How we call a service down

Most “is it down” sites only count people who bothered to complain. That is useful, and it is also late, easy to spam, and blind when a failure is quiet. This desk keeps the crowd and adds two other signals.

User reports

Anyone can report a service without an account. We compare the last 15 and 60 minutes with the typical volume for that UTC hour over the past week. A small service can spike on a handful of reports. A busy one has to move well past its own baseline. Ordinary noise stays quiet.

Active checks

A cron trigger on Cloudflare fetches each service about every five minutes. Two failures in a row count as down. One failure is only degraded, so a single blip does not page the internet. Timeouts and HTTP 5xx fail the check. HTTP 403 and 429 still count as reachable, because a lot of sites refuse bots while serving people. Latency is degraded when recent successful checks are several times slower than the last day and slower than 800 ms.

Official status

When a vendor publishes a Statuspage or Slack-style JSON feed, we read the indicator. A major or critical incident is a major outage here even if the marketing homepage still returns 200 — the summary says so when our own check still succeeds. A missing feed is not treated as an incident.

How the three combine

Owner alerts

A person who controls the domain can claim the page, prove it with a DNS TXT record or a homepage meta tag, and subscribe. Stripe Checkout takes the payment. Resend sends the mail. Alerts require an active subscription with no dispute, fraud, or full-refund hold.

Damn is it down is operated by Big Toy Market LLC and is not affiliated with the services it tracks.