See what happened just before a request began failing
When a saved request turns red, its notice says what happened just before - a deploy, a new API version or an edit - and offers the one next step.
In short
When a saved request that passed starts failing, {{brand}} looks once at what happened just before. That can be a deploy, a new API version or an edit of the request. Then {{brand}} adds one line, such as "Began failing 12 minutes after the deploy of 1.4.2." It offers one next step: open the change, or create a bug.
Before you start
- Who: anyone who can open the failed request sees the line. "Create bug" needs "Change cards" on a board connected to the project.
- Plan: every plan.
- First: the request must have passed in that environment before it failed.
Steps
- Open the failure notice in your inbox, the item under Needs you on Home, the run's page or a monitor's health sheet.
- Read the line about what happened just before, and the sentence about what differs from the last pass.
- Press "Open the change" when the event has a page, or "Create bug" on the failing run.
Result: you go straight to the change, or file a bug with the run as evidence.
Good to know
- Three things count. One is a deploy reported to the environment where it failed. Another is a new version of the API whose changes name the request's operation. The last is a saved edit of the request itself.
- It counts only if it happened after the request last passed in that environment, no later than the first failure, and within 24 hours. The nearest one wins.
- The line says how long after, never a clock time, and never claims a cause. The second sentence names one difference. It is what the new version changed, else the status ("It answered 200 at the last pass and 500 now."), else that the shape of the response changed. It never repeats a value a response held.
- Requests that turned red after the same event in the same environment share one notice - forty requests after one deploy are one unread row.
- On a phone, the notification offers "Open the change" beside "Open".
If something goes wrong
- There is no line: the request never passed in that environment, or nothing happened within 24 hours before. The notice reads as before.
- A whole project run, and a run the CLI reports, are not looked at this way. A monitor's alert emails and webhooks are unchanged.
Related
- /help/ai/why-did-this-fail
- /help/api-testing/compare-runs
- /help/api-testing/tests-after-a-deploy