Uptime monitoring
Six kinds of check, run from the regions your customers are actually in, with the confirmation logic to keep one flaky network path from paging the whole team.
Pick the check that matches the failure
HTTP(S)
Status code, redirects followed, and a keyword matched in the response body.
Keyword
Fails when the expected text is missing, so a 200 rendering an error page still counts as down.
TCP port
Confirms a service is accepting connections — databases, mail, queues, anything not speaking HTTP.
Ping
Plain ICMP reachability, where an application-layer check does not apply.
DNS
Watches a record resolve to the value you expect, catching a bad zone edit or a lapsed registration.
Heartbeat
Your cron job or worker pings ServerPal on a schedule; a missed window is the alert.
One region disagreeing is not an outage
Checks run from several places at once and you decide how many have to agree before anything fires, so a single bad route never wakes anyone at three in the morning.
- Choose which regions run each check
- Set the number that must confirm before alerting
- Intervals as fast as every 30 seconds
- SSL certificate expiry on its own warning schedule
Watch it get slower before it stops
Every check records its full timing, not a sample. Set a response-time threshold per check and a slow stretch shows as degraded — on your dashboard and, if you choose, on your public status page.
api.acme.com · 60 days
Start monitoring in about a minute.
Twenty checks and one server, free and with no card. Upgrade only when you outgrow it.