By default, the Stream Operator will create the error table. If the error table already exists, Stream Operator will terminate the job with an error message. The exception is on a job restart. On a job restart, Stream Operator will continue.
- When the value of the AppendErrorTable attributes is YES, the Stream operator reuses the error table and displays the number of rows already existing in the error table.
- If the error table does not exist, the Stream operator creates the error table before data loading and continue.
- If the structure of the existing error table is not compatible with the error table the Stream operator expects, it terminates the job with an error message before data loading.
A new column, RowInsertTime, was added to the Stream Operator error table for version 13.10. Therefore, you can no longer reuse error tables from prior versions, as valid Stream Operator error tables now have ten columns rather than nine.
If the user requested a queue error table and the max size of the job's schema is 64K, the Stream operator will create the HOSTDATA column as VARBYTE(63666) and the job will continue.
If the user requested a queue error table and the size of the job's schema is or could be greater than 64K, the Stream operator will terminate the job with a TPT16190 error message:
TPT16190: Error: Row size (128005 bytes) is too big to create queue error table.
If the HOSTDATA column is defined as VARBYTE(63666) in the operator's error table and the job's schema is or could be greater than 64K, the Stream operator will terminate the job with the TPT16191 error message:
TPT16191: Error: Row size (128005 bytes) is too big for the error table of which the HOSTDATA column is defined as VARBYTE(63666).