Use logrotate to clean up from the log-queue-lenghts (sic) job

Right now these files never go away, and some are a million lines long.
This means splunk eats all the disk IO if we ever restart.

playbooks/roles/rabbitmq/tasks/main.yml

@@ -54,6 +54,15 @@
     name: "log-queue-lenghts"
     job: "{{ rabbitmq_app_dir }}/log-rabbitmq-queues.sh >/dev/null 2>&1"
 
+- name: install logrotate configuration
+  template:
+    src: etc/logrotate.d/rabbitmq.j2
+    dest: /etc/logrotate.d/rabbitmq
+  tags:
+    - "install"
+    - "install:configuration"
+    - "logrotate"
+
 # Defaulting to /var/lib/rabbitmq
 - name: create cookie directory
   file: >

playbooks/roles/rabbitmq/templates/etc/logrotate.d/rabbitmq.j2

+# We want to hit the top level queues and any vhost queues
+# such as notifier and fulfillment
+{{ rabbitmq_log_dir }}/*.log {{ rabbitmq_log_dir }}/*/*.log {
+  compress
+  dateext
+  dateformat -%Y%m%d-%s
+  missingok
+  daily
+  rotate 3
+}