Handling Failed Records in Batches

A batch can have a Completed state even if some or all of the records have failed. If a subset of records failed, the successful records aren't rolled back. Likewise, even if the batch has a Failed state or if a job is aborted, some records could have been completed successfully.

When you get the batch results, it's important to look at the Success field for each result row to ensure that all rows were processed successfully. If a record was not processed successfully, the Error column includes more information about the failure.

To identify failed records and log them to an error file:

1. Wait for the batch to finish processing. See Getting Information for a Batch and Interpreting Batch State.
2. Get the batch results.

   The following sample CSV batch result shows an error for the last record because the LastName field was missing:

   1"Id","Success","Created","Error"
   2"003D000000Q89kQIAR","true","true",""
   3"003D000000Q89kRIAR","true","true",""
   4"","false","false","REQUIRED_FIELD_MISSING:Required fields are missing: [LastName]:LastName --"

3. Parse the results for each record:
   1. Track the record number for each result record. Each result record corresponds to a record in the batch. The results are returned in the same order as the records in the batch request. It's important to track the record number in the results so that you can identify the associated failed record in the batch request.
   2. If the Success field is false, the row was not processed successfully. Otherwise, the record was processed successfully and you can proceed to check the result for the next record.
   3. Get the contents of the Error column.
   4. Write the contents of the corresponding record in the batch request to an error file on your computer. Append the information from the Error column. If you don't cache the batch request that you submitted, you can retrieve the batch request from Salesforce.

After you have examined each result record, you can manually fix each record in the error file and submit these records in a new batch. Repeat the earlier steps to check that each record is processed successfully.