Dry run a batch import before it writes anything

Before loading a VSAM dataset or an IMS segment, you can now dry run it. Turn on Dry run in Trigger batch job and choose how many records to check per source (1,000 by default, or blank for all of them).

A dry run reads and maps the records just as a real run does, then checks every row against its target table instead of writing it: empty values in NOT NULL columns, values too long, numbers too large, and values of the wrong type. It counts rows a real run would skip, those whose key is already in the table or was read earlier in the run, and it tells you how many rows would load. Nothing is written and no table changes. Its problems are listed as causes, as a real run's are, so Fix it works on them.

The job page marks a dry run with a dashed Dry run badge and banner, and offers Dry run again, to check a fix, and Run for real. A source that cannot be read doesn't stop the dry run: the other sources are still checked, and the page names the one that failed.

A dry run doesn't see foreign keys, check constraints or triggers; only a real run finds those. AI clients can start one over MCP and fix the mapping from its causes.