The Distributed FileManager/MVS support runs under an EUT FRR. For special situations, the recovery environment is temporarily changed to an ESTAE environment. When a system logic or hardware error is detected by Distributed FileManager/MVS, first failure data capture (FFDC) is performed. This results in a dump that includes all of the pertinent data necessary for debugging the error. The error is also recorded in the system LOGREC file. After the error is recorded, retry and/or end-of-memory processing is performed. This process consists of flushing the buffers and cache, closing all open data sets, and freeing all outstanding locks, latches, and other resources.

If the detected error impacts the entire conversation, a "permanent agent error" (AGNPRMRM) reply message is returned to the source system, and the conversation is terminated. If the detected error impacts only the agent, AGNPRMRM is returned to the source, and the agent is terminated. If the detected error only impacts the command being processed, a "command check" (CMDCHKRM) reply message is returned to the source system, and the command process is terminated. For details about AGNPRMRM and CMDCHKRM reply messages, see DDM reply messages returned by Distributed FileManager/MVS.