n8n-io/n8n · error · NodeOperationError

${error.message}

Error message

${error.message}

What it means

Catch-all in the Google Gemini router's per-item execution loop. When the dispatched operation (execute.call(this, i)) throws and the node is NOT in continueOnFail mode, the original error is re-wrapped as a NodeOperationError carrying the itemIndex and the original description. The message is the wrapped error's own message.

Source

Thrown at packages/@n8n/nodes-langchain/nodes/vendors/GoogleGemini/actions/router.ts:64

			break;
		default:
			throw new NodeOperationError(
				this.getNode(),
				`The operation "${operation}" is not supported!`,
			);
	}

	for (let i = 0; i < items.length; i++) {
		try {
			const responseData = await execute.call(this, i);
			returnData.push(...responseData);
		} catch (error) {
			if (this.continueOnFail()) {
				returnData.push({ json: { error: error.message }, pairedItem: { item: i } });
				continue;
			}

			throw new NodeOperationError(this.getNode(), error, {
				itemIndex: i,
				description: error.description,
			});
		}
	}

	return [returnData];
}

View on GitHub (pinned to 5ac6606e81)

Solutions

  1. Read the wrapped error's message/description — it identifies the real underlying failure (model not supported, upload failed, non-empty prompt, etc.) — and fix that root cause using its own guidance.
  2. If you want the loop to keep processing the remaining items, enable the node's 'Continue On Fail' setting (or set the workflow execution to continue on fail); failures are then pushed as { json: { error } } instead of thrown.
  3. For batch robustness, split the input so one bad item doesn't abort the whole run, or pre-validate each item upstream.

Example fix

// before — one bad item aborts the whole loop
// (continueOnFail off)

// after — toggle on the node so failures are captured per item
// In the node settings: Settings → 'Continue On Fail' = true
Defensive patterns

Strategy: try-catch

Try / catch

try {
  const out = await geminiNode.execute();
} catch (e) {
  // NodeOperationError carries itemIndex on the failing item
  const itemIndex = (e as NodeOperationError).itemIndex;
  if (itemIndex != null) {
    // isolate the bad item, continue with the rest
  }
  throw e;
}

Prevention

When it happens

Trigger: Any inner operation throws (model validation, API HTTP error, upload failure, malformed response) on item i, and the workflow/execution settings do not have 'Continue On Fail' enabled. The original error type and description are preserved on the wrapped error.

Common situations: One item in a batch has a bad prompt / model / binary while others are fine; the upstream API rate-limits or returns 4xx for one request; a credential is expired so every item fails on the first one.

Related errors

AI-assisted analysis of n8n-io/n8n@5ac6606e81 (2026-08-12). Data as JSON: /api/errors/a933778672b9ba0f. Report an issue: GitHub.