XCF does not provide any recovery for the status user routine. Routines that require recovery must establish their own. XCF does place sufficient information into the SDWA to identify the status user routine that was in control. The multisystem application must provide whatever diagnostic data is required for problem determination for the status user routine.

XCF will reschedule a status user routine that suffers an error. However, the status user routine should not rely on this as a means of recovery.

Members that identify a status user routine should allow for SRB-to-task percolation. (Note: SRB-to-task percolation does not work for address space associated members. See Member Association for more information.) If XCF processing fails, and XCF does not retry, XCF abnormally ends the task that the member is associated with (either the task or the job step task as specified on IXCJOIN) with a retryable system completion code 00C and reason code 05070000. However, the task's recovery routine does not have to take any action, because the status user routine completed its function before giving up control.

To ensure that the member's recovery can intercept SRB-to-task percolation, the task that the member is associated with must ensure that its recovery routine always receives control when a task abnormally ends. To accomplish this, the associated task should issue the WAIT macro and continue waiting indefinitely while other tasks perform the member's work.

SRB-to-task percolation does not occur while the task's recovery routine is running. XCF waits until the task is not in recovery before abnormally ending the task.