int() argument must be a string, a bytes-like object or a number, not 'NoneType'

[bitshiftnetau]

Summary

When I try to create a container on Proxmox, it fails saying that "fatal: [central.local]: FAILED! => {"changed": false, "msg": "creation of lxc VM 810 failed with exception: int() argument must be a string, a bytes-like object or a number, not 'NoneType'"}"

Issue Type

Bug Report

Component Name

community.general.proxmox

Ansible Version

$ ansible --version
ansible [core 2.12.0]
  config file = /etc/ansible/ansible.cfg
  configured module search path = ['/root/.ansible/plugins/modules', '/usr/share/ansible/plugins/modules']
  ansible python module location = /usr/lib/python3.9/site-packages/ansible
  ansible collection location = /root/.ansible/collections:/usr/share/ansible/collections
  executable location = /usr/bin/ansible
  python version = 3.9.9 (main, Nov 20 2021, 21:30:06) [GCC 11.1.0]
  jinja version = 3.0.3
  libyaml = True

Community.general Version

$ ansible-galaxy collection list community.general
# /usr/lib/python3.9/site-packages/ansible_collections
Collection        Version
----------------- -------
community.general 4.0.2  

# /root/.ansible/collections/ansible_collections
Collection        Version
----------------- -------
community.general 4.1.0 

Configuration

$ ansible-config dump --only-changed
INVENTORY_ENABLED(/etc/ansible/ansible.cfg) = ['yaml', 'ini']

OS / Environment

Arch Linux in LXC on Proxmox (although I get the exact same errors running from a bare-metal install of Arch Linux)

Steps to Reproduce

---Playbook extracted from proxmox documentation---

- name: Create new LXC container in Proxmox
  hosts: central.local
  tasks:
    - name: testing
      proxmox:
        vmid: 810
        node: central
        api_user: root@pam
        api_password: <password>
        api_host: central
        password: 123456
        hostname: example.org
        ostemplate: 'local:vztmpl/archlinux-base_20210420-1_amd64.tar.gz'


---Custom playbook---

- name: Create new LXC container in Proxmox
  hosts: 'central.local'
  tasks:
    - name: 'Create container'
      proxmox:
- name: Create new LXC container in Proxmox
  hosts: 'central.local'
  tasks:
    - name: 'Create container'
      proxmox:
        vmid: '810' # specifying container ID
        api_user: 'root@pam' # Proxmox user
        api_password: '' # password in plaintext
        api_host: 'central.local' # proxmox hostname
        password: 'password' # container password in plaintext
        hostname: 'CT-Prometheus-ArchLinux' # container hostname
        node: 'central' # name of proxmox host
        cores: '1'
        cpus: '2'
        cpuunits: '1000'
        ostemplate: 'local:vztmpl/archlinux-base_20210420-1_amd64.tar.gz'
        storage: 'VMs'
        disk: '16'
        memory: '512'
        nameserver: '8.8.8.8'
        netif: '{"net0":"name=eth0,ip=dhcp,bridge=vmbr10"}'
        state: 'present'

Expected Results

I expect it to connect and create a container.

Actual Results

ansible-playbook playbooks/vm_template_2.yml --user root@pam --ask-pass
SSH password: 

PLAY [Create new LXC container in Proxmox] *********************************************

TASK [Gathering Facts] *****************************************************************
ok: [central.local]

TASK [testing] *************************************************************************
fatal: [central.local]: FAILED! => {"changed": false, "msg": "creation of lxc VM 810 failed with exception: int() argument must be a string, a bytes-like object or a number, not 'NoneType'"}

PLAY RECAP *****************************************************************************
central.local              : ok=1    changed=0    unreachable=0    failed=1    skipped=0    rescued=0    ignored=0   

Code of Conduct

• I agree to follow the Ansible Code of Conduct

[ansibullbot]

Files identified in the description:

• plugins/modules/cloud/misc/proxmox

If these files are incorrect, please update the component name section of the description or use the !component bot command.

click here for bot help

[ansibullbot]

cc @Ajpantuso @Thulium-Drake @joshainglis @karmab @tleguern
click here for bot help

[bitshiftnetau]

This issue is resolved following a full system reboot of proxmox. Cannot say for other systems but it should be noted in the community docs if anyone else can replicate this error.

EDIT: this was a false positive. System reboot did not restore. See below.

[bitshiftnetau]

I spoke too soon. It appears incapable of creating a new container.

If I use the following playbook:

• name: Create new LXC container in Proxmox
  hosts: central.local
  tasks:
  • name: testing
    proxmox:
    vmid: 810
    node: central
    api_user: root@pam
    api_password:
    api_host: central
    password: 123456
    hostname: example.org
    ostemplate: 'local:vztmpl/archlinux-base_20210420-1_amd64.tar.gz'

It will return "ok", HOWEVER a container at id 810 already exists on the proxmox installation.

