ODS activation process random error in process chain

Question

Dear All,
In my process chains, the process of ODS activation is sometimes in error. It is a random error, it is not always the same ODS and process chain.
The message error is DBIF_RSQL_SQL_ERROR.
I noticed that, on right-clic -> Change variant to the process of ODS activation, I have the following message:
"Not all infoobjects can be read".
For information, in the processes of activation of the ODS, the unique parameter is the name of the ODS to be activated.
Thank you very much for your help,
Arnaud

Answer 1

Hi......
Check SAP Note : 631668......
According to this note :
Symptom
This note applies only to BW systems under ORACLE.
Loading data into InfoCubes terminates with an error: ORA 60, "deadlock detected". The monitor log displays the entry "CALLER 70 missing" and the request is flagged as red.
Other terms
Caller 70, DEADLOCK, ora0060, bitmap, indexes, DBIF_RSQL_SQL_ERROR
CX_SY_OPEN_SQL_DB
Reason and Prerequisites
In SAP BW, there are two fact tables for including transaction data for Basis InfoCubes: The F and the E fact table.
Unlike the E fact table, the F fact table contains the information about the request from where the transaction data originates.
Therefore, transaction data is always written to the F fact table and the package dimension contains the request to which the loaded data belongs.
Using 'Request compression', the system reads the data of a request from the F fact table and writes it to the E fact table without request ID and compresses.
For good reporting performance, you must keep only the data of a few requests in the F fact table since the F fact table is partitioned according to the key of the package dimension and the data of all requests is always read. Therefore, you should compress the request if you are sure that the loaded data is correct and that you no longer have to delete the request.
When you install BW on an ORACLE database, BITMAP indexes are created on the fact tables to improve the reporting performance of the system.
These have a negative effect on the performance when loading data and the deadlock mentioned above (ORA 60) may occur as a result. The DEADLOCKs occur during parallel insert operations because ORACLE does not support a blocking concept for BITMAP indexes at data record level.
Solution
Before loading movement data, delete secondary indexes from the F fact table and create them again after the loading process. If the F fact table is small, no performance problems occur.
Alternatively, you can also load requests serially into InfoCubes by first loading the data only into the Persistent Staging Area (PSA) and then serially into the InfoCube. If you use the data transfer process (DTP) for loading in BW 7.0, this option is no longer available.
NOTE
           This avoids the deadlock problem. However, you should note that after some loading processes, it is necessary to reorganize the BITMAP indexes because they degenerate very heavily and, therefore, the read and writing performance deteriorates dramatically.
Hope this helps you.....
Thanks==Points as per SDN.........
Regards,
Debjani..........
Edited by: Debjani  Mukherjee on Sep 29, 2008 3:14 PM