mem0ai/mem0 · error · Error

Unsupported distance metric: ${config.distanceMetric}

Error message

Unsupported distance metric: ${config.distanceMetric}

What it means

The distance metric is stored in the vector index DDL and used to convert distance to a similarity score, so it must be one of the six supported Oracle VECTOR_DISTANCE functions: COSINE, EUCLIDEAN, EUCLIDEAN_SQUARED, DOT, HAMMING, MANHATTAN. The value is uppercased before the check, so 'cosine' is accepted; anything else fails, echoing the original config value in the message.

Source

Thrown at mem0-ts/src/oss/src/vector_stores/oracledb.ts:351

    this.collectionName = quoteIdentifier(config.collectionName || "mem0");
    this.indexName = quoteIdentifier(
      config.indexName || `${config.collectionName || "mem0"}_VEC_IDX`,
    );

    this.embeddingModelDims = config.embeddingModelDims ?? 1536;
    if (
      !Number.isInteger(this.embeddingModelDims) ||
      this.embeddingModelDims <= 0
    ) {
      throw new Error("`embeddingModelDims` must be a positive integer");
    }

    const distanceMetric = (config.distanceMetric ??
      "COSINE") as string as DistanceMetric;
    this.distanceMetric = distanceMetric.toUpperCase() as DistanceMetric;
    if (!DISTANCE_METRICS.includes(this.distanceMetric)) {
      throw new Error(`Unsupported distance metric: ${config.distanceMetric}`);
    }

    const indexType = (config.indexType ?? "HNSW") as string;
    this.indexType = indexType.toUpperCase() as IndexType;
    if (this.indexType !== "HNSW" && this.indexType !== "IVF") {
      throw new Error(`Unsupported index type: ${config.indexType}`);
    }

    this.indexAccuracy = config.indexAccuracy;
    if (
      this.indexAccuracy !== undefined &&
      (!Number.isInteger(this.indexAccuracy) ||
        this.indexAccuracy <= 0 ||
        this.indexAccuracy > 100)
    ) {
      throw new Error("`indexAccuracy` must be an integer between 1 and 100");
    }

View on GitHub (pinned to 001c235229)

Solutions

1. Use one of: COSINE, EUCLIDEAN, EUCLIDEAN_SQUARED, DOT, HAMMING, MANHATTAN (case-insensitive).
2. Map other backends' names: Qdrant 'Dot' → DOT, 'Euclid' → EUCLIDEAN; pgvector 'l2' → EUCLIDEAN, '<#>' → DOT.
3. Omit distanceMetric to get the COSINE default, which matches most normalized-embedding setups.
4. If porting a config file, grep it for old metric names before switching stores.

Example fix

// before
new OracleDB({ connectionParams, distanceMetric: 'l2' });

// after
new OracleDB({ connectionParams, distanceMetric: 'EUCLIDEAN' });

const ORACLE_METRICS = new Set(['COSINE','EUCLIDEAN','EUCLIDEAN_SQUARED','DOT','HAMMING','MANHATTAN']);
function normalizeMetric(m?: string): string | undefined {
  if (!m) return undefined;
  const upper = m.toUpperCase();
  if (!ORACLE_METRICS.has(upper)) throw new RangeError(`Unsupported distance metric '${m}'. Valid: ${[...ORACLE_METRICS].join(', ')}`);
  return upper;
}

const isOracleMetric = (v: string): v is 'COSINE'|'EUCLIDEAN'|'EUCLIDEAN_SQUARED'|'DOT'|'HAMMING'|'MANHATTAN' => ORACLE_METRICS.has(v.toUpperCase());

When it happens

Trigger: distanceMetric: 'euclidean' works (uppercased) but 'l2' fails; 'cosine-similarity', 'IP' (a Qdrant name), 'l2_sq' (pgvector name), or undefined-typo metrics fail. Note the message prints the raw config value, e.g. "Unsupported distance metric: l2".

Common situations: Porting configs from other vector stores: Qdrant uses 'Cosine'/'Dot'/'Euclid', pgvector uses '<->' or 'l2'; misspellings like 'COSIN'; binary-embedding setups where 'hamming' is correct but users type 'hamming-distance'.

Related errors

• Must provide at least one of `connectionParams` and `client`
• `embeddingModelDims` must be a positive integer
• Unsupported index type: ${config.indexType}
• `indexAccuracy` must be an integer between 1 and 100
• Unsupported ${this.indexType} index parameter '${key}'. Allo

AI-assisted analysis of mem0ai/mem0@001c235229 (2026-08-15). Data as JSON: /api/errors/b4d07a6ac703a716. Report an issue: GitHub.