Publishing your project

Operations

Error tracking, logs, server status, and releases in the day-to-day operation of your project.

On this page

After the first deployment, day-to-day work begins: spotting errors, keeping an eye on servers, and tracing releases. This page shows which tools are already set up for that in every project.

Finding errors: Sentry

Every generated project is connected to Sentry from the first commit:

  • The Sentry DSN is already in env/shared.generated.env (app) or in the deployment configuration (backend) – nothing to set up.
  • Errors from app and backend automatically appear in your project’s Sentry project, including the release version.
  • You’ll find the link to your Sentry project in the platform’s project overview.

More on the concept: Error tracking.

Tip
Check Sentry briefly after every prod release: new errors usually show up there minutes after the deployment – long before users report them.

Keeping an eye on servers

You can see the state of your servers (reachability, running services) in the platform’s server section. How to create and connect servers is covered in Servers.

Logs of the deployed applications live on the server in the Docker containers – if you run into problems, support can help you with access.

Tracing releases

Every merge into main produces a new, automatically computed version:

  • Which version runs where is recorded in the gitops-configuration repository in configurations/dev/versions.yaml and configurations/prod/versions.yaml.
  • The corresponding builds and jobs are in GitLab under CI/CD → Pipelines.
  • Errors in Sentry are linked to the respective release version – so you can see immediately from which release a problem appears.

The complete flow from commit to deployment is described in Git workflow & deployment.

When something goes wrong

  1. Check Sentry – are there new errors since the last release?
  2. Check the pipeline – did the deployment complete in GitLab (code repository and gitops-configuration)?
  3. Roll back to the last working state: re-run publish-docker-prod in the pipeline of the last good main state – the previous version is live again.
  4. If you’re stuck, contact support.

See also