VOL-393 VOL-395 This update implements a replicated filesystem allowing
fluentd logs to be accessible from all cluster servers. The replicated
filesystem also limits the aggregate size of the logs to limit the
amount of disk space that they consume. This update also
creates a barrier filesystem for consul ensuring that the consul data
can't be affected by a disk full condition. Finally, some serious
changes were made to the fluentd clustering implememtation which was
very sub-optimal and wouldn't start correctly 9 times out of 10.
Added improved debygguing log collection that checks if the log file is
still growing before it kills the collection process. Added consul
release 0.9.2 to the production containers list. Removed fluent/fluentd
from containers.cfg. Added configuration options to the installer to set
the size of the barrier files.

Change-Id: I89389d76b081b0e6c3961e62d24277b1d2cf3cfb
diff --git a/install/ansible/roles/glusterfs/tasks/main.yml b/install/ansible/roles/glusterfs/tasks/main.yml
index e744d75..b8e17da 100644
--- a/install/ansible/roles/glusterfs/tasks/main.yml
+++ b/install/ansible/roles/glusterfs/tasks/main.yml
@@ -1,6 +1,6 @@
-- name: The glusterfs volume is created
+- name: The glusterfs registry volume is created
   gluster_volume:
-    bricks: "{{ replicated_fs_dir }}/brick1"
+    bricks: "{{ barrier_fs_dir }}/reg_brick1"
     force: true
     cluster: "{{groups.cluster | join(\",\")}}"
     replicas: 3
@@ -8,7 +8,21 @@
     state: present
   run_once: true
 
-- name: Start the gluster volume
+- name: The glusterfs logging volume is created
   gluster_volume:
-    name: registry_volume
+    bricks: "{{ barrier_fs_dir }}/log_brick1"
+    force: true
+    cluster: "{{groups.cluster | join(\",\")}}"
+    disperses: 3
+    name: logging_volume
+    state: present
+  run_once: true
+
+- name: Start the gluster volumes
+  gluster_volume:
+    name: "{{ item }}"
     state: started
+  with_items:
+    - registry_volume
+    - logging_volume
+