Why is this a problem? If you are running a command that is non-idempotent, the command will be started over again which may cause a failure in your playbook execution. Recently I ran into a situation where Ansible was rerunning on a task on a node even though the task was not configured to be retried. It turns out that the ssh connection will retry actions auto-magically if the ssh connection fails while waiting for the task to execute. Here's how I was able to reproduce the issue.... Environment Setup I started by setting up a fake environment using containers. I have a scale-ssh project that can be used to launch a set of containers running ssh that can be used to run Ansible against. $ git clone https://github.com/mwhahaha/scale-ssh $ cd scale-ssh $ ./build-container.sh $ cp ~/.ssh/id_rsa.pub authorized_keys $ ./run-containers.sh 10 After the containers have launched, we now have an Ansible inventory file that can be used to execute Ansible via SSH to our fre...