When the service receives sustained traffic above ~500 req/s, the database connection pool exhausts and new requests start timing out after waiting for an available connection.
Steps to reproduce:
- Run load test at 600 req/s for 2+ minutes
- Observe connection wait times increasing in metrics
- Requests begin returning 503 after ~90s
Expected: pool should queue and serve requests within acceptable latency.
Actual: timeouts cascade and recovery requires a restart.
Possible fix: increase pool ceiling or add queue with backpressure.
When the service receives sustained traffic above ~500 req/s, the database connection pool exhausts and new requests start timing out after waiting for an available connection.
Steps to reproduce:
Expected: pool should queue and serve requests within acceptable latency.
Actual: timeouts cascade and recovery requires a restart.
Possible fix: increase pool ceiling or add queue with backpressure.