Testing can be tricky. What most people do is run a query against the relational data warehouse and retrieve a single value, such as the total sales for Product X for the current year to date. Then they run an MDX query against the cube and retrieve the same value. If the numbers don't match they know they have a problem. You can build a set of queries like this and automate the process. I've even seen people run the queries in the ETL after moving the data and processing the cubes.

Sorry, I missed your message.

Normally you will load a star schema in the ETL. Once that is done, you kick off the cube processing.

Once the cubes are processed, you can issue queries against the relational warehouse such as "Year-to-date sales." You then issue an MDX query against the cube to get the same data. If the values don't match you can send an email or otherwise notify the BI team of a problem.