Understanding the behavior of failed Salesforce batch operations that do not roll back successfully inserted records in the batch

When a batch insert failed due to incorrect data in the second or subsequent batches, the driver inserted one or more records of non-primitive data types into Salesforce.

Oct 18, 2024
Knowledge

Title

URL Name

Salesforce-Failed-batch-operations-no-roll-back-successfully-inserted-records

Article Number

000265689

Environment

Product: Connect for JDBC Salesforce driver Version: All supported versions OS: All supported applications Database: Saleforce Application: All supported applications

Question/Problem Description

When a batch insert failed due to incorrect data in the second or subsequent batches, the driver inserts one or more records of non-primitive data types into Salesforce.

Steps to Reproduce

To reproduce this issue, insert two rows of data into the ACCOUNT table using a batch operation. The second of the inserted rows should include a column of a non-primitive type. For example, the ParentID column that is of the Hierarchy type This behavior can be observed by running a SQL query that inserts two rows using the following sample data String sql = 'insert into Account (Name, ParentId) values (?,?)'; PreparedStatement createStatement = con.prepareStatement(sql); createStatement.setString(1, 'batch1'); createStatement.setString(2, '0017F000007xE1f'); createStatement.addBatch(); createStatement.setString(1, 'batch2'); createStatement.setString(2, '123'); createStatement.addBatch(); Note that the second row in the sample data has invalid data '123' for the ParentID(Hierarchy type) column. When the user attempts to insert these values, an error is thrown. Observed behavior: The first row in the batch is inserted despite the mentioned error being returned.

Clarifying Information

Although Salesforce supports batches, the service does not support them atomically. If an error occurs when performing a batch operation, the service does not roll back the entire batch.

For non-primitive types, the driver is unable to verify if data is compatible with the corresponding column type. This is expected, and the responsibility of performing the verification falls to the Salesforce service in this scenario. If the data is determined to be incorrect, the service returns an error, however, because Salesforce does not support batches atomically, any inserted rows are not rolled back.

For primitive types, the driver implicitly checks whether a cast to the corresponding SQL type is successful. If the cast is unsuccessful, the driver returns an error and does not process the request. As a result, no rows are inserted if the batch operation fails.

Error Message

java.sql.BatchUpdateException: [DataDirect][SForce JDBC Driver][SForce]Parent Account ID: id value of incorrect type: 123

Defect Number

Enhancement Number

Cause

Resolution

The behavior described in this article is the expected for a Salesforce database.

Refer to the following Salesforce documentation for more information:
https://developer.salesforce.com/docs/atlas.en-us.api_asynch.meta/api_asynch/asynch_api_batches_failed_records.htm

Workaround

You can work around this issue by correcting failed records and executing the batch again. To identify failed records, refer to the update count array at the appropriate index for the failed records.

Notes

Keyword Phrase

Last Modified Date

10/18/2024 6:05 AM

Understanding the behavior of failed Salesforce batch operations that do not roll back successfully inserted records in the batch

Files(0)

Was this article helpful?

Was this article helpful?

Understanding the behavior of failed Salesforce batch operations that do not roll back successfully inserted records in the batch

Information

Related Articles