Each of the objects described under Master Data must be validated using the respective object validation program indicated in its Table Names section.
In a limited number of cases object validation is available for Transaction Data objects, where customers may convert transaction data that is still pending. For the same objects you may also be converting historic records. You may not want to perform validation on completed records. As a result the background processes provided for transaction data allow you to limit the validation to records in a given status.
We strongly recommend validating each object in the following steps:
Execute each object's validation program in random-sample mode to highlight pervasive errors. When you execute a validation in random-sample mode, you are actually telling it to validate every X records (where X is a parameter that you supply to the job). Refer to Submitting Object Validation Programs for more information about the parameters supplied to these background processes.
You can view errors highlighted by validation programs using the Validation Error Summary transaction.
Correct the errors using SQL. Note, you can use the base package's transactions to correct an error if the error isn't so egregious that it prevents the object from being displayed on the browser.
After all pervasive errors have been corrected; re-execute each object's validation program in all-instances mode to highlight elusive, one-off errors. Refer to Submitting Object Validation Programs for more information about the parameters supplied to these background processes.
Take note! Whenever an object validation program is run, it is necessary to delete all previously recorded errors associated with its tables from the validation error table before it inserts new errors.
After the various object validation programs run cleanly, run the referential integrity validation programs as described in the next section.
Copyright © 2011, Oracle and/or its affiliates. All rights reserved.