After spending time debugging my analysis services cube for the second time having this error, I guess it will be a good idea to blog about it, so I can remember it to another time 🙂
I got the above error after I had modified a dimension, so used it three fields for the key, instead of one. I made the relationship from my facts, based on the 3 keys, and then I deployed and processed the cube. And then I got the above error.
The reason for the error is, that I had the partitions for the fact table, based on some SQL sentences, and I had forgotten to modify these partitions after I had made the new relation.
So the conclusion: When you get the error “Invalid column name ‘x’.; 42S22” in analysis services, it is a problem related to the partitions design.