ProbeRun

Docs / Using ProbeRun

Running your own server

One command installs the server, its batch worker and the dashboard on a machine with Docker:

The installer is published with ProbeRun's open-source release.

  • What it installs. Five containers under the Docker Compose project proberun-selfhost: Postgres, a one-shot database migration, the API, the worker that runs triggered batches, and the dashboard. Nothing else on the machine changes.
  • Private by default. The dashboard (port 3000) and the API (port 8000, for probe push) listen on 127.0.0.1 only; the database and worker are not published at all. Anyone who can open the dashboard acts as its Admin, which on loopback means you.
  • Sharing it with a team. Set PROBE_BIND to a network address and configure sign-in with your OIDC provider (Keycloak, Entra ID, Google): the installer refuses to expose the server without it, and on a shared machine use OIDC even on loopback. Put your own TLS proxy in front.
  • Your API key. The installer prints the server’s API key when it finishes; probe push sends it as PROBE_API_KEY. It lives, with every other generated secret, in the install directory’s .env (readable only by you).
  • Where data lives. In the Docker volume proberun-selfhost_postgres-data. Re-running the installer keeps every secret and every Project and Run.
  • First look. Open an empty Project and press Load an example comparison for sample data (see “Your first Project”).
  • Stop and remove. docker compose -p proberun-selfhost stop stops it; docker compose -p proberun-selfhost down removes the containers and keeps the data; down -v deletes the data too.

Settings, upgrades and what each installer exit code means are in the operator guide (docs/SETUP.md, “Self-hosted install”).