Mintplex-Labs/anything-llm · error · Error

Could not embed document chunks! This document will not be r

Error message

Could not embed document chunks! This document will not be recorded.

What it means

In AstraDB's addDocumentToNamespace, after embedding the document the code builds vector records from vectorValues; when the embedder returned no vectors at all (vectorValues empty/absent), the else branch throws this error and the document is skipped from the vector store. It guards against writing a document with zero embeddings.

Source

Thrown at server/utils/vectorDbProviders/astra/index.js:246

      this.logger("Snippets created from document:", textChunks.length);
      const documentVectors = [];
      const vectors = [];
      const vectorValues = await EmbedderEngine.embedChunks(textChunks);

      if (!!vectorValues && vectorValues.length > 0) {
        for (const [i, vector] of vectorValues.entries()) {
          if (!vectorDimension) vectorDimension = vector.length;
          const vectorRecord = {
            _id: uuidv4(),
            $vector: vector,
            metadata: { ...metadata, text: textChunks[i] },
          };

          vectors.push(vectorRecord);
          documentVectors.push({ docId, vectorId: vectorRecord._id });
        }
      } else {
        throw new Error(
          "Could not embed document chunks! This document will not be recorded."
        );
      }
      const { client } = await this.connect();
      const collection = await this.getOrCreateCollection(
        client,
        namespace,
        vectorDimension
      );
      if (!(await this.isRealCollection(collection)))
        throw new Error("Failed to create new AstraDB collection!", {
          namespace,
        });

      if (vectors.length > 0) {
        const chunks = [];

        this.logger("Inserting vectorized chunks into Astra DB.");

View on GitHub (pinned to 3aec848f28)

Solutions

Check the embedding engine logs/credentials first - the embedder silently returning [] is the usual root cause.
Verify the document actually contains extractable text (re-upload as .txt/.md to test).
Test the embedder directly with a one-line input (e.g. curl the embeddings endpoint) to confirm non-empty vectors.
Once the embedder returns vectors, delete and re-embed the failed document.

Defensive patterns

Strategy: validation

Validate before calling

const embedded = await LLMConnector.embedChunks(textChunks);
if (!Array.isArray(embedded) || embedded.length === 0 || embedded.some((v) => !v?.length)) {
  throw new Error('Embedder returned no/empty vectors - aborting before vector DB write.');
}

Try / catch

try {
  await vectorDb.addDocumentToNamespace(/* ... */);
} catch (e) {
  if (/Could not embed document chunks/i.test(e.message)) {
    // embedder-side issue: surface to user, do not retry unchanged
    return { ok: false, reason: 'embedding-engine' };
  }
  throw e;
}

Prevention

Smoke-test the embedder at deploy time with a known input and assert vector length.
Alert on zero-vector batches in embedder wrappers instead of passing them downstream.
Validate documents have extracted text before embedding.

When it happens

Trigger: Embedding a document whose textChunks/vectorValues came back empty: embedder endpoint down but returning 200 with empty data, document text extraction produced nothing, or a workspace embed with zero usable chunks.

Common situations: Embedding engine credentials revoked or quota exhausted so batches return empty results; empty/ scanned PDFs; LM Studio/Ollama model unloaded returning no embeddings; mismatch between chosen embedder and the model it serves.

Related errors

AI-assisted analysis of Mintplex-Labs/anything-llm@3aec848f28 (2026-08-18). Data as JSON: /api/errors/7228d162a116e7cd. Report an issue: GitHub.