Hi,
XCF has its own recovery. If your message exit recovery fails to
field the problem, it percolates to XCF recovery. By default, XCF recovery
will always attempt to retry. Since the error was retriable, XCF retried.
Thus no SRB to Task percolation occurs. You would need to create a
non-retriable error so that XCF itself cannot retry.
Is there a need for an interface change here? For example, one might
envision a keyword on the IXCJOIN macro that permits the member to indicate
that they want SRB to Task percolation to occur if the exit routine fails
to field the error. Or perhaps a means for the message exit to tell XCF
recovery whether it wants this particular failure to be percolated out to
the task.
The hard part for the task that gets thumped is that there is no
context. As it stands now, when the task recovery gets control it can
inspect an abend reason code set by XCF to indicate whether something was
lost or not. In the case where something is lost, there is no information
as to what the something might have been. That is, with respect to a
message exit failure, there is no information about which message was being
processed at the time of the failure.
Mark A. Brooks
z/OS Sysplex design and development
845-435-5149 T/L 8-295-5149
Poughkeepsie, NY
[email protected]
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
