Course contentsShow
← System Design
Lesson 1056 of 1,91927. Circuit Breakers and BulkheadsFree lesson

Circuit Breaker State Machine

Understanding the three states (Closed, Open, Half-Open) and state transitions based on failure thresholds and timeouts.

Circuit Breaker State Machine

You'll learn: How a circuit breaker's state machine orchestrates transitions between Closed, Open, and Half-Open states based on real-time failure metrics.

The State Machine Core

A circuit breaker is fundamentally a state machine that tracks system health and makes decisions about allowing or blocking requests. It continuously monitors three key inputs:

  1. Failure count (or rate) in the current window
  2. Success count during recovery attempts
  3. Time elapsed since state changes

The state machine uses these inputs to determine when to transition between states.

State Transition Logic

From Closed → Open:
When failures exceed the configured threshold (e.g., 50% error rate over 10 seconds), the circuit trips to Open. This is your "break the circuit" moment.

From Open → Half-Open:
After a cooldown timeout expires (e.g., 30 seconds), the circuit automatically transitions to Half-Open to test if the downstream service has recovered. Time-based transition — no manual intervention needed.

From Half-Open → Closed:
If a configured number of trial requests succeed (e.g., 3 consecutive successes), the circuit resets to Closed, resuming normal operation.

From Half-Open → Open:
If any trial request fails during the Half-Open phase, the circuit immediately trips back to Open, restarting the timeout period.

The Control Loop

Think of it as a feedback control system: the state machine continuously samples failure metrics, compares them against thresholds, and adjusts the circuit state to protect your system. This creates a self-healing mechanism that automatically detects problems, protects services, and attempts recovery without human intervention.

Key Takeaway: The circuit breaker state machine is a deterministic control flow that uses failure thresholds and timeouts to automatically transition between Closed, Open, and Half-Open states, enabling self-healing behavior in distributed systems.