consecutive failures
Trips after a specific number of failures occur sequentially.
How it works
Monitors sequential failure counts, ignoring time windows to focus on the string of events. This is the fastest method for detecting a hard service outage.
When to use
- Critical external API dependencies.
- Database connection health checks.
- Logic where a single success should reset the failure counter.
When not to use
- Noisy environments where intermittent expected errors are frequent.
State explanations
closed
No sustained failure streak detected.
half open
Waiting for a successful request to confirm recovery.
open
The service has failed sequentially enough to indicate an outage.