Timeout behavior inconsistency across Vagrant provisioner types

Direct Answers

Which provisioner types cause Vagrant to halt the VM on timeout? Only provisioners that Vagrant can directly terminate—typically the shell provisioner and any provisioner that exposes a native --timeout (or equivalent) flag that Vagrant passes through—will trigger a vagrant halt when the timeout is exceeded. Provisioners that run their own client executables and do not expose a timeout flag (e.g., ansible, chef, puppet) usually leave the VM running after a Vagrant‑level timeout because Vagrant cannot forcefully stop the child process chain.

Is the halt behavior controllable via Vagrantfile settings or environment variables? No, Vagrant does not provide a global switch to force all provisioners to halt the VM on timeout. The behavior is determined by whether Vagrant can signal the provisioner’s process tree. You can influence it per‑provisioner by:

• Using a wrapper script that enforces a timeout (e.g., timeout on Linux).
• Passing a native timeout flag if the provisioner supports one.
• Setting config.vm.provision :shell, timeout: N only for the shell provisioner.

Can users guarantee a uniform VM state (always running or always halted) regardless of the provisioner used? Not out of the box. To guarantee a uniform outcome you must either:

• Wrap every provisioner invocation in a shell wrapper that enforces the timeout and handles the exit code, or
• Use a custom provisioner plugin that implements consistent timeout handling across all types.

Likely Explanation

Vagrant’s timeout logic is implemented in the core and applies only to the command it starts for a provisioner. For the shell provisioner, Vagrant starts /bin/sh directly and can send a SIGTERM when the timer expires, causing the VM to halt if the shell exits non‑zero. For provisioners that launch external executables, Vagrant can only signal the top‑level process; if that process spawns long‑running children (common in Chef, Ansible, Puppet), the signal may not reach them, and the VM remains running. Additionally, some provisioners use SSH to execute commands and treat Vagrant’s timeout as an SSH session timeout, not a playbook execution timeout.

Steps for Your Current Setup

1. Run vagrant up --debug and locate the line that starts the provisioner. Verify if a --timeout flag appears in the command.
2. For ansible or other non‑shell provisioners, check the provisioner’s own documentation for a timeout option (e.g., --timeout or --forks).
3. If no native timeout exists, create a wrapper script:

   # wrapper.sh
   #!/usr/bin/env bash
   # Use the Unix timeout utility
   exec timeout $VAGRANT_TIMEOUT_CMD your_provisioner_command
   

   Add to Vagrantfile:

   config.vm.provision :shell, path: 'wrapper.sh', timeout: 300
   

4. After modifying, run vagrant up to confirm that a timeout now halts the VM as desired.

Diagnostic Detail Needed

To refine the recommendation, please confirm whether the provisioner you are using exposes a native timeout flag that Vagrant can forward. If it does, Vagrant will normally halt the VM on timeout; if not, you’ll need a wrapper approach.