Availability checks

An availability check calls one of your canisters from outside, the way your users do, and records whether it answered and how long it took. The canister needs no code changes, and the check never changes anything.

1. Open Monitor › Availability and choose Add a check. Or select the Availability signal on a canister’s page.

2. Choose what to call: a query that takes no arguments, or a public HTTP GET path for a canister that serves a website.

3. Choose how often, in minutes. The review shows the cost per call and per day.

4. Save the check. Results start arriving at the next call.

Read the results

Each check shows the last six hours as an uptime bar, one cell per five minutes:

• A full cell answered every call.
• A red cell had at least one failed call.
• An empty cell had no call. That’s a gap, never counted as down.

Beside it: availability as a percentage, the p95 round trip, the last call and today’s cost. Open a check for its trend against the previous six hours and its latest results. A failed call shows the error the canister returned.

Good to know

• Checks run on Mainnet only, while Monitor is on.
• Calls are paid from your Agent’s cycles, within a daily limit. The review shows the cost before you save.
• The round trip includes the network’s own processing, not just your code.
• Pause stops calls but keeps the history. Remove asks you to type the canister’s name, and also keeps the history.
• Alert on a check like any other signal: Alert rules.

Next

• Work an alert