spring-projects/spring-ai · error · IllegalArgumentException

Field {0} has unsupported type {1}

Error message

Field {0} has unsupported type {1}

What it means

RedisVectorStore.schemaField() maps each MetadataField to a RediSearch schema field, supporting only NUMERIC, TAG and TEXT field types. A MetadataField with any other fieldType hits the default branch and throws this IllegalArgumentException. It prevents building an FT.CREATE schema with an unmappable field type.

Source

Thrown at vector-stores/spring-ai-redis-store/src/main/java/org/springframework/ai/vectorstore/redis/RedisVectorStore.java:711

			.attributes(vectorAttrs)
			.as(this.embeddingFieldName)
			.build());

		if (!CollectionUtils.isEmpty(this.metadataFields)) {
			for (MetadataField field : this.metadataFields) {
				fields.add(schemaField(field));
			}
		}
		return fields;
	}

	private SchemaField schemaField(MetadataField field) {
		String fieldName = jsonPath(field.name);
		return switch (field.fieldType) {
			case NUMERIC -> NumericField.of(fieldName).as(field.name);
			case TAG -> TagField.of(fieldName).as(field.name);
			case TEXT -> TextField.of(fieldName).as(field.name);
			default -> throw new IllegalArgumentException(
					MessageFormat.format("Field {0} has unsupported type {1}", field.name, field.fieldType));
		};
	}

	private VectorAlgorithm vectorAlgorithm() {
		if (this.vectorAlgorithm == Algorithm.HNSW) {
			return VectorAlgorithm.HNSW;
		}
		return VectorAlgorithm.FLAT;
	}

	private String jsonPath(String field) {
		return JSON_PATH_PREFIX + field;
	}

	@Override
	public VectorStoreObservationContext.Builder createObservationContextBuilder(String operationName) {
		VectorStoreSimilarityMetric similarityMetric = switch (this.distanceMetric) {

View on GitHub (pinned to 98a7beda4f)

Solutions

1. Use only FieldType.NUMERIC, FieldType.TAG or FieldType.TEXT when defining metadata fields for the Redis store
2. Coerce unsupported types: map BOOLEAN to TAG, dates to NUMERIC (epoch) before building the store
3. Check for null/default FieldType values coming from shared configuration or other store adapters

Example fix

// before
new MetadataField("active", FieldType.BOOLEAN)
// after
new MetadataField("active", FieldType.TAG)

java.util.Set<FieldType> allowed = Set.of(FieldType.NUMERIC, FieldType.TAG, FieldType.TEXT);
fields.forEach(f -> { if (!allowed.contains(f.fieldType)) throw new IllegalArgumentException("Unsupported field type for Redis store: " + f.fieldType); });

static boolean isSupportedFieldType(MetadataField f) { return f.fieldType == FieldType.NUMERIC || f.fieldType == FieldType.TAG || f.fieldType == FieldType.TEXT; }

try { RedisVectorStore.builder().metadataFields(fields).build(); } catch (IllegalArgumentException e) { if (e.getMessage().contains("unsupported type")) { /* remap field type */ } }

When it happens

Trigger: Constructing the RedisVectorStore with MetadataField instances (via builder metadataFields or an observed metadata mode) whose fieldType is not one of NUMERIC/TAG/TEXT, e.g. a null or custom FieldType value.

Common situations: Copy-pasting MetadataField definitions from another vector store that supports more types (e.g. BOOLEAN or CHUNK); refactor renamed/extended FieldType enum; building fields programmatically with a wrong enum constant.

Understand the failure class

Background: Invalid enum value errors: "Unknown type", "Invalid scope", "must be one of" — when a string is not on the library's allowed list — this error's family across 23 libraries.

Related errors

• Failed to create index:
• Not allowed filter identifier name:
• Field type {0} not supported
• Expression type {0} not supported for numeric fields
• Numeric value must be a Number

AI-assisted analysis of spring-projects/spring-ai@98a7beda4f (2026-09-11). Data as JSON: /api/errors/ce22fa5b2936c432. Report an issue: GitHub.