Error Rate Percentage vs. Absolute Error Count for Worker Alerts

Answer to the Core Questions

Which metric gives better visibility for low‑volume, high‑criticality endpoints? An absolute error count rule. Because such endpoints may only see a handful of requests, a 100 % error rate can still be below a typical percentage threshold (e.g., 5 %). An absolute count fires on every failure, no matter how small the traffic, ensuring that a single catastrophic error is immediately noticed.

How do CPU time‑limit exceptions influence the choice? These exceptions are logged as errors. In high‑traffic periods they can accumulate and overwhelm an absolute‑count rule, while a percentage rule can smooth them out. However, for low‑traffic endpoints the same logic applies: the absolute rule will alert on even a single CPU‑limit exception, whereas a percentage rule may ignore it if the total request volume is very small.

Why Absolute Count Works Better for Low‑Volume Endpoints

• Direct correlation: one error = one alert, no dilution by traffic volume.
• Immediate visibility: critical failures are seen instantly, which is essential for high‑criticality services.
• Predictable behavior: the threshold does not change with load spikes.

Impact of CPU Time‑Limit Exceptions

• High‑traffic bursts can generate many CPU‑limit errors; a percentage rule reduces noise in that scenario.
• During low traffic, a single CPU‑limit error will trigger an absolute‑count alert, ensuring the issue isn’t missed.
• Consider a hybrid approach: use an absolute rule for the endpoint itself and a separate percentage rule for overall traffic spikes.

Practical Steps for This Case

1. Inspect the current alert configuration. In the Cloudflare dashboard, locate the rule that monitors Error Rate for the Worker endpoint.
2. Switch to an absolute threshold. Set the trigger to ≥ 1 error (or a small number if you want to tolerate a single transient failure).
3. Add a separate rule for CPU time‑limit exceptions. Configure a CPU Time Limit Exceeded alert with a percentage threshold (e.g., 10 %) to catch bursts without flooding your channel.
4. Verify in a low‑traffic window. Trigger a controlled error (e.g., return 500) and confirm that the absolute rule fires immediately.
5. Adjust alert routing. For critical endpoints, route the absolute‑count alerts to a dedicated channel or pager; route the percentage rule to a broader monitoring channel.

Missing Diagnostic Detail

To fine‑tune the absolute threshold, it would help to know the typical request rate for the endpoint (requests per minute). If that rate is consistently below 5 req/min, a threshold of ≥ 1 error is appropriate; if it can reach 10 req/min, you might consider ≥ 2 errors to avoid single‑shot noise.

Conclusion

For low‑volume, high‑criticality Cloudflare Worker endpoints, an absolute error count provides the clearest signal. CPU time‑limit exceptions, being error events, are best monitored with an absolute rule for the endpoint and a complementary percentage rule for overall traffic spikes. This dual strategy balances sensitivity with noise suppression and ensures that critical failures are never missed.