Which means two things:

1. this "ok" is a false positive
2. Nothing happens, no container is created

Altering this number exactly to any other number not yet assigned to a VM or container returns the error in OP. This example comes directly from the examples in your docs.

[bitshiftnetau]

Anyone going to address this? Not expecting an immediate fix, but this error renders Ansible completely useless in Proxmox.

[felixfontein]

Nobody seems to be working on this. I think this is caused because you don't set the onboot parameter; the code handling it ignores that it can not be specified, and will likely (judging from glancing over it) produce this exact error when you don't specify it.

Changing line 608 to

                            onboot=int(module.params['onboot']) if module.params['onboot'] is not None else None,

should fix it (assuming my guess is correct).

[russoz]

Wouldn't it be slightly better to add default=False to the argument spec?

[felixfontein]

No, because we just spent quite some amount of work to get rid of that default.

[bitshiftnetau]

onboot=int(module.params['onboot']) if module.params['onboot'] is not None else None

Could we do is not None else 0 ?

[felixfontein]

No, for exactly the same reason: that would re-introduce the default.

[bitshiftnetau]

Oh right of course haha I totally read that wrong.

What about making onboot a mandatory value? That way there is no default. The user either defines it to boot on startup or defines it to not boot on startup.

[felixfontein]

That's a breaking change and will not be accepted without a long deprecation period.

[bitshiftnetau]

That's a breaking change and will not be accepted without a long deprecation period.

Technically yes? But also isn't that kind of a moot point as this is a major bug which results in this value needing to be set as a workaround (tested and confirmed) regardless? Else it doesn't work at all?

Either way, software development protocol yadda yadda... all good, I gets how it be 😅

Ok cool, so should we define that as a workaround? For now we note that without defining this value, it throws this error and won't work.

[felixfontein]

If someone wants to touch the module, how about simply fixing the bug, instead of adding some other random restriction which will avoid the bug to happen? Fixing the bug requires less effort (considering possible deprecations, required documentation, explanation in changelog etc.) than all the other things proposed here.

[bitshiftnetau]

Happy to do that, but do you have a known solution? Do we know what the fundamental issue is under the hood?

If not, then I don't have the resources atm to commit for known unknowns. From my pov, it's simply easier to push through the deprecation cycle and stick with a workaround.

Either way, the workaround exists in the interim.

[felixfontein]

I'm pretty sure #3844 (comment) fixes the issue (since the fundamental reason is that int(None) gets called when you don't specify onboot as a module parameter), but I have no way to verify this because I don't use proxmox.

[bitshiftnetau]

We seem to be going in circles. That solution is unfortunately out of the question, given the removal of default behaviour which I happen to agree with at the very least from a maintainer's pov.

Making this flag mandatory has other benefits in terms of basic operations hygene and makes sense also from the Ansible/config mgmt pov.

Does it boot? Does it not boot? Unacceptable unknown imo. Setting should be forced on the user.

Deprecation cycles aren't so bad, workarounds are very normal.

[felixfontein]

We seem to be going in circles. That solution is unfortunately out of the question, given the removal of default behaviour which I happen to agree with at the very least from a maintainer's pov.

I don't understand what you are saying. The comment I linked to is handling the option correctly when it is not provided by the user.

[bitshiftnetau]

We seem to be going in circles. That solution is unfortunately out of the question, given the removal of default behaviour which I happen to agree with at the very least from a maintainer's pov.

I don't understand what you are saying. The comment I linked to is handling the option correctly when it is not provided by the user.

Oh sorry, I'm really not reading properly today.

Well no, coz the error is related to passing through None when the option isn't set in the first place right? If we then just give it None then we're not really fixing anything.

[felixfontein]

Please look closely. None is not passed to int(). But it is passed on as the value of onboot (which is correct and required) if the user provided None.

[Ajpantuso]

I can pick this up later today. I'll add the inverse operation of

community.general/plugins/module_utils/proxmox.py

Lines 52 to 54 in 80d650f

	def proxmox_to_ansible_bool(value):
	'''Convert Proxmox representation of a boolean to be ansible-friendly'''
	return True if value == 1 else False

to the module_utils and wrap all the bool parameters in it. The other bool parameters have defaults so they are not affected currently, but obviously could be if defaults are removed.

EDIT: By inverse I mean True -> 1, False -> 0, None -> None.

[bitshiftnetau]

Please look closely. None is not passed to int(). But it is passed on as the value of onboot (which is correct and required) if the user provided None.

I was operating under the assumption that the value passed to onboot was the issue in question, not the value passed to int() but now I understand.