That's... strange. This seems like an Ansible bug. I was able to reproduce it on all the versions which this role supports. I'll have to find some time to look more closely at this unless you've since found a resolution.

I ran into a similar issue where the redis_sentinel boolean set in the playbook's vars section wasn't being picked up by the DavidWittman.redis role.

I also tried moving any variables from group_vars or host_vars directly into the role similar to this:
- { role: Community/DavidWittman.redis, redis_sentinel: true, redis_sentinel_port: 26379 }

I downgraded from Ansible 2.2.1.0 to 2.1.4.0 as a workaround.

Also running into this error. When I specify the variable as part of the role item, it fails even when the port is 6379:

- name: Redis master
  hosts: redis-master
  roles:
    - role: DavidWittman.redis
      redis_port: "{{ redis_master_port }}"

I've managed to get rid of the error when I changed handlers/main.yml file from this:

 ---
 - name: "restart redis {{ redis_port }}"
   service:
     name: "{{ redis_service_name }}"
     state: restarted
   when: redis_as_service

to:

 ---
 - name: "restart redis {{ redis_port }}"
   service:
     name: redis_{{ redis_port }}
     state: restarted
   when: redis_as_service

Was having RUNNING HANDLER [DavidWittman.redis : restart redis {{ redis_port }}] ********** fatal: [192.168.33.10]: FAILED! => {"changed": false, "failed": true, "msg": "Unable to restart service redis_6379: Failed to restart redis_6379.service: Interactive authentication required.\nSee system logs and 'systemctl status redis_6379.service' for details.\n"}
and I got rid of the error by making the change as suggested above.