We are experiencing problems with S80A abends in some batch jobs.
In these jobs we see DD statements being allocated to the View database.
What is causing the abend and the View database extents to be allocated?
The reason for a batch job to show the allocation of all of the View database extents would be because the job output goes through CA Deliver's pre-spool processing and the output is being written directly to a View database (RMOPARM ARCHn=DIRECT[/...]).
As matter is spread throughout a View database, index and data, when it comes to doing a write of data to View, from Deliver, all of the extents would need to be allocated because there is no way up front to tell as to where that writing will be occurring, as it can occur to multiple places in the database.
In Deliver, rather than use of the ARCHn=DIRECT[/...] setting, the ARCHn of ARCHn=class/dest/form can be used, which would match the criteria for a corresponding View database's collection criteria, whereas the database extents would not be allocated.
A review of a client dump showed that 8MB of below-the-line storage was exhausted in a sort, which was all of the below-the-line storage.
It was recommended that the client change sort OPTION RESALL, in the below statement, to a larger amount, like 1M:
Also, it was recommended to add field RESINV=1M.
The sort OPTION statement can be overridden with a JCL //SORTCNTL DD statement, in the job step having the problem.