System Copy : error in Table Splitting Preparation

[ashish_vikas]

Hello friends,

We were doing System Copy for system based on NW7.01 with oracle. During export phase on source system in phase table splitting preparation, we had a error.

Tablesplitfile:

APQD%5

CDCLS%3

PCL2%10

PCl4%5

PPOIX%10

Error Message:

An error occurred while processing option SAP ERP 6.0 EHP4 Ready - Support Release 1 > Software Life-Cycle Options > System Copy > Oracle > Source System Export > High-Availability System > Based on AS ABAP and AS Java > Table Splitting Preparation( Last error reported by the step :SQL statement or script failed. DIAGNOSIS: Error message: ORA-20100: Column name undefined invalid ORA-06512: at "SAPSR3.TABLE_SPLITTER", line 775 ORA-06512: at line 1 Disconnected from Oracle Database 10g Enterprise Edition Release 10.2.0.2.0 - 64bit Production With the Partitioning and Data Mining options . SOLUTION: For more information, see ora_sql_results.log and the Oracle documentation.). You can now:

Choose Retry to repeat the current step.

Choose View Log to get more information about the error.

Stop the option and continue with it later.

Log files are written to /tmp/export/sapinst_instdir/ERPEhP4/LM/COPY/ORA/EXP/HA/AS/PRETABSPLIT.

this error looked like like it requires a column name too?  The system copy guide is saying the format is only <table>%<nr of splits> 

So I'm not sure what format should be used.. and what should be used for column name if this is equired ?

Regards

ashish

[former_member189725 · Accepted solution]

Please refer the SAP note Note 1043380 - Efficient Table Splitting for Oracle Databases

for details on using the PL/SQL splitter for Oracle.

Please refer to Page 6 of the doc below

http://www.sdn.sap.com/irj/scn/go/portal/prtroot/docs/library/uuid/b070e7b4-a5ca-2c10-1dab-a4b53fd1c...

You will have to provide table_name, no of splits and column for any table . For cluster table do not provide PAGENO as the column for split and do not use ROWID splitting.

Regards

Ratnajit

[ashish_vikas · Accepted solution]

Hi Ratnajit,

Is it compulsory to use Oracle PL/SQL splitter for Oracle database(page 7).. Can't we use R3ta table splitter.

I wanted to use R3ta table splitter in my case.

Regards

ashish

[former_member215961 · Accepted solution]

Hi ashin Vikas,

You can use R3TA table splitter. After you enter the file with the particions sapinst will read the file and shows you the contents.

Cheers

[former_member189725 · Accepted solution]

If you want to use R3ta , then you can , refer to page 7 of the link

http://www.sdn.sap.com/irj/scn/go/portal/prtroot/docs/library/uuid/b070e7b4-a5ca-2c10-1dab-a4b53fd1c...

With R3ta , you only put the table name and the number of splits. R3ta will find out column with highest cardinality for the split if the column name is not available in the R3ta hnts file.

But I recommend you use PL/SQL spliiter if the source database is Oracle and if the target is non-Oracle do not use ROWID , use any column of the primary key with highest column. You can refer the column with highest cardinality from the SAP provided R3ta hint file for reference. PL/SQL splitter is optimized for such splitting and performs better than R3ta.

Regards

Ratnajit

[ashish_vikas · Accepted solution]

For Cluster tables.. what should be <col_name> .. means how to find value for <col_name> ?

Regards

ashish

[former_member189725 · Accepted solution]

You check the primary key , do not select the PAGENO field .

For the rest of the fields , run the following

select count(distinct(<column_name>)) from SAP<SID>.<CLUSTER_TABLE>;

The column which returns the maximum number of distinct values , should be used as the column for split.

Regards

Ratnajit

[former_member215961]

Hi ashih,

The format is correct for non cluster tables but make sure that you dont put weird characters.did you edit your split.txt file on your computer and then you copy to the server?

The table split file should be the format:

<table_name>%<no_of_splits>;

And cluster tables will require a column name at the end of it.

<table_name>%<no_of_splits>;<col_name>

Cheers.

[ashish_vikas]

Hi Sargan,

For Cluster tables.. what should be <col_name> .. means how to find <col_name> ?

And if these are not Cluster table.. <col_name> is not required.. right ??

Regards

ashish