Error handling & debugging

By default, if a task fails on one host, Ansible stops running further tasks on that host — but keeps going on every other host in the play. Knowing that default, and how to override it deliberately, avoids a lot of confusion mid-incident.

# ignore a failure on this one task and keep going
- name: Attempt to clean old logs (ok if none exist)
  shell: rm /var/log/app/*.old
  ignore_errors: true

# a command that "fails" normally, but here 1 is actually fine
- name: Check for a lockfile
  command: test -f /tmp/app.lock
  register: lock_check
  failed_when: lock_check.rc not in [0, 1]

# custom, more readable failure condition
- name: Verify service is actually listening
  shell: curl -sf http://localhost:8080/health
  register: health
  failed_when: health.rc != 0

The single most useful debugging tool is register + debug — capture a task's result, then print it to see exactly what Ansible saw.

- name: Get disk usage
  command: df -h /
  register: disk_result

- name: Show what we got
  debug:
    var: disk_result.stdout

# run-time flags that make debugging any playbook easier
ansible-playbook site.yml -v       # more output
ansible-playbook site.yml -vvv     # a LOT more output, incl. module internals
ansible-playbook site.yml --start-at-task="Deploy config"   # resume mid-playbook
ansible-playbook site.yml --limit=web3.example.com               # retry on just the host that failed