Using the Ansible Stat Module on a Loop

Using the Ansible stat module on a loop

Hi again, it’s been a while since I wrote something on this blog. This time I was working on a Ansible playbook and I get this:

Problem

I want to verify if a file exists. Depending on the registered output I want to perform some other actions. On my specific use case, I want to use in in conjunct with the file module to define the state on my next task.

graph TD;
A(Stat over file) --> B{Does the file exists?};
B -->|Yes| C[state: file];
B -->|No| D[state: touch];

Solution

Here is the solution that worked for me, using loops, loop_control,and jinja2 filters.

- name: Stat over the files
  stat:
    path: "{{ my_loop }}"
    loop:
      - /etc/cron.allow
      - /etc/at.allow
    loop_control:
      loop_var: my_loop
    register: my_stat_var

- name: Create a file if not exists
  file:
    path: "{{ item.my_loop }}"
    owner: root
    group: root
    mode: og-rwx
    state: "{{ 'file' if item.stat.exists else 'touch' }}"
  loop: "{{ my_stat_var.results }}"

Basically, I’m using the new loop syntax to iterate over all the files that I want to check.

In order to avoid some warnings about the loop using item I implement the loop_control and loop_var syntax to control the loop behavior and on this specific case, instead of using the word item I will substitute it with the one that I define as my loop_var in this case is called my_loop (Remember this, I will use it on the next task).

I’m registering the result of the stat task on a variable, for this case is my_stat_var

Here is an example of the output, when the 2 files do not exists:

ok: [instance-amazon2] => {
    "my_loop": {
        "changed": false,
        "msg": "All items completed",
        "results": [
            {
                "ansible_loop_var": "my_loop",
                "at_cron_restricted_touch": "/etc/cron.allow",
                "changed": false,
                "failed": false,
                "invocation": {
                    "module_args": {
                        "checksum_algorithm": "sha1",
                        "follow": false,
                        "get_attributes": true,
                        "get_checksum": true,
                        "get_md5": false,
                        "get_mime": true,
                        "path": "/etc/cron.allow"
                    }
                },
                "stat": {
                    "exists": false
                }
            },
            {
                "ansible_loop_var": "my_loop",
                "at_cron_restricted_touch": "/etc/at.allow",
                "changed": false,
                "failed": false,
                "invocation": {
                    "module_args": {
                        "checksum_algorithm": "sha1",
                        "follow": false,
                        "get_attributes": true,
                        "get_checksum": true,
                        "get_md5": false,
                        "get_mime": true,
                        "path": "/etc/at.allow"
                    }
                },
                "stat": {
                    "exists": false
                }
            }
        ]
    }
}

On the next task I access the results of the registered variable my_stat_var using, according to the previous output, I need to use the results variable to access it and gives me 2 arrays (each one for each file)

Also, I extract the path from the same variable my_stat_var.results but as it is part of a loop, I will access it using item.my_loop as on the the second task loop I’m not using a loop_control, also I can access it using item.item but I prefer the first syntax. In case you implement a loop_control on you can access it as my_second_loop.my_loop

Also, I extract if the file exists with the variable item.stat.exists but I’m putting that on a jinja2 filter, that way it will evaluate and set the correct option on the state

# When the file exists
state: file
# When the file does not exists
state: touch

This behavior also brings idempotency to your task, because:

• If the files does not exists on the first iteration it will be created and,
• The next time you run the tasks as the files already exists it will works as a stat and will return the current state of path.

That is all for now. See you soon!