Customer is on ZMF 8.2 Patch 3. When the error started, the users who were already logged into ChangeMan ZMF could not continue with their work. Users were also unable to login in ChangeMan after the error started. Customer had to recycle the ChangeMan ZMF task to clear the problem.  Problem has happened more frequently since first reported.

Support informed customer the ‘stalled’ CMN subtasks in the task output could be batch jobs that completed, but were still connected to the ZMF task. The ZMF task is detaching 'stale' (no longer active) tasks that have connected to it. The S33E abend (which should not bring the task down) is working as designed. This abend simply means that a task was detached rather than having terminated on its own. This does not indicate an error, as it is done many times routinely during our normal processing. The abend S33E is simply the completion code of the task that was detached.

Customer has implemented the ‘TIMEOUT’ parm to detach inactive users/sessions, but problem still happens. Support asked about XML requests running without the proper disconnect, and the customer did not know of any.