Team Deployment

A team deployment is a shared DAGZ instance that multiple developers and CI agents connect to. It stores test history, coverage, and selection state in a central database, so selection works across machines and across runs.

DAGZ ships as a Helm chart and a Docker Compose stack; both run the same components. This section covers configuration that applies to either.

A team deployment adds three things over the local daemon:

• Authentication. Users sign in with SSO. CI agents use client certificates.
• Authorization. Admins grant access by user or by email domain, with roles.
• Shared storage. All clients read and write the same database, so test selection sees results from teammates and CI.

Setup is one-time per deployment. The pages in this section cover each step.

Setup checklist

1. Review the architecture and pick a deployment target.
2. Deploy DAGZ.
   • Install with Helm (Kubernetes)
   • Install with Docker Compose (single VM)
3. Configure an SSO provider so users can sign in.
   • Microsoft Entra ID
   • Google
4. Sign in as the first admin and grant access to your team.
5. Wire the pytest plugin into CI.

How clients connect

Once DAGZ is running and SSO is configured, developers point their local DAGZ at it through .dagz/config.yaml:

project:
  public_url: https://dagz.example.com

CI agents authenticate with a client certificate issued from the deployment. See zb login for the developer-side sign-in flow.