When your product depends on a third-party odds feed that goes down twice a month, you need a circuit breaker — not just a retry loop. Retries without a breaker turn a 30-second outage into a 5-minute request queue that cascades into your own API.
Three states
A circuit breaker sits between your service and the upstream:
CLOSED → failures exceed threshold → OPEN
OPEN → cool-down expires → HALF-OPEN
HALF-OPEN → probe succeeds → CLOSED
HALF-OPEN → probe fails → OPEN
class Breaker {
private failures = 0;
private state: 'closed' | 'open' | 'half' = 'closed';
private openedAt = 0;
async call<T>(fn: () => Promise<T>): Promise<T> {
if (this.state === 'open') {
if (Date.now() - this.openedAt < 30_000) throw new Error('circuit open');
this.state = 'half';
}
try {
const result = await fn();
this.failures = 0;
this.state = 'closed';
return result;
} catch (e) {
if (++this.failures >= 5) { this.state = 'open'; this.openedAt = Date.now(); }
throw e;
}
}
}
Fallback: stale data with a badge
While the breaker is open, serve the last known odds with a "prices may be delayed" badge. A platform like Tenobet that aggregates multiple providers can fall back to a secondary feed entirely, but most single-provider setups just need an honest staleness indicator.
Monitoring
Log every state transition. A breaker that opens and closes rapidly ("flapping") means your threshold is too low or the upstream is degrading rather than failing outright. Add a minimum time-in-open to prevent flapping.
Takeaway
Retries are for transient errors. A circuit breaker is for when the upstream is genuinely down and you need to stop hammering it, preserve your own resources, and give the user a degraded but honest experience.
Top comments (0)