Incidents

An incident is one problem, tracked from the check that found it until a check no longer finds it. Incidents are what send alerts, so knowing how they open and close tells you when you’ll hear from SEOComet.

When an incident opens

When a check finds an issue, SEOComet opens an incident for it. Every issue opens one, info included. Your alert level then decides whether that incident sends an alert.

• A check has at most one open incident per kind of issue. When the next check finds the same issue, it’s added to the incident that’s already open, not reported as new.
• A suggestion never opens an incident. Nothing is broken, so it only shows on the check.
• “Couldn’t check” opens an incident too, usually as a warning. Uptime checks are different: after enough failures in a row they open Site down, which is critical. While a check can’t fetch its address, its other open incidents stay open.
• A robots.txt or ads.txt change found on the same check as a warning or critical issue doesn’t open its own incident. It’s added to that issue’s incident as evidence, so you see the change and the problem together.

Open, muted and resolved

• Open: the problem is still there.
• Muted: still open, but quiet. See Mute.
• Resolved: a check no longer found the problem, someone marked it resolved, or its check or site was paused.

Escalation

If a later check finds the same issue at a higher severity, the incident takes the new severity. For example, a certificate warning becomes critical as the expiry date gets closer. Your channels get a Got worse alert, if your alert level covers the new severity.

An incident’s severity never goes down while it’s open. If the problem gets less serious, the incident keeps its highest severity until it’s resolved.

Reminders

Reminders are off until you turn them on. In Alerts › What, tick Send a reminder every 4 hours while an incident is open. The owner and people who can edit all sites can change it.

With reminders on, each open incident sends a Still open alert every 4 hours, counted from when the problem was first seen, if your alert level covers it. SEOComet looks for due reminders once an hour, so one can arrive up to an hour late. Muted incidents don’t remind.

A reminder only goes to channels that got an alert about the incident. If its alert was held back because the site was down or had sent too many, it doesn’t remind there either. See Was the site down?.

Mute

Mute on the incident page quiets one incident for an hour, for 24 hours or until it’s fixed.

While an incident is muted, no reminders or escalation alerts go out for it. You still get the resolved alert when it’s fixed.

Muting doesn’t close anything. The incident stays open, and when a timed mute ends it’s an ordinary open incident again. Mute again to pick a new length, or Unmute to end it early.

Mark resolved

Mark resolved closes the incident now and sends the resolved alert. Use it when you’ve fixed the problem and don’t want to wait for the next check.

If the problem is still there, the next check that finds it opens a new incident, which alerts again.

The owner, people who can edit all sites and people who can edit that site can mute an incident or mark it resolved. People who can only view can’t.

Evidence

Evidence on the incident page shows the latest check that found the problem: the message, when it ran, the result, its HTTP status and response time, and the final address after redirects. For a robots.txt or ads.txt change, it shows the lines that changed. Links take you to the check and, for those files, to Compare versions.

Below it, Timeline lists everything that happened to the incident, from when it opened to how it closed.

Alerts sent

Alerts sent lists the alerts this incident sent to your channels, newest first, and whether each one went out. It’s shown to the owner and people with access to all sites.

When it closes on its own

• A check no longer finds the issue. The incident is resolved and a resolved alert goes out. It’s judged by the highest severity the incident reached, so it goes where the alert about the problem went. A check that couldn’t fetch its address doesn’t count: it can’t tell whether the problem is gone.
• The issue was a change. Some issues compare one check with the one before it: robots.txt or ads.txt changed, a Sitemap line removed from robots.txt, fewer URLs in a sitemap, a changed redirect, title, description or H1. The next check can’t find the same change again, so these close on the next check that gets an answer.
• The check or site is paused. Pausing a check or a site resolves its open incidents, and no alerts go out for them. The same happens when a plan change pauses a check.