup
session starting, and them reported the session timing out because of an hour
with no activity about an hour and two minutes later. On the client a ps
command showed the scheduler process still running. The dsmerror.log file did
not show any error messages from the failed backup. The dsmsched.log file
ended in the middle of a line. The truncated line indicated that it was
starting to back up /dsmsched.log.

I used the 'define clientaction' command to request another incremental
backup. After about half an hour of pending status I killed the scheduler
process and restarted the scheduler process in the foreground on the client.
The backup started almost immediately and hung up within a couple of minutes,
apparently in the same way as it had the night before. The terminal output
from the scheduler indicated that it was starting to back up /dsmsched.log. At
that point ps commands showed the scheduler process still running and having
one child process, which was defunct. A netstat command showed an established
TCP connection between the client system and port 1500 on the MVS system.
Repeated 'query session' commands on the ADSM server showed the same amount of
input and output data, with several megabytes sent to the client and 711 bytes
received from the client. The backup was to be sent directly to tape. The
server did not mount a tape for the session.

Does anyone recognize this as a known problem?