Ansible - starting a task after cpu load is below 2.0

I am trying to create a playbook where I want to perform a simple debug task after cpu load is below 2.0.

I have this so far in cpu-load.yml:

---
- name: Check CPU load and wait
  hosts: localhost
  gather_facts: yes
  
  tasks:
    - name: Check cpu load
      shell: uptime | awk -F 'load average:' '{print $2}' | awk -F ', ' '{print $1}'
      register: cpu_load
      
    - name: Wait until cpu load is bellow 2.0
      wait_for:
        timeout: 300
        delay: 10
        shell: Do something here
        msg: "cpu load is bellow 2.0"
      
    - name: Continue jobs
      debug:
        msg: "CPU load is bellow 2.0. Continue!!!"

Now I am not sure how to make the task wait for the cpu load to go bellow 2.0. Can you guys help?

You need to put an until loop around your "check cpu load" task:

- hosts: localhost
  gather_facts: false
  tasks:
    - name: Check cpu load
      shell: uptime | awk -F 'load average:' '{print $2}' | awk -F ', ' '{print $1}'
      register: cpu_load
      until: cpu_load.stdout|float < 2.0
      retries: 300
      delay: 1

    - name: Some other task
      debug:
        msg: hello world

This will wait up to five minutes (300 retries with a 1-second delay) for the load average to drop below 2.0.

There are probably better ways to get the current 1-minute CPU load; reading from /proc/loadavg may be easiest:

- hosts: localhost
  gather_facts: false
  tasks:
    - name: Check cpu load
      command: cat /proc/loadavg
      register: cpu_load
      until: cpu_load.stdout.split()|first|float < 2.0
      retries: 300
      delay: 1

    - name: Some other task
      debug:
        msg: hello world

Building up on @larsks answer and going further regarding the sentence
> There are probably better ways to get the current 1-minute CPU load

There's actually a fact containing this information, at least on linux, e.g.

$ ansible localhost -m setup -a gather_subset='!all,!min,loadavg'
localhost | SUCCESS => {
    "ansible_facts": {
        "ansible_loadavg": {
            "15m": 0.669921875,
            "1m": 0.48974609375,
            "5m": 0.4501953125
        },
        "gather_subset": [
            "!all",
            "!min",
            "loadavg"
        ],
        "module_setup": true
    },
    "changed": false
}

Notes:

1. gather_subset='!all,!min,loadavg' ensures that stricly only the needed facts are gathered (and refreshed in the below playbook). For more information, see the option documentation
2. the loadavg subset is not documented at time I write these lines but is listed in the allowed options from the module error message in my ansible 2.14.6 / 2.15.0 versions (see below)

Knowing this and applying the same recipe as in @larsk's answer, the check can be implemented as:

---
- name: Demo play to check load and continue
  hosts: localhost
  gather_facts: false

  tasks:
    - name: Check load before next tasks
      ansible.builtin.setup:
        gather_subset:
          - '!all'
          - '!min'
          - 'loadavg'
      retries: 30
      delay: 5
      until: ansible_loadavg['1m'] < 2.00

    - name: Now do something on a cooled down system
      debug:
        msg: I'm doing something without pressure

Regarding the undocumented subset for the setup module, here is a quick and dirty solution to list all accepted subsets in your version of the module:

$ ansible localhost -m setup -a gather_subset='toto'
localhost | FAILED! => {
    "changed": false,
    "msg": "Bad subset 'toto' given to Ansible. gather_subset options allowed: all, all_ipv4_addresses, all_ipv6_addresses, apparmor, architecture, caps, chroot, cmdline, date_time, default_ipv4, default_ipv6, devices, distribution, distribution_major_version, distribution_release, distribution_version, dns, effective_group_ids, effective_user_id, env, facter, fibre_channel_wwn, fips, hardware, interfaces, is_chroot, iscsi, kernel, kernel_version, loadavg, local, lsb, machine, machine_id, mounts, network, nvme, ohai, os_family, pkg_mgr, platform, processor, processor_cores, processor_count, python, python_version, real_user_id, selinux, service_mgr, ssh_host_key_dsa_public, ssh_host_key_ecdsa_public, ssh_host_key_ed25519_public, ssh_host_key_rsa_public, ssh_host_pub_keys, ssh_pub_keys, system, system_capabilities, system_capabilities_enforced, user, user_dir, user_gecos, user_gid, user_id, user_shell, user_uid, virtual, virtualization_role, virtualization_tech_guest, virtualization_tech_host, virtualization_type"
}