There is a generic load job that reads a Dataset and writes it to an Oracle OCI stage. The job uses RCP entirely.
The error we are getting is:-
Item #: 14
Event ID: 57786
Timestamp: 2013-07-07 06:15:18
Type: Fatal
User Name: dsops
Message Id: IIS-DSEE-TFIP-00034
Message: ociTarget,0: Could not find input field "COST_CATEG".
Item #: 15
Event ID: 57787
Timestamp: 2013-07-07 06:15:18
Type: Fatal
User Name: dsops
Message Id: IIS-DSEE-TFDR-00050
Message: ociTarget,0: Fatal Error: Unexpected runtime failure in data set preparation.
The error is random because we can then re-run the job without any modification and it will succeed.
The field it is complaining about is actually named "COST_CATEGORY_SID: decimal[38,0];"
After reading on the forum about problems with reading and writing datasets using different configurations I checked what configurations were being used.
The Dataset is created in Overwrite mode under 2 nodes. It is then read in a different job using 1 node.
This job that has the random error later reads the Dataset under 2 nodes again.
Has anybody seen this error before or could shed any light on what might be happening here?
Dataset to Oracle OCI - Random RCP Error
Moderators: chulett, rschirm, roy
While I have no idea what could be happening, I would recommend adding "$OSH_PRINT_SCHEMAS" to your job in order to be able to trace if the column "COST_CATEG" is read in from the dataset and, if so, where it disappears in the stream.
The Dataset will be repartitioned as needed (but in this case it was written with a 2 node and subsequently read with the same 2-node configuration, so that shouldn't be part of the problem). You could explicitily repartition when reading the dataset if you wish.
The Dataset will be repartitioned as needed (but in this case it was written with a 2 node and subsequently read with the same 2-node configuration, so that shouldn't be part of the problem). You could explicitily repartition when reading the dataset if you wish.
</a>