'beat' host_var breaks restart handler

[seanpringle]

Beats product:

filebeat

Beats version

7.0.0

Role version: (If using master please specify github sha)

5ec7143

OS version (uname -a if on a Unix-like system):

Ubuntu 18.04

Description of the problem including expected versus actual behaviour:

Defining the beat variable as a host_var breaks the restart handler on a fresh playbook run, with: ERROR! 'beat' is undefined. It isn't, though because role tasks referencing the variable succeed. Repeated attempts to run the playbook move the error to the next task that triggers a handler notification.

Think it might due to this limitation: ansible/ansible#48466 (comment)

Expected behaviour is that the beats role can use a host_var.

Playbook:
Please specify the full playbook used to reproduce this issue.

playbook:

- hosts: appservers
  roles:
    - elastic.beats

host_vars:

beat: filebeat
beat_conf:
  inputs:
    - type: log
      enabled: true
      paths:
        - /opt/app/log/*.log

Provide logs from Ansible:

...
TASK [beats : Debian - Ensure apt-transport-https is installed] **********************************************************************************************
ERROR! 'beat' is undefined

Repeated runs move the error to the next task that triggers a handler notify.

[ktibi]

Same issue with group_vars