Occurrences such as a network failure, no disk space on the database, or a bad aggregate request result in aggregate persistence errors.
When a series of aggregates are being created, and the creation of one aggregate fails, the aggregate persistence engine skips creation of the failed aggregate and its dependencies and proceeds to the next aggregate in the list. Check the log files to identify failed aggregates.
If there are errors, you must remove the failed aggregates in one of the following ways:
Manually remove the aggregates from the metadata and the database. To identify the aggregate metadata, you can query the repository using the
Is System Generated filter for physical tables and logical table sources. See Querying the Repository.
Automatically remove the failed aggregates using the Delete Aggregates specification. In particular, use this technique to remove any orphan aggregate dimensions, those not joined to any other fact table.
Run the Model Check Manager to ensure that your repository does not contain modeling problems that will affect Oracle BI Summary Advisor and aggregate persistence performance and results. See Using Model Check Manager to Check for Modeling Problems.