resilience

Circuit Breaker

Prevent cascading failures — CLOSED → OPEN → HALF_OPEN state machine

StateCLOSED
Consecutive Failures0 / 3
CLOSEDrequests flowOPENfail fastHALF_OPENprobe 1 requestfailures ≥ 3timeoutprobe succeeds → resetprobe failssuccess → reset count
Failure Counter
·
·
·
0/3 to trip

// event log

No events yet. Click "Simulate Success" or "Simulate Failure" to begin.

// how it works

  • CLOSED: Requests flow normally. Track consecutive failures.
  • If failures ≥ threshold → switch to OPEN
  • OPEN: All requests fail fast immediately. Start a timer.
  • After timeout → switch to HALF_OPEN
  • HALF_OPEN: Allow ONE probe request through.
  • Probe succeeds → CLOSED (reset). Probe fails → back to OPEN.

// trade-offs

  • Prevents cascading failures across services
  • Fast failure response — no waiting for timeouts
  • Automatic recovery via HALF_OPEN probing
  • Needs careful threshold tuning
  • Can mask intermittent errors
  • Single probe in HALF_OPEN may not be representative