Failed to obtain the embedding dimensions from the…

Error message

Failed to obtain the embedding dimensions from the embedding model and fall backs to default: ${embeddingDimension}

What it means

MilvusVectorStore.embeddingDimensions() attempts embeddingModel.dimensions(); on exception or a non-positive value it warns and falls back to the OPENAI_EMBEDDING_DIMENSION_SIZE (1536) constant rather than the configured embeddingDimension shown in the message text. The collection schema may then be created with the wrong vector field dimension.

Solutions

Set the dimension explicitly via MilvusVectorStore.builder(...).withDimension(n) so no detection is attempted
Ensure the embedding service is reachable and credentials valid before store initialization
Check that your EmbeddingModel implementation returns a positive dimension
If the collection was created with the wrong dimension, drop it and re-create after fixing configuration

Example fix

// before
MilvusVectorStore.builder(milvusServiceClient, embeddingModel).build();

// after
MilvusVectorStore.builder(milvusServiceClient, embeddingModel)
    .withDimension(384) // e.g. all-MiniLM-L6-v2
    .build();

Defensive patterns

Strategy: validation

Validate before calling

int dims = embeddingModel.dimensions();
if (dims <= 0) {
    throw new IllegalStateException("Set Milvus withDimension(n) explicitly; model reported " + dims);
}

Try / catch

try {
    milvusVectorStore.afterPropertiesSet();
} catch (Exception e) {
    logger.error("Milvus store init used fallback dimension", e);
}

Prevention

Always pass withDimension(n) in MilvusVectorStore.builder
Ensure embedding provider is reachable before collection creation
Test EmbeddingModel.dimensions() in unit tests with a stub
Drop and recreate the collection if it was created with the fallback dimension

When it happens

Trigger: Creating/initializing the Milvus store (embeddingFieldType or dim call paths) when embeddingModel.dimensions() throws — remote embedding service unavailable, model misconfigured, or returns <= 0.

Common situations: Startup without network access to the embedding provider; using a local model (Ollama, transformers) that lacks a dimensions() implementation; wrong API key configured.

Related errors

AI-assisted analysis of spring-projects/spring-ai@98a7beda4f (2026-09-11). Data as JSON: /api/errors/ade546b2f310fc95. Report an issue: GitHub.

Appendix: source

Thrown at vector-stores/spring-ai-milvus-store/src/main/java/org/springframework/ai/vectorstore/milvus/MilvusVectorStore.java:584

		if (indexStatus.getException() != null) {
			throw new RuntimeException("Failed to create Index", indexStatus.getException());
		}
	}

	int embeddingDimensions() {
		if (this.embeddingDimension != INVALID_EMBEDDING_DIMENSION) {
			return this.embeddingDimension;
		}
		try {
			int embeddingDimensions = this.embeddingModel.dimensions();
			if (embeddingDimensions > 0) {
				return embeddingDimensions;
			}
		}
		catch (Exception e) {
			if (logger.isWarnEnabled()) {
				logger.warn(
						"Failed to obtain the embedding dimensions from the embedding model and fall backs to default: "
								+ this.embeddingDimension,
						e);
			}
		}
		return OPENAI_EMBEDDING_DIMENSION_SIZE;
	}

	// used by the test as well
	void dropCollection() {

		R<RpcStatus> status = this.milvusClient
			.releaseCollection(ReleaseCollectionParam.newBuilder().withCollectionName(this.collectionName).build());

		if (status.getException() != null) {
			throw new RuntimeException("Release collection failed!", status.getException());
		}

View on GitHub (pinned to 98a7beda4f)