Steps
- On Alerts, add an address. Your own address is offered during set-up. A client’s address works too: its owner confirms by e-mail first.
- Open the confirmation link in that inbox. Until then the address shows as waiting for confirmation and receives nothing. Send the link again sends a new one.
- Press Send a test message and look for a message headed
[TEST] Monitoring test notification. The page shows when the last test was sent. - Optionally add Slack or a webhook, or Teams, Discord, Telegram, Google Chat, PagerDuty, Opsgenie or ntfy, as well. Each one is confirmed by a test message.
What an alert says
An incident alert lists what each region saw, the result and the time, and links to the incident page. The incident page shows the full evidence of each check: for a web page the status code, the redirects, the certificate, the DNS answer and the timings (what is on it). An alert is sent when two regions agree that a site is down (why two regions), and another when the site has recovered. The same channels carry warnings about expiring certificates and domain registrations, search-visibility risks and slowdown notices.
When nothing arrives
- Check the spam folder, and check that the address is confirmed.
- Send a test message. If the test arrives, alerts are working and no incident has been confirmed.
- One region seeing an error is not an alert on purpose (why).
- Your plan limits the number of alert channels; the Billing page shows how many are in use.
- An address you remove stops receiving alerts at once.
- Planned work holds alerts back on purpose: during a maintenance window nothing is sent for failures inside it (how).