Hey all
This is not as simple as it sounds...
My cube is fully optimized, there for there are no joins to dimension tables
in its process select (which my cause records to fall).
Also i see no alerts about missing dimension keys errors nor do i find
missing records in the cube's processing log.
Finaly cubes process finishes successfully.
Still number of processed rows is significantly lower than number of records
in facts.
My cube is partitioned over time dimension.
Partitions are not sliced with specific time members but include a filter
clause
which specifies the time condition which should be met by facts records.
I have checked this field and it is not missing any values in facts.
I am using AS2000 sp3 on a 2000 server.
Cube's data source is in oracle 8i.
My line of thoughts goes to comunication problems while processing
but..usually when this kind of problem occurs i get a massage reporting
connection to data source was lost.
I am stuck here.
Would appriciate any educated guess or ideas about where/how to look further.
Thanks alot
Rea

I would suspect: "a filter clause which specifies the time condition
which should be met by facts records." If you have a "count" measure in
the cube, can you check which partition(s) are falling short (assuming
that you can translate each filter to some MDX filter on time)? Then you
could trace the SQL query to Oracle for that partition, and compare row
counts returned by the query to the cube count.


- Deepak

Deepak Puri
Microsoft MVP - SQL Server

*** Sent via Developersdex http://www.developersdex.com ***

Thanks Puri
The problematic partition is the last one in cube (it is the only one that
is being processed).
Its select is very simple:
"select list of fields from facts where time field meets time condition".
When this select is run localy at oracle number of records is
significantly different then when it is run from AS during this partition
process.

Thanks again
Rea

"Deepak Puri" wrote: