Skip to main content
The deployment check proves that a running instance is wired up correctly: the app answers, Postgres and Redis are healthy, your API key works, a real survey can be answered in a browser and its response read back, file storage round-trips a file, and the JavaScript SDK is served. It runs from a Docker image, talks to your instance over HTTP only, and needs no database access. Run it after the first install, after every upgrade, and after changing a proxy, domain, or storage setting.

Run The Check

1

Create a dedicated workspace

In Formbricks, create a workspace named Deployment check. The check creates and deletes its test surveys there, so keep it separate from real data.
2

Create an API key

Go to Organization settings → API keys and create a key with write access to the Deployment check workspace only. No organization-level access is needed.
3

Run the image

Pass the workspace id when the key has no organization access, which is the recommended setup.
If you serve surveys from a separate domain, also set FORMBRICKS_PUBLIC_URL:
Use latest only when your app runs the latest stable release. Otherwise, use the published check image tag matching your Formbricks version. Confirm that tag exists before scheduling the check.

Configuration

The container must be able to reach your instance. When file storage points at a bucket that only the app can reach, the storage check reports that as a failure, because browsers would not reach it either.

Read The Results

Each check prints one line:
  • ✅ passed. ❌ failed, with the broken component and the next step. ⏭ skipped, because an earlier check it depends on failed. Fix the first ❌ and run again.
  • The exit code is 0 when nothing failed. Skipped checks do not fail the run.
  • report/html/index.html is a browsable report, with a trace and screenshot for failed browser steps. report/junit.xml is for CI systems.
If the instance is unhealthy, the infrastructure checks fail first and everything else is skipped, so you see one root cause instead of a wall of errors.

Common Failures

What It Creates

Each run creates up to two surveys named [deployment-check] ... in the workspace and submits one response to one of them. Everything is deleted at the end of the run, including when a check fails. Surveys older than one hour that carry the prefix are deleted too, so a run that was killed does not leave data behind. The check only ever touches surveys with that prefix. The single test response is stored with its survey and removed with it, so it does not remain in your usage counts.