- Ensure that level-1 trace points are in the special set for all CICS components.
Select special tracing for all user tasks, by setting up special tracing for
all user transactions and all terminals, and disable standard tracing by setting
the master system trace flag off.
- Use
the CETR transaction to set up the tracing options, and select auxiliary trace
as the trace destination. When you get the symptoms that tell you that the
storage violation has occurred, take a system dump—unless the error causes
a system dump to be taken.
- Format the system dump, and format and print the auxiliary trace data
set. If you know which area of storage the violation occurred in, you can
use appropriate dump formatting keywords. Otherwise, you need to format the
entire system dump. The dump formatting program may report that it has found
some incorrect data. If not, you need to find the overlaid area by other means.
- Locate all the entries in the trace table that address the overlaid area.
Operations involving GETMAIN and FREEMAIN in particular are likely pointers
to the cause of the error.
- When you have found a likely trace entry, possibly showing a GETMAIN or
FREEMAIN addressing the area, find the ID of the associated transaction by
locating the trace entry for TASK ATTACH. Rather than locating this manually,
it is probably better to reformat the auxiliary trace data set selectively
to show just trace entries corresponding to the task’s unit of work.
- Having found the identity of the transaction, take a look at all the programs
belonging to the transaction. It is likely that one of these caused the overlay,
and you need to consider the logic of each to see if it could have caused
the error. This is a long job, but it is one of the few ways of resolving
a storage violation affecting an innocent transaction.
What to do if you still cannot find the cause of the overlay
If you are unable to identify the cause of the
storage violation after carrying out the procedures of the preceding section,
contact your IBM® Support
Center. They might suggest coding a global trap/trace exit to
detect the storage violation.