An effective alert tells the right person what failed and gives them enough context to act. Too many low-value notifications encourage people to ignore real incidents. Tune checks carefully while preserving visibility into problems that affect customers.
IN THIS ARTICLE
01 Review the alert before changing thresholds
- Open the monitor's event history and record the failed condition, location, start time and duration. Compare the exact URL or port with the service you intended to check.
- Check whether another monitoring location saw the same result. Review public service status and test from another network.
- Compare website results with server metrics and application logs. A timeout, unexpected content, certificate expiry and HTTP error need different responses.
- If the website blocks the monitoring bot, inspect the matching CDN or firewall event. Use the provider's current bot and IP information for a scoped exception only when it is appropriate.
- Review sensitivity and timing. Use supported confirmation, delay or recovery settings to avoid notifying on an isolated blip, while keeping the threshold appropriate to business impact.
- Check recipients and escalation ownership. A notification sent to an abandoned mailbox is not an effective response process.
02 Handle planned maintenance
Use a maintenance or pause facility if provided by your plan, and record when normal checking must resume. Do not delete a useful production monitor simply to silence it. If you publish a status page, describe the affected service and the information you can verify.
03 Investigate certificate alerts early
Check the actual certificate presented by the monitored hostname, including any CDN or reverse proxy in front of the origin. An origin certificate and an edge certificate can have different expiry dates. Confirm renewal and installation, then wait for the monitor to observe the corrected certificate.
What evidence should I collect before closing a monitoring incident?
Test the failed customer journey after a repair, not only the server's power status. Confirm monitoring shows sustained recovery and that alert notifications also close as expected. Keep a short record of the cause, remedy and preventive work.
Monitoring detects conditions; it does not automatically fix the website or establish eligibility for a hosting service credit. For a support review, supply the timestamps, affected service and relevant evidence under the applicable terms.