First playbook, first problem

[gpow81]

Hello, I hope I am asking this in the right place. I will be working on a project which will involve running ansible playbooks on the z/OS. I wanted to start learning with something simple and I think I am close but I stuck on a python syntax error.

SyntaxError: Non-UTF-8 code starting with '\xa4' in file on line 2, but no encoding declared; see https://python.org/dev/peps/pep-0263/ for details

On the z/OS I have

• python 3.10.5
• zoau-1.2.0.1.
• USS is working
• group_vars with all.yml from one of the samples and host_vars.yml with paths.

On my workstation I have ansible with installed ibm_zos_core collection.
Whichever module I would try to run (or even just -m ping) , I am getting same error about the encoding. Funny thing is that documentation says to set "pipelining = True" to make that particular error go away but in my case the error started showing up after making it "true". With having pipelining set to False, it of course transfers a python script to z/OS and tries to run which fails for some other reason:

"The TSO command "LU GPUSER" execution failed.", "changed": false, "output": [{"command": "LU GPUSER", "rc": 16, "content": ["\udcc2\udcd7\udce7\udce6\udcf0\udcf0\udcf0\udcf0\udcc9@\u0167\udc85\udc83@\udc95\udc96\udca3@\udc86\udc96\udca4\udc95\udc84\u0015"], "lines": 1}], "failed": true, "invocation": {"module_args": {"commands": ["LU GPUSER"]}}}

Interestingly, if I intercept the temporary python script uploaded to the z/OS and manually run it via python interpreter on my z/OS unix shell it actually works and gives me good output.

Does it have something to do with conversion (or lack of it) from ascii to ebdic?
Please let me know if you know what I have missed. thanks!

[rexemin]

Hi, thanks for reaching out to the team, could you tell us which version of the collection you're using as well as which version of Ansible? It's worth noting we currently don't certify that our collection will work completely with ZOAU versions greater than 1.1.1 nor with Python 3.10, we're working on that for our next releases.
It could be that your issue is related to the environment variables you have set in all.yml and host_vars.yml. When you run the python script manually, are there any differences on the environment/path set on your shell and the one you configure in Ansible?
It would also help if you could run the playbook in verbose mode with -vvv and share the output it displays when running the task that is causing you problems.

[gpow81]

Can I ask someone to run this simple test?
printf 'print("Hello, world")' | ssh user@zos "/bin/sh -c /path_to_python/bin/python"

[gpow81]

I've switched to python3.8 and now hello world works! But, ansible is giving me exact same error as I had before with pipelining OFF.
So now it looks like this:

Using module file /root/.ansible/collections/ansible_collections/ibm/ibm_zos_core/plugins/modules/zos_tso_command.py
Pipelining is enabled.
  3039 1661179177.82748: _low_level_execute_command(): starting
  3039 1661179177.82755: _low_level_execute_command(): executing: /bin/sh -c '/tmp/python/usr/lpp/IBM/cyp/v3r8/pyz/bin/python3 && sleep 0'
<zos1> ESTABLISH SSH CONNECTION FOR USER: gpowied
<zos1> SSH: EXEC ssh -C -o ControlMaster=auto -o ControlPersist=60s -o KbdInteractiveAuthentication=no -o PreferredAuthentications=gssapi-with-mic,gssapi-keyex,hostbased,publickey -o PasswordAuthentication=no -o 'User="gpowied"' -o ConnectTimeout=10 -o ControlPath=/root/.ansible/cp/6911bd2538 zos1 '/bin/sh -c '"'"'/tmp/python/usr/lpp/IBM/cyp/v3r8/pyz/bin/python3 && sleep 0'"'"''
  3039 1661179177.83109: Sending initial data
  3039 1661179177.91927: Sent initial data (117092 bytes)
  3039 1661179178.58019: stdout chunk (state=3):
>>>
{"msg": "The TSO command \"netstat\" execution failed.", "changed": false, "output": [{"command": "netstat", "rc": 16, "content": ["\udcc2\udcd7\udce7\udce6\udcf0\udcf0\udcf0\udcf0\udcc9@\u0167\udc85\udc83@\udc95\udc96\udca3@\udc86\udc96\udca4\udc95\udc84\u0015"], "lines": 1}], "failed": true, "invocation": {"module_args": {"commands": ["netstat"]}}}
<<<

[rexemin]

Thank you for the additional information, a couple of things to note: since you're using Ansible 2.9.6, you should keep using pipelining = False. Additionally, Ansible is not using the environment variables you configured in your inventory, in the output you shared the task is run like this:
SSH: EXEC ssh -C -o ControlMaster=auto -o ControlPersist=60s -o KbdInteractiveAuthentication=no -o PreferredAuthentications=gssapi-with-mic,gssapi-keyex,hostbased,publickey -o PasswordAuthentication=no -o 'User="gpowied"' -o ConnectTimeout=10 -o ControlPath=/root/.ansible/cp/6911bd2538 zos1 '/bin/sh -c '"'"'/usr/lpp/IBM/cyp/v3r10/pyz/bin/python3.10 && sleep 0'"'"''

When it should look like this:
SSH: EXEC ssh -C -o ControlMaster=auto -o ControlPersist=60s -o StrictHostKeyChecking=no -o Port=22 -o KbdInteractiveAuthentication=no -o PreferredAuthentications=gssapi-with-mic,gssapi-keyex,hostbased,publickey -o PasswordAuthentication=no -o ‘User=“omvsadm”’ -o ConnectTimeout=10 -o ControlPath=/Users/iamorenosoto/.ansible/cp/de618e5fd4 -tt [ec01134a.vmec.svl.ibm.com](http://ec01134a.vmec.svl.ibm.com/) ‘/bin/sh -c ‘“’”‘_BPXK_AUTOCVT=ON ZOAU_HOME=/zoau/v1.2.0 PYTHONPATH=/zoau/v1.2.0/lib LIBPATH=/zoau/v1.2.0/lib:/python3/usr/lpp/IBM/cyp/v3r8/pyz/lib:/lib:/usr/lib:. PATH=/zoau/v1.2.0/bin:/python3/usr/lpp/IBM/cyp/v3r8/pyz/bin:/bin:/var/bin _CEE_RUNOPTS=‘“’”‘“’”‘“’”‘“’”‘FILETAG(AUTOCVT,AUTOTAG) POSIX(ON)‘“’”‘“’”‘“’”‘“’”' _TAG_REDIR_ERR=txt _TAG_REDIR_IN=txt _TAG_REDIR_OUT=txt LANG=C /python3/usr/lpp/IBM/cyp/v3r8/pyz/bin/python3.8 /.ansible/tmp/ansible-tmp-1660955267.8000598-43378-135487875126096/AnsiballZ_zos_mvs_raw.py && sleep 0'“‘”’'

Are you sourcing the environment variables in your playbook? If you take a look at this sample playbook you can source them with this line before the tasks: environment: "{{ environment_vars }}"

[gpow81]

This was it! At some point in my desperation I've dropped that line and I was sure that variables were still being sourced anyway because they were still showing up in my "list all variables" task . But I hadn't noticed that they have stopped showing on the ssh line!

pipelining = True - works only with python 3.8 and not 3.10.5
pipelining = False works with python 3.8 and 3.10.5

Big thank you!

[ddimatos]

We are working with the Python team to see if they can support pipelining=True for Python 3.10 and later.

[ddimatos]

The latest PTF/patch from IBM Enterprise Python for 3.10 and 3.11 supports a new environment variable that allows for pipelining=true to operate as it did in the past. Tha environment var is PYTHONSTDINENCODING: "cp1047". See git discussion #657 for an example.