When you do not add rule input columns to the scoreboard, the mapping fails.

When you run a mapping that is not open in the editor, the mapping fails to run with "java.lang.IllegalArgumentException" error message. After the error occurs, the option to run the mapping is grayed out.

Random masking in a Data Masking transformation does not retain the original number of characters in the masked result if you do not enter a range.

When the Spark engine runs a mapping, the mapping fails to write to the target when the following conditions are true:

The mapping reads from a binary complex file data object.
The mapping writes to a flat file target.
The mapping uses Kerberos authentication.
The mapping runs on HDP 3.1.5 or HDI 3.6 ADLS Gen 1 clusters.

When a domain refresh update takes more than the refresh interval time, the instance needs to be logged in node log file.

When you run a dynamic mapping that uses an input rule to rename generated ports and retain the port order of the nearest transformation with static ports, the mapping does not rename the ports and does not retain the port order in the target.

A mapping that runs on Blaze, and that has "Support mixed-case identifiers" enabled in the Hive connection with any SQL Identifier Character, fails with an error like: "Failed to fetch the location of the database."

When you search for a relational object to import, the search takes an excessive amount of time.

When the Data Integration Service runs a large number of rules against a moderately sized data set for multiple hours, and the mapping contains more than one Joiner transformation having a single source, the JVM where the Data Integration Service resides crashes.

When you configure a Data Masking transformation, the

Select Source

button for the dictionary connection is not visible unless you maximize the dialog box.

When you cancel a data preview job on the Spark engine, the behavior is unpredictable.

The Data Integration Service does not recover mappings configured to run on the Spark engine that have mapping outputs even if you enable data engineering recovery for the mapping.

When you enable data engineering recovery, the Monitor view of the Administrator tool does not display summary statistics for recovered jobs if the Data Integration Service stops after the application is submitted to the cluster.

Effective in version 10.4.1, a Spark job that runs on a Hadoop cluster uses the delegation tokens fetched from the cluster, and not from the Data Integration Service.