Catch errors raised when we try to delete objects that don't exist.

This will happen from time to time as we're calling #delete_container_objects
on each worker instance.

The calls to OpenShift are asynchronous so the next worker we tell
to #delete_container_objects may try to delete objects that have
already been removed.

Additionally, this makes it difficult to track when objects have
been deleted. Ideally we would scale a deployment down and have the
last worker delete the deployment iteslf, but a previous worker may
not have deleted the worker record by the time we tell the next worker
to exit.

lib/container_orchestrator.rb

@@ -47,18 +47,26 @@ def delete_deployment_config(name)
     scale(name, 0)
     connection.delete_deployment_config(name, my_namespace)
     delete_replication_controller(rc.metadata.name) if rc
+  rescue KubeException => e
+    raise unless e.message =~ /not found/
   end
 
   def delete_replication_controller(name)
     kube_connection.delete_replication_controller(name, my_namespace)
+  rescue KubeException => e
+    raise unless e.message =~ /not found/
   end
 
   def delete_service(name)
     kube_connection.delete_service(name, my_namespace)
+  rescue KubeException => e
+    raise unless e.message =~ /not found/
   end
 
   def delete_secret(name)
     kube_connection.delete_secret(name, my_namespace)
+  rescue KubeException => e
+    raise unless e.message =~ /not found/
   end
 
   private