Mload upsert - 6705 - SAP BW as source


Mload upsert - 6705 - SAP BW as source

Greetings Experts,

We use Informatica to load data from SAP BW to Teradata staging layer and could see a job not dropping the ET table (some records exist with error code 6705) during Mload operation (upsert) and succeeds.  During the next run of the job, it fails in its first phase of Mload as it couldn't create the ET table which already exists.

How to identify the source records (for which we don't have access to)  that have violated and landed into ET table?  (The source data is landed into Informatica server through the flat file, data is kept in pipe in Informatica which is deleted once the job is completed)

If Teradata has database_link like Oracle, we could have accessed the source data of SAP BW right from Teradata and could have troubleshooted the issue.  Can you explain why Teradata implicitly doesn't support db link (I guess).  I have gone through the following link and couldn't find a implicit db_link support (for external source systems like SAP, oracle,...) rather than through external tools.

One option is to find all the distinct values on the column which is recorded in ET table of Teradata from the source SAP BW system and identify the records missing in source which may be a tedious task or

use a flat file in Informatica rather than pipe so that we could read the file.

Are there any provision to find the missing records right from Teradata?

Tags (3)
Teradata Employee

Re: Mload upsert - 6705 - SAP BW as source

Hi Cheeli,

Please refer to "Handling Teradata MultiLoad Errors" section in MultiLoad manual to for error recroding.




Re: Mload upsert - 6705 - SAP BW as source

Error code 6705:Is an error in a string having Multi-byte characters. So you need to correct the SQL or data as Mload doc suggests.

Same like in all ETL tools, you should code a logic in your job such that if ET table exists, then

fail the job. You need to correct the data etc and re run the job.But you need to take care of restartability features.

You should have a perfect clean load and then delete file.Check Acquisition error and application error table in a script maybe. If pipe is a problem, then think about alternative, like landing a file

in a directory and flush them post successful load.

Teradata has dblink-like feature and it is even more advanced. Now  it connects only to Oracle.

There will always be problems when you move data from one environment to another, but they are few and far between, since logical programs take care.