DSXchange

hi,
Job was running fine and when i migrated the job, i am getting the error that there was a "ds_uvput() - Write failed for record id ' 880'". Can any one please help me out with the problem.
Thanks
Raj

You can not insert NULL value into a key column of a Hashed File, this error is pointing to the 880th record, check whether this row from source is containing any NULL in the key column or not.

i did check and there is no NULL Values in the Key Column. please let me know what can be the problem.
Thanks
Raj

raj9176 wrote:i did check and there is no NULL Values in the Key Column. please let me know what can be the problem.
Thanks
Raj

Raj can you let us know how did you check for nulls from source. Go to the director pick the select query that it shows for the source data and try executing it directly against your database and see for any null or empty strings in the columns.
Is there any transformation happening after the source before writing to the Hashed File? Are you applying any transformation from your key columns?

Can you put a derivation of OCONV(In.Key,'MCP') for your key as a test, this will replace unprintable chars with a period ".". This would show that your key perhaps does contain a field mark or value mark (which is not allowed) or a null.

loveojha2 wrote:You can not insert NULL value into a key column of a Hashed File, this error is pointing to the 880th record, check whether this row from source is containing any NULL in the key column or not.

Just a slight correction, it's not the 880th record, the message:

Write failed for record id ' 880'".

means that the primary key value (the record id) is invalid for some reason. See that space before the 880? I suspect that the space is really a reserved character, something really low or really high on the ASCII chart. Arnd has suggested that the poster investigate the row where the primary key contains reserved characters and that is the proper solution.

[quote]Just a slight correction, it's not the 880th record, the message:
Write failed for record id ' 880'".
means that the primary key value (the record id) is invalid for some reason.[quote]

Correction. Something somewhere in record with ID = '880' is invalid.
It need not be in the key column, but can be in any of the columns that is attempted to be written.

Have you given the full error message?

Why dont you try to write in source query that where key column is not null.

Thanks,
Anupam

Write the data into a text file then use a hex editor (UltraEdit is good, and free for 45 days) to see exactly what you're trying to write. Post results here.