Best way to run powershell command?

[Ab Generette]

On my (winrm) target host, I am able to run the following:

PS C:\Users\agenerette> c:/Python27/python.exe c:/xx/run/yy/tests/confluence_api_test.py

When I try to incorporate that into a playbook, however, I get a great block of mostly encoded error output that I'm not able to make sense of.  I've attached a couple of screen-shots to give a more complete picture of what I'm seeing, but here's an excerpt:

PLAY [873320-wrk04-ia] *********************************************************

TASK [Do a test run of Sharefiles and Confluence API scripts...] ***************

Traceback (most recent call last):

  File "/usr/lib/python2.6/site-packages/ansible/plugins/connection/winrm.py", line 274, in exec_command

    result = self._winrm_exec(cmd_parts[0], cmd_parts[1:], from_exec=True)

  File "/usr/lib/python2.6/site-packages/ansible/plugins/connection/winrm.py", line 219, in _winrm_exec

    response = Response(self.protocol.get_command_output(self.shell_id, command_id))

  File "/home/agenerette/.local/lib/python2.6/site-packages/winrm/protocol.py", line 398, in get_command_output

    self._raw_get_command_output(shell_id, command_id)

  File "/home/agenerette/.local/lib/python2.6/site-packages/winrm/protocol.py", line 417, in _raw_get_command_output

    res = self.send_message(xmltodict.unparse(req))

  File "/home/agenerette/.local/lib/python2.6/site-packages/winrm/protocol.py", line 244, in send_message

    fault = root.find('soapenv:Body/soapenv:Fault', xmlns)

TypeError: find() takes exactly 2 arguments (3 given)

failed: [873320-wrk04-ia] (item=c:/t3/run/ae/tests/sharefile_api_test.py) => {"item": "c:/t3/run/ae/tests/sharefile_api_test.py", "msg": "failed to exec cmd PowerShell -NoProfile -NonInteractive -ExecutionPolicy Unrestricted -EncodedCommand

And here's a copy of the playbook:

---

- hosts: "{{ targets }}"

  user: "{{ admin }}"

  gather_facts: yes

  vars:

    python_exec: c:/Python27/python.exe

    working_dir: c:/xx/run/yy

    sharefile_script: c:/xx/run/yy/tests/sharefile_api_test.py

    confl_script: c:/xx/run/yy/tests/confluence_api_test.py

  tasks:

    - name: Do a test run of Sharefiles and Confluence API scripts...

      win_shell: "'{{ python_exec }} {{ item }}'"

#     win_shell: "Get-Item Env:ldap_user"

      with_items:

        - "{{ sharefile_script }}"

        - "{{ confl_script }}"

      environment:

        LDAP_USER: "{{ my_ldap_user }}"

        LDAP_REG_CODE: "null"

        LDAP_PASS: "{{ my_ldap_pass }}"

        GDOC_PASS: "{{ my_gdoc_pass }}"

        SHARE_USER: "{{ my_sharefiles_user }}"

        SHARE_PASS: "{{ my_sharefiles_pass }}"

        WIKI_USERNAME: "{{ my_ldap_user }}"

        WIKI_PASSWORD: "{{ my_ldap_pass }}"

      ignore_errors: yes

      register: result

    - debug: var=result

[Jordan Borean]

The main issue you have is that there is a compatibility issue with pywinrm 0.3.0 and Pythoon 2.6 as shown here https://github.com/diyan/pywinrm/issues/204. This is a controller side issue and you either need to downgrade pywinrm to 0.2.2 or use Python 2.7, 3.5+.

Also what version of Ansible are you running, it doesn't look like it is pipelining the modules which was added in 2.3?

Thanks

Jordan

[Ab Generette]

Ok, I think I'm going to pull back and work on upgrading both Python (to 2.7.5) and Ansible.  We're 2.2.1.0, where the latter is concerned.

Thanks for your help, Jordan.

-Anthony

[Jordan Borean]

No worries, I was just curious as to why it wasn't pipelining. The syntax you had there seemed good to me, the only thing I would recommend is to use win_command instead of win_shell when calling the Python executable. The win_shell module should be used to execute things in a shell interpreter like PowerShell commands while win_command is used to execute executables.

Thanks

Jordan

[Ab Generette]

Ah, ok.  I thought I needed to use win_shell, because of the need to set environment variables.  I see that's not the case, though.  Thx.

[Jordan Borean]

Environment variables are set for all modules, except raw (and maybe script I can't remember). It is a mechanism done by Ansible before running the module code.

Thanks

Jordan