Use the distributed memcached locked dbm cache backend

This commit switches the backend used on the api server to be the dbm
file backend with using memcached for distributed locking. This enables
having an async worker, but avoids the size limitations with storing
the cached data in memcached directly.

Change-Id: If506bbca1eb867135430c7f435c5e4a1db1a0232
Depends-On: Ied241ca1762c62a047bd366d7bd105028a884f30

manifests/api.pp

@@ -105,7 +105,7 @@ class openstack_health::api(
   }
 
   exec { 'requirements':
-    command     => "${virtualenv_dir}/bin/pip install -U -r ${source_dir}/requirements.txt pylibmc",
+    command     => "${virtualenv_dir}/bin/pip install -U -r ${source_dir}/requirements.txt",
     require     => [
       Python::Virtualenv[$virtualenv_dir],
       Package['libmemcached-dev'],

templates/openstack-health.conf.erb

@@ -1,7 +1,6 @@
 [default]
 query_dir = <%= @elastic_recheck_dir %>/queries
 cache_expiration = <%= @cache_expiration %>
-cache_backend = dogpile.cache.pylibmc
 cache_url = 127.0.0.1
 db_uri = <%= scope.lookupvar("openstack_health::api::db_uri") %>
 <% if @ignored_keys.length > 0 -%>