[kolla] [rabbitmq] Another RMQ config question

We’re running kolla-ansible Train and implementing durable queues. I was able to make it work by adding these lines to etc/kolla/config/global.conf:

[oslo_messaging_rabbit] amqp_durable_queues = True

But we want to do it one cluster at a time, so I tried commenting that out and adding line: amqp_durable_queues: "True" to group_vars/<region>/vars.yml but that seems to have no effect. Where can I put this value to enable durable queues per-cluster?