apache/flink · error · IllegalArgumentException

Only simple types are supported in the second level nesting

Error message

Only simple types are supported in the second level nesting of fields '%s' but was: %s

What it means

Validation from the TypeInformation overload of CsvRowSchemaConverter.validateNestedField(): when converting an ARRAY or ROW field, each element type must itself be a simple number/string/boolean. If a nested element is itself complex (map, another row/array), conversion aborts with 'Only simple types are supported in the second level nesting' — CSV's Jackson schema has no representation for deeper structure.

Source

Thrown at flink-formats/flink-csv/src/main/java/org/apache/flink/formats/csv/CsvRowSchemaConverter.java:242

            for (LogicalType fieldType : rowType.getChildren()) {
                validateNestedField(fieldName, fieldType);
            }
            return CsvSchema.ColumnType.ARRAY;
        } else {
            throw new IllegalArgumentException(
                    "Unsupported type '"
                            + type.asSummaryString()
                            + "' for field '"
                            + fieldName
                            + "'.");
        }
    }

    private static void validateNestedField(String fieldName, TypeInformation<?> info) {
        if (!NUMBER_TYPES.contains(info)
                && !STRING_TYPES.contains(info)
                && !BOOLEAN_TYPES.contains(info)) {
            throw new IllegalArgumentException(
                    "Only simple types are supported in the second level nesting of fields '"
                            + fieldName
                            + "' but was: "
                            + info);
        }
    }

    private static void validateNestedField(String fieldName, LogicalType type) {
        if (!NUMBER_TYPE_ROOTS.contains(type.getTypeRoot())
                && !STRING_TYPE_ROOTS.contains(type.getTypeRoot())
                && !BOOLEAN_TYPE_ROOTS.contains(type.getTypeRoot())) {
            throw new IllegalArgumentException(
                    "Only simple types are supported in the second level nesting of fields '"
                            + fieldName
                            + "' but was: "
                            + type.asSummaryString());
        }
    }

View on GitHub (pinned to 2f3c205e92)

Solutions

1. Flatten nested rows inside arrays into delimited strings or separate columns before CSV writing.
2. Move to a hierarchical format (json/avro/parquet) for such schemas.
3. Restructure the row so arrays contain only scalars.

// Check array/row elements before schema build:
void checkNested(String name, TypeInformation<?> el) {
    if (!isCsvSimpleType(el)) throw new IllegalArgumentException(
        "Nested field " + name + " too deep for CSV: " + el);
}

static boolean isCsvSimpleType(TypeInformation<?> info) {
    return NUMBER_TYPES.contains(info) || STRING_TYPES.contains(info)
        || BOOLEAN_TYPES.contains(info);
}

When it happens

Trigger: A RowTypeInfo containing BasicArrayTypeInfo<ObjectArrayTypeInfo<RowTypeInfo...>> or ObjectArrayTypeInfo<MapTypeInfo...>: arrays-of-rows or arrays-of-maps inside the row.

Common situations: Reusing rich event types (orders with line-item arrays that are themselves rows) for a CSV DataStream sink without flattening.

Related errors

• Unsupported type information '%s' for field '%s'.
• Unsupported type '%s' for field '%s'.
• Unsupported type: %s
• Csv does not support TIME type with precision: %s, it only s
• Serializing the source elements failed: {e.getMessage()}

AI-assisted analysis of apache/flink@2f3c205e92 (2026-08-14). Data as JSON: /api/errors/8f7e9ca457f0e6f6. Report an issue: GitHub.