- I have set up a number of aggregate tables and explicitly defined these in the schema using AggPattern.
- One of my shared dimensions has a parent-child hierarchy (with a closure table).
Every time I drill into this specific dimension with parent-child hierarchy - mondrian ignores all aggregate tables and runs SQL on the underlying fact table.
If I drill into any other dimension without parent-child hierarchy - mondrian uses the appropriate aggregate tables.
As an experiment - if I eliminate only the parent-child hierarchy, mondrian uses the aggregate tables.