[silug]

Adds a control_branch parameter to the basic_aio plan, defaulting to production, and a matching CONTROL_BRANCH variable to the Vagrantfile. This lets an environment run a branch other than production, such as a PR branch in CI. That's the next step, in a separate PR.

With control_repo, the plan:

• Still deploys every branch with r10k, as before. It then fails if the chosen branch's environment wasn't deployed, before touching any agent.
• Runs puppet config set environment <branch> --section agent on every target before the first agent run, so both the first runs and the puppet service use that environment.

The branch has to be a valid environment name, meaning letters, digits, and underscores. Both the plan parameter and the Vagrantfile check this. r10k would otherwise skip a branch like my-feature or deploy it under a different name. The README notes that the control repo still needs a production branch, because the server fails to process reports without that environment. #210's CI turned this up. basic_aio::preflight accepts the parameter so command lines still work with only the plan name changed.

Testing

• CONTROL_BRANCH=ci_test vagrant up on a clone with a local ci_test branch built production and ci_test environments. Both puppet and agent have environment = ci_test and report Using environment 'ci_test', and a second agent run on each exited 0.
• CONTROL_BRANCH=no_such_branch vagrant provision failed with "The control repo has no no_such_branch branch" before changing the agents.
• puppet-lint passes on the Bolt plans, and bolt plan show parses both plans.

🤖 Generated with Claude Code