Circuit breaker

When a client sends a request to a server, the circuit breaker monitors failure rates to detect persistent issues.

It prevents clients from wasting resources on a failing service and gives the server time to recover.

This pattern protects both client performance and server stability.

---

The circuit breaker counts well-formed requests and failure exceptions.

When the number of exceptions exceeds a threshold, it concludes the failure is not transient.

The client stops sending requests for a defined period.

This pause allows the server to recover from overload.

---

The circuit breaker operates through three states:

• Closed: Allows requests to pass through.
• Open: Blocks all requests and returns an exception.
• Half open: Sends one trial request to test service availability.

Initially, the circuit breaker is in the closed state.

When failure count exceeds a threshold, it transitions to open.

After a configured duration, it moves to half open.

If the trial request fails, it returns to open.

If the trial request succeeds, it transitions to closed.

If another exception occurs (not the monitored one), it stays in half open and retries once.

---

To set up a circuit breaker, specify:

• The exception type to monitor.
• The number of consecutive failures before opening.
• The duration the circuit remains open.

Example: Five consecutive failures trigger opening; one-minute open duration.

---

Use established open-source libraries like Resilience4j or Poly.

Implement the circuit breaker with a thread-safe instance.

Multiple threads must share the same instance to measure overall failure rates.

Use locking during state reads and recalculations.

---

When a request is rejected by the open circuit breaker, one of these actions is taken:

• Buffer
• Failover
• Callback
• Apply back pressure up the stack
• Cancel

---