Docker Deployment

Deploy Arbiter by first preparing a local Docker staging directory, then installing that checked staging directory as a Linux system service.

Reploy

Reploy is the deployment tool used by Arbiter's Docker install path. It creates the staging directory, manages generated Docker files, runs deployment-scoped app commands, and installs a checked staging directory as a Linux service.

Install Reploy on the operator machine before starting:

curl -fsSL https://reploy.yadan.net/install.sh | sh

If you prefer to install through Python packaging, Reploy is also available on PyPI:

python3 -m pip install reploy

The staging directory is ordinary user-owned state: it contains the selected package bundle, generated Docker files, config, env placeholders, and a local helper script. You can start it locally, test the Arbiter server, and iterate before anything is installed under /opt.

When staging works, install copies that directory into place and rewrites it from a staging identity to the installed service identity.

Deployment Flow

1. Prepare Docker Deployment: stage and test an Arbiter instance as an unprivileged operator.
   • Initialize a Docker staging directory.
   • Prepare the installation bundle by selecting Arbiter server and service plugin packages.
   • Bootstrap configuration for plugin accounts and policies.
   • Bring up the staged Docker instance.
   • Test manually or with an agent against the staged Arbiter server.
2. Linux Install: install the checked staging directory as a system service with sudo.

References

• Bundle deep-dive: package roots, wheelhouse behavior, upgrades, and maintainer bundle refreshes.
• Reploy Command Reference: generated helper commands for inspection, staging, and bundle maintenance.
• Security Model: deployment trust boundaries and host access assumptions.