gluster_volume task fails when removing brick from disconnected peer

[mtruneck]

The scenario:

• gluster_volume is correctly configured across an ansible group
• one of the group's peer disconnects permanently (cluster scaled down in our case)
• run gluster_volume ansible without the disconnected peer in the group

Expected result

• gluster_volume task automatically removes the bricks from the disconnected peer

Real behaviour

• it fails with ValueError: invalid literal for int() with base 10.

Reason

The problem is on the line 430 in gluster_volume.py:

gluster-ansible-collection/plugins/modules/gluster_volume.py

Lines 426 to 431 in 3b3d406

	def reduce_config(name, removed_bricks, replicas, force):
	out = run_gluster(['volume', 'heal', name, 'info'])
	summary = out.split("\n")
	for line in summary:
	if 'Number' in line and int(line.split(":")[1].strip()) != 0:
	module.fail_json(msg="Operation aborted, self-heal in progress.")

Because it expects output from gluster volume heal [name] info like this:

Brick 10.10.1.102:/opt/volume
Status: Connected
Number of entries: 0

But in case the peer disconnected, the output is

Brick 10.10.1.102:/opt/volume
Status: Transport endpoint is not connected
Number of entries: -

So the condition on line 430 fails, because ' -' is not an int.

Task used:

- name: Configure Gluster volume.
  gluster_volume:
    state: present
    name: "{{ brick_name }}"
    brick: "{{ brick_dir }}"
    replicas: "{{ groups[gluster_group] | length }}"
    cluster: "{{ groups[gluster_group] | map('extract', hostvars, 'ansible_all_ipv4_addresses') | map('select', 'search', '^10\\.') | map('first') | list }}"
    host: "{{ ansible_all_ipv4_addresses | select('search', '^10\\.') | first }}"
    force: yes
  run_once: true