Behavioral changes

NIFI-11523 introduced a fix that ensures the ScriptedTransformRecord processor uses the correct schema defined for the record writer. Previously, if the schema name attribute was set in the writer but not in the flow, it was ignored, defaulting to the reader schema. This behavior has been corrected, which may cause the processor to fail after upgrading if the schema name attribute is not set in the flow.

org.apache.nifi.schema.access.SchemaNotFoundexception: ${schema.name} did not provide appropriate Schema Name

To prevent failures, ensure that the schema name attribute is properly configured in the flow or match it to the schema defined for the record reader for identical behavior.

In Avro, when a schema is present, all required fields are defined by the schema. If a field is missing in the input data, a java.lang.NullPointerException is raised. Conversely, when there is an extra field in the input data not defined by the schema, the field is ignored, and the RecordWriter excludes it from serialization.

Avro 1.11 has introduced a more strict schema validation, which leads to changes in default value handling. This change might impact existing flows. In previous Avro versions, you cloud set the default value of a complex type as null' to indicate that it is not required.

However, with Avro 1.11 introduced, if a field is missing, parsing fails with org.apache.avro.AvroTypeException. To address this, it is now necessary to specify null as a valid type in the schema before you can assign it as a default value.