My flow monitor fails or can't run
Tell a broken sign-in apart from a flow that needs fixing, and get the monitor passing again.
Last updated
A flow monitor shows one of two problem states. Failing means a test ran and an expectation didn't hold, which alerts. Cannot run means Uptime couldn't finish the run, which never alerts. Open the monitor's page and click Run now to see the current result.
If a step's expectation fails
The run is Failing and Last failure names the test and the line it stopped on. Usually your service really changed: the sign-in rejected the account, an endpoint answered 401, a field moved in the JSON.
Read the failure detail against that line of the flow. If the service is right and the flow is out of date, open Edit, update the step or the expected value, click Save changes, then Run now.
If the test account stopped working
A flow that signs in depends on its account. A changed or expired password, a locked account or a new second-factor prompt makes the sign-in step fail on every run.
Sign in as that account yourself, fix it, and update the password in the flow with Edit.
If the flow reaches a host no verified domain covers
Every address in the flow has to sit under one of your team's verified domains, checked before every run. A redirect to another domain, or a domain your team has since removed, stops the run, and it shows Cannot run. A host outside your verified domains is refused when you save the flow, with a message naming it.
Ask a team owner or admin to verify the domain under Settings → Domains, or change the flow to use a host you've verified. See Team Domains.
If the run hit a limit
One run may take 30 seconds and make 20 requests, and each response is read up to 1 MB. A run past any of these shows Cannot run, with a message such as "This flow made too many requests".
Split a long flow into two monitors, and point steps at the endpoint or page you assert on rather than at a large export.
If the flow doesn't parse
A typo, a phase out of order (given, then when, then then) or a URL built at run time instead of written out stops the flow before it starts. The message gives the reason.
Fix the source with Edit. Every URL must be written in full in the flow. See Writing a Flow.
Why didn't I get an alert when it showed Cannot run?
Cannot run means the run didn't find out anything about your service, so it doesn't page anyone. Check the monitor's page after you change a flow, and use Run now to try it straight away.