Choose a monitoring interval and sensitivity based on how quickly the business needs to detect a sustained failure. More frequent checks improve observation, while requiring several failures can reduce noise at the cost of later notification.
01 Set a useful detection policy
- Define the monitored function and who responds. A customer checkout and an internal report page may justify different detection times.
- Open the monitor's settings and select an interval available on your plan. Choose a timeout appropriate for a healthy response rather than hiding a slow application with a very long value.
- Set sensitivity, which controls how many failed checks are required before an alarm. Consider how brief network interruptions affect your service.
- Select monitoring locations that represent important users. Compare failures across locations before assuming the whole service is unavailable.
- Observe normal operation and review actual incidents. Change one setting at a time and record the reason.
- Test detection on an isolated endpoint, then confirm both failure and recovery notifications reach the assigned responders.
Common questions
Will three failures at one-minute intervals always notify me in exactly three minutes?
No. The outage can begin between checks, and timeouts, confirmation checks and notification processing affect detection. Treat the configuration as a policy to test, not an exact response-time guarantee.
Should I increase sensitivity when alerts are noisy?
First investigate whether failures are real, the target is wrong or a firewall is blocking probes. Increasing the threshold can conceal a persistent configuration problem.
Can frequent checks overload my site?
A lightweight public endpoint is usually easier to monitor than an expensive report or search. Avoid checks that trigger large jobs, writes or third-party charges.