mem0ai/mem0 · error · Error

Failed to search vectors: ${error instanceof Error ? error.m

Error message

Failed to search vectors: ${error instanceof Error ? error.message : String(error)}

What it means

Catch-all around Cloudflare Vectorize similarity search: any failure while querying the index (auth, missing index, network, malformed request) is logged and rethrown wrapped as 'Failed to search vectors: <message>'. The inner message carries the actual Cloudflare error text.

Source

Thrown at mem0-ts/src/oss/src/vector_stores/vectorize.ts:121

        {
          account_id: this.accountId,
          vector: query,
          filter: filters,
          returnMetadata: "all",
          topK: topK,
        },
      );

      return (
        (result?.matches?.map((match) => ({
          id: match.id,
          payload: match.metadata,
          score: match.score,
        })) as VectorStoreResult[]) || []
      ); // Return empty array if result or matches is null/undefined
    } catch (error) {
      console.error("Error searching vectors:", error);
      throw new Error(
        `Failed to search vectors: ${error instanceof Error ? error.message : String(error)}`,
      );
    }
  }

  async get(vectorId: string): Promise<VectorStoreResult | null> {
    await this.initialize();
    try {
      const result = (await this.client?.vectorize.indexes.getByIds(
        this.indexName,
        {
          account_id: this.accountId,
          ids: [vectorId],
        },
      )) as any;

      if (!result?.length) return null;

View on GitHub (pinned to 001c235229)

Solutions

1. Read the inner message and map it: 404 -> indexName/accountId wrong; 401 -> token; 400 dimension -> align embedding model with index dims.
2. Verify the index exists: wrangler vectorize list, and compare with config indexName.
3. For transient failures, retry the search — reads are idempotent.

if (query.length !== indexDims) throw new Error(`Query vector is ${query.length} dims; index expects ${indexDims}`);

try { return await memory.search(q, opts); } catch (e) { const m = e instanceof Error ? e.message : ''; if (m.startsWith('Failed to search vectors:')) { if (isTransientInner(m)) return retryWithBackoff(() => memory.search(q, opts), 3); throw new Error(m); } throw e; }

When it happens

Trigger: memory.search() against a deleted or renamed index; expired API token; account_id mismatch; dimension of the query vector differing from the index dims (Cloudflare rejects the query); network failure from the runtime.

Common situations: Index recreated after a config change while old code still references the prior indexName; token permissions narrowed to write-only; embeddings model changed without recreating the index.

Related errors

• Failed to insert vectors: ${error instanceof Error ? error.m
• Failed to update vector: ${error instanceof Error ? error.me
• Failed to insert vectors: ${response.status} ${errorText}
• Failed to get vector: ${error instanceof Error ? error.messa
• Failed to update vector: ${response.status} ${errorText}

AI-assisted analysis of mem0ai/mem0@001c235229 (2026-08-15). Data as JSON: /api/errors/f67c5fcade97bd45. Report an issue: GitHub.