When automating RabbitMQ administration with Ansible, many developers encounter a puzzling scenario where commands work locally but fail remotely with:

fatal: [IP]: FAILED! => {"changed": false, "cmd": "rabbitmqctl list_vhosts", 
"failed": true, "msg": "[Errno 2] No such file or directory", "rc": 2}

The root cause typically stems from environment and PATH differences between local and remote execution. When Ansible connects remotely via SSH, it doesn't load the same shell environment as an interactive login session.

Key factors contributing to this behavior:

• The full path to rabbitmqctl isn't specified
• Environment variables (like ERLANG_HOME) aren't loaded
• The SSH session doesn't source the same profile scripts

1. Absolute Path Specification

The most reliable approach is using the full path to rabbitmqctl:

- name: get vhosts
  command: /usr/sbin/rabbitmqctl list_vhosts
  register: vhosts
  changed_when: false

2. Environment-Aware Execution

For complex setups where environment variables matter:

- name: get vhosts
  shell: |
    source /etc/profile
    /usr/sbin/rabbitmqctl list_vhosts
  args:
    executable: /bin/bash
  register: vhosts
  changed_when: false

3. Become User Context

When RabbitMQ runs under specific user permissions:

- name: get vhosts
  become: yes
  become_user: rabbitmq
  command: /usr/sbin/rabbitmqctl list_vhosts
  register: vhosts
  changed_when: false

For production environments, consider these robust patterns:

Custom Facts for Path Discovery

- name: find rabbitmqctl path
  shell: which rabbitmqctl
  register: rabbitmqctl_path
  changed_when: false

- name: use discovered path
  command: "{{ rabbitmqctl_path.stdout }} list_vhosts"
  register: vhosts
  changed_when: false

Using RabbitMQ Module (Ansible 2.8+)

- name: get vhosts
  rabbitmq_vhost:
    list: yes
  register: vhosts

When issues persist, add these debugging steps:

- name: debug environment
  shell: env
  register: env_output

- debug: var=env_output.stdout_lines

This reveals exactly what environment variables are available during remote execution.

Remember that RabbitMQ commands often require:

• Proper Erlang environment setup
• Correct cookie file permissions
• SELinux context considerations

Always test your playbook with -vvv flag for detailed execution tracing.

---

When running Ansible playbooks that execute rabbitmqctl commands remotely, you might encounter this frustrating error:

fatal: [IP]: FAILED! => {"changed": false, "cmd": "rabbitmqctl list_vhosts", 
"failed": true, "msg": "[Errno 2] No such file or directory", "rc": 2}

The puzzling aspect is that the same playbook works fine when run locally on the target machine. This suggests the issue relates to how Ansible executes commands remotely versus locally.

The fundamental reason this fails remotely stems from how Ansible handles command execution:

• Environment variables: Remote execution doesn't inherit the same environment as a local shell session
• Path resolution: The full path to rabbitmqctl isn't available in the restricted remote execution environment
• User context: The command may run under a different user than expected

After testing multiple approaches, here are the most reliable solutions:

1. Using Absolute Paths

The most straightforward fix is specifying the full path to rabbitmqctl:

- name: get vhosts
  command: /usr/sbin/rabbitmqctl list_vhosts
  register: vhosts
  changed_when: false

You may need to verify the exact path on your system with:

which rabbitmqctl

2. Setting the Correct Environment

RabbitMQ commands often require specific environment variables. This approach ensures they're available:

- name: get vhosts
  shell: |
    . /etc/profile
    . ~/.bashrc
    rabbitmqctl list_vhosts
  args:
    executable: /bin/bash
  register: vhosts
  changed_when: false

3. Using the RabbitMQ Ansible Module

For better reliability, consider using the dedicated rabbitmq_* modules:

- name: get vhosts
  rabbitmq_vhost:
    list: yes
  register: vhosts

The common suggestion to switch from command to shell often fails because:

• It still doesn't address the missing PATH issue
• Environment variables needed by rabbitmqctl aren't automatically loaded
• The execution context may lack necessary permissions

To better understand the remote execution context, add this debug task:

- name: debug environment
  shell: |
    echo "PATH: $PATH"
    echo "USER: $(whoami)"
    echo "HOME: $HOME"
  register: env_debug

- debug: var=env_debug.stdout_lines

For production environments, combining absolute paths with environment setup provides the most reliable solution:

- name: get vhosts reliably
  shell: |
    . /etc/profile
    /usr/sbin/rabbitmqctl list_vhosts
  args:
    executable: /bin/bash
  register: vhosts
  changed_when: false