How can you change the execution of a playbook from sequential to parallel on machines?

To change task execution from sequential to parallel in Ansible, you can use the following approaches:

1. serial parameter: Allows running a playbook on a group of hosts in subsets of a specified size. This is not exactly parallel execution of all hosts at once, but it divides the group into "batches".

   # playbook.yml
   ---
   - name: Apply configuration
     hosts: webservers
     serial: 5 # Run on 5 hosts simultaneously
     tasks:
       - name: Update packages
         apt:
           name: '*'
           state: latest
   

2. throttle parameter: Limits the number of processes executing a specific task or block simultaneously. It is applied within a task or block.

   # playbook.yml
   ---
   - name: Apply configuration
     hosts: webservers
     tasks:
       - name: Deploy application
         command: /opt/deploy_app.sh
         throttle: 3 # Run this command on no more than 3 hosts at a time
   

3. free_strategy parameter: Changes the playbook execution strategy from default (sequential task execution on hosts) to free (tasks are executed on hosts as they become ready, without waiting for previous tasks to complete on the same host). This affects task execution within a single host.

   # playbook.yml
   ---
   - name: Configure servers with free strategy
     hosts: all
     strategy: free
     tasks:
       - name: Install nginx
         ansible.builtin.apt:
           name: nginx
           state: present
       - name: Install redis
         ansible.builtin.apt:
           name: redis
           state: present
   

4. Configuring forks parameter in ansible.cfg: This parameter controls the number of parallel processes that ansible launches to interact with hosts. Increasing forks allows Ansible to work with more hosts simultaneously.

   # ansible.cfg
   [defaults]
   forks = 30 # Increase the number of parallel processes to 30
   

5. Asynchronous tasks with polling (async and poll): For tasks that may take a long time, you can use asynchronous mode. Ansible starts the task on the remote machine and immediately moves on to the next hosts or tasks. It then periodically checks the status of the asynchronous task.

   # playbook.yml
   ---
   - name: Run long-running task asynchronously
     hosts: databases
     tasks:
       - name: Backup database
         command: /opt/backup_db.sh
         async: 1800 # Maximum execution time in seconds (30 minutes)
         poll: 15    # How often to check task status (every 15 seconds)
   

The most direct way to influence parallelism at the host level within a playbook execution (if serial is not used) is to use the forks parameter in the configuration. The strategy: free, serial, throttle, and asynchronous tasks provide more granular control over the order or concurrency of execution.

Method	Application level	Impact
forks	Ansible configuration	Maximum number of parallel hosts
serial	Playbook	Execution in groups (batches)
throttle	Task/Block	Limit on parallel task executions
strategy: free	Playbook	Parallel task execution within a host
async / poll	Task	Asynchronous task launch