The July 2025 release includes changes in behavior for the following connectors.
Delta files
Effective in this release, Amazon S3 V2 and Microsoft Azure Data Lake Storage Gen2 Connectors include the following changes in behavior when you write to a delta file:
•The Secure Agent ignores the target file name override specified in advanced target properties and retains the existing target file name. Previously, the Secure Agent replaced the existing file name with the specified override.
•The Secure Agent applies Snappy compression by default, regardless of whether a compression format is selected. Previously, files were uncompressed by default, and the selected compression format was applied only when specified.
Amazon S3 V2 Connector
Effective in this release, when you write decimal values to an Avro file, the values are correctly written to the target.
Previously, certain decimal values were incorrectly written to an Avro file.
For example, the value 441265.1234 in the source was written as 4.412651234 in the target.
IBM MQ Connector
Effective in this release, when you run an IBM MQ mapping task to read message data from queues in real time and a transformation in the pipeline abruptly stops, the Secure Agent terminates the mapping task and stops reading message data from queues in IBM MQ.
Previously, even if a transformation in the pipeline abruptly stopped, the Secure Agent did not terminate the mapping task and continued to read message data from queues.
Microsoft Azure Synapse SQL Connector
Effective in this release, when you read data from Microsoft Azure Synapse SQL and the source table uses a UTF-8 compatible collation, you can read data directly from Microsoft Azure Synapse SQL without staging the data.
Previously, if the source table used a UTF-8 compatible collation, the mapping failed as it used Polybase queries that do not support UTF-8 collations.
Microsoft Fabric Data Warehouse Connector
Effective in this release, when you create a Microsoft Fabric Data Warehouse target at runtime where incoming string fields have a precision greater than 4000, the mapping preserves the precision in the new target.
Previously, the mapping did not preserve the precision of the incoming field and set the precision to 8000 by default.
MongoDB V2 Connector
Effective in this release, when you run a MongoDB V2 mapping to write data to a MongoDB V2 target configured with a primary key update column in the Target transformation, the mapping updates or upserts data in all rows that contain the same primary key value.
Previously, the mapping updated or upserted data only in the first row that contains the same primary key value.
This changed behavior doesn't apply to a mapping in advanced mode and it updates or upserts data only in the first row.
Snowflake Data Cloud Connector
Effective in this release, Snowflake Data Cloud Connector includes the following changes in behavior:
•A Snowflake mapping rolls back the partial data written to the target and fails with an error message in the session log when all the following scenarios are true:
- The mapping writes to a Snowflake target.
- The target is configured with the onError=ABORT_STATEMENT and oneBatch=true load properties in the Additional Write Runtime Parameters field in the Target transformation.
- A transformation in the mapping pipeline fails.
Previously, the mapping failed without an error message and wrote the partial data to the Snowflake target.
If you want to retain the previous behavior, set the -DSNOWFLAKE_ENABLE_TARGET_ROLLBACK_ON_ERROR JVM option property value to false in the Secure Agent properties.
This changed behavior doesn't apply to a mapping in advanced mode and it fails without an error message and writes the partial data to a Snowflake target.
•When you run a mapping that writes a large record with the binary fields of high precision and optimize the write operation by setting the INFA_DTM_STAGING_ENABLED_CONNECTORS staging property in the Secure Agent properties, the mapping creates a single staging file that contains the binary fields.
Previously, the mapping created multiple staging files for the binary fields of high precision.
If you want to retain the previous behavior, set the DTMDisablePartialEORStagingSupport custom property value to yes in the Secure Agent properties.
This changed behavior doesn't apply to a mapping in advanced mode and it creates multiple staging files for the binary fields of high precision.