It seems that this can be safely done on our openstack cluster. The procedure is as follows:

*On the node that you wish to reset*
- `sudo rabbitmqctl stop_app` (no systemctl stop otherwise the reset will fail)
- `sudo rabbitmqctl reset`
- `sudo rabbitmqctl join_cluster rabbit@anothermember`
- `sudo rabbitmqctl start_app`

*On one of the other nodes*
- `sudo rabbitmq-queues grow rabbit@resetnode all`
- `sudo rabbitmq-queues rebalance all` || quorum