I have an ETL job that performs cube refresh daily, but it sometimes fails
with key errors. It appears that this failure occurs whenever there are new
keys added to dimensions. It processes in the following order:

1) refresh dimensions ( I have tried both update and full process)
2) refresh the cubes

What are the requirements around dimension refresh? Should they be set up
to refresh in the order of their key usage (i.e., Dimension 1 followed by
Bridge Dimension which relies on Dimension 1, etc, etc) or is there
something else I'm missing?

If I run a manual refresh, I don't get the missing key errors.

I can post ETL if necessary, but I'm hoping it's something on the stupid
side that I've overlooked.

Thanks.

-Tim

1. think of it as a FK problem... Add the dimension member, then the fact,
then the M-M relationship if you need it.
2. Process the dimension first, before the facts... Otherwise the fact will
try to reference a dimension that doesn't yet exist.. ( maybe this is where
your problem is)
--
Wayne Snyder MCDBA, SQL Server MVP
Mariner, Charlotte, NC

I support the Professional Association for SQL Server ( PASS) and it''s
community of SQL Professionals.


"Tim Dot NoSpam" wrote:

Thanks.

I bet it's in the settings for the dimension updates. I always process the
dimensions first, but I bet I don't have 'process affected' setting enabled.

Appreciate it. It confirms where I thought I'd need to start looking.

-Tim

"Wayne Snyder" <wayne.nospam.snyder (AT) mariner-usa (DOT) com> wrote