Description of problem: Currently, the API and RPC worker counts are calculated to match the number of processors. Given that multiple components support multiple worker processes for different services, this can result in a lot of processes depending on the node role. Providing the ability to control the worker counts would allow purpose-driven tuning.
The relevant configuration items are found in /etc/neutron.conf and they are: api_workers = [integer value >= 0] rpc_workers = [integer value >= 0] for kilo: 0 means no separate process for worker threads
This bug is against a Version which has reached End of Life. If it's still present in supported release (http://releases.openstack.org), please update Version and reopen.