Workers don’t register themselves to Conductor servers. Instead, they poll tasks from a specified Conductor instance. To handle multiple instances, configure the worker with the desired server endpoint or use a load balancer to abstract multiple servers.