Re: Strange processing error.

I've a sneaking suspicion that this may be down to memory, but nothing to
base that on other than the fact that I have seen memory problems with
Analysis Services previously (especially when trying to share a box w/ a SQL
Server).

If you can, try moving the SQL Server onto a (topologically close) separate
server, and see if this helps at all (if this is even an option).

Do you have any entries in your event log / SQL logs?

"Radu Colceriu" <radu_colceriu@xxxxxxxxxxx> wrote in message
news:%23dD2JhNXIHA.3400@xxxxxxxxxxxxxxxxxxxxxxx
The SQL and AS share the same server. The datasource for the AS database
is on the same SQL.

Error from last night processing.

Error: 2008-01-22 08:05:16.52 Code: 0xC1000007 Source: cube
Analysis Services Execute DDL Task Description: Internal error: The
operation terminated unsuccessfully. End Error Error: 2008-01-22
08:05:16.52 Code: 0xC1000007 Source: cube Analysis Services
Execute DDL Task Description: Internal error: Th... The package
execution fa... The step failed.

Processing the cubes manually this morning worked perfectly. Nothing
changed in the datasource.

It is very strange and it's driving me nuts :). Also my customer is not
very happy at the moment. :/

So, any help would be highly appreciated.

Cheers,
Radu

Will Alber wrote:
Haven't seen this one before. Perhaps running a server side trace on the
server(s) from which the data is being acquired would help shed some
light on where the problem is occurring? The fact that it's only
sometimes is, as always, a concern.

"Radu Colceriu" <radu_colceriu@xxxxxxxxxxx> wrote in message
news:eQnkFMBXIHA.4532@xxxxxxxxxxxxxxxxxxxxxxx
I have a strange problem processing a cube.

The cube has 6 big measure groups (MOLAP) each with 2-4 partitions each
with 40 - 70 million records.
There are also some smaller measure groups (MOLAP) feed it with fact
tables in range of 2 - 9 million records.

There are around 7-10 dimension connected / measure group.
I have 2 "special" dimensions:
Articles with 2.4 million records and Customers with around 300.000. The
dimension design I think is pretty good (attribute relationship a.s.o.)
as the query and processing performance is quite good.

The server is an 4x dualcore @ 3GHz with 8G RAM. The SQL server and AS
is running on the same machine.

SQL server has allocated an max of 5G ram, the AS has the low memory
threshold by 40%. The server is x64 Enterprise SP 2 build 3215.

I'm doing full processing. First the dimensions and then the cube.
When it is running ok it takes around 1h.

Problem:

Sometime, especially in the night, during processing I get an OLEDB
error like:


Started: 5:13:35 PM
Error: 2008-01-20 17:30:04.67
Code: 0xC1060000
Source: cube Analysis Services Execute DDL Task
Description: OLE DB error: OLE DB or ODBC error: Unspecified error.
End Error

I process the cube via SSIS but I get this error sometimes also when I
process them manually.

Does anybody have an ideea?

I've tried making more and smaller partitions but it didn't help.

If someone have some ideas please let me know.

Thanks,

Radu


.

Quantcast