Protect downstream APIs with a circuit breaker using Data Tables

Quick overview

Stop n8n workflows from repeatedly calling an unavailable API by opening a circuit after consecutive failures, enforcing a cooldown, and allowing a controlled half-open recovery trial.

How it works

1. Receives a POST request on the /webhook/circuit-breaker-check endpoint with a service_key to decide whether a downstream API call should proceed.
2. Looks up the service’s circuit state in the service_circuits n8n Data Table and evaluates whether to allow, block, or start a half-open trial based on the current state and cooldown window.
3. Creates or updates the circuit record when entering half-open, then responds with allow_request, the current state, and retry_after_seconds for the caller.
4. Receives a POST request on the /webhook/circuit-breaker-result endpoint with the service_key plus success and optional error to record the outcome of an allowed call.
5. Updates the service_circuits record to close the circuit on success or increment failures and open the circuit (with open_until) when the failure threshold is reached or a half-open trial fails, then returns a confirmation response.

Setup

1. Create an n8n Data Table named service_circuits with the columns listed in the template (including service_key, state, failure_count, failure_threshold, and the timestamp fields) and select it in each Data Table step.
2. Copy the two webhook URLs from n8n and configure your calling service to POST to /webhook/circuit-breaker-check before an external API call and to /webhook/circuit-breaker-result after each allowed call.
3. Update the circuit breaker defaults in the configuration step (for example failure_threshold and cooldown_minutes) to match your reliability requirements.

Requirements

• Data Tables enabled
• No credentials or paid services are required.

Customization

• Use a separate service_key for each provider or endpoint.
• Adjust thresholds for critical or rate-limited services.
• Route blocked requests to a queue, fallback response, or retry workflow.
• Replace Data Tables with PostgreSQL for strict concurrency guarantees.