An uptime alert looks wrong

Most odd uptime alerts come from a setting or a firewall, not from a fault in the check. By the end of this page you will know which case you have. You will also know how to fix it.

Start with the cause in the alert mail or on the Incidents tab. It points to the right section below.

The site loads for you, but the monitor says down#

A failure only counts when a second location confirms it. So two locations saw the problem, even if your own browser did not. How downtime is confirmed walks through the steps.

Common reasons:

• A firewall blocks the checker. Security plugins and bot filters often send a 403 to automated requests. The cause then reads "HTTP 403". Allow requests whose user agent starts with BrokenLinkChecker-Uptime/1.0 in your firewall.
• The page redirects too often. The check follows up to five redirects and expects a 200 at the end. A longer chain ends in a failure.
• The outage was short. A site can fail for a minute and recover before you look. Check the Duration on the Incidents tab.

A keyword monitor went down, but the page looks fine#

The cause reads "keyword check failed". The page loaded, but your keyword was missing from it.

• The match is case-sensitive. "Sale" does not match "sale".
• Someone may have changed the page text. Removing a banner or renaming a heading is enough.
• A redirect to a login page replaces the content the check sees.

Click Edit on the monitor's row and pick a keyword that stays on the page for good. Or clear Keyword (optional) to drop the check.

An alert arrived during planned work#

A maintenance window only silences the monitors ticked on it. Open the Maintenance tab and check three things:

1. The window lists the monitor that alerted.
2. The start and end times are right in UTC, not your local time.
3. The window covered the full stretch of the work.

An incident that was open before the window started stays open. Its alerts resume once the window ends. Maintenance windows covers the details.

The alert came late, or not at all#

• Alert after is set above 1. The alert waits for that many confirmed failures in a row. Each one adds a full check interval. On Free, checks run every 5 minutes, so detection alone can take that long.
• The channel is not ticked on the monitor. A new alert channel reaches no monitor until you tick it under Alert channels. Click Edit on the monitor's row to check.
• An email channel replaced your address. Once a monitor has an email channel ticked, your login address stops getting its mail.
• The monitor is paused. A paused monitor runs no checks. Click Resume.
• The outage fell between two checks. A short outage can end before the next check runs.

The mail says two locations confirmed it, but it is a heartbeat#

Heartbeat and expiry monitors open incidents from a single reading. The mail wording is the same for every type. For a heartbeat, the alert means no ping arrived in time. Check that your job ran and called the ping URL with GET.