In case of network error it may happend that bubuku won't start

[antban]

If unhandled exception will occur in run_daemon_loop cycle, that bubuku process will try to restart it's own work cycle, and recreate all internal entities (having in mind existing kafka process).
But if in that case AmazonEnvProvder.get_region() will return null, than it won't be able to refresh exhibitor address list, and that will start unlimited cycle of exceptions (see AWSExhibitorAddressProvider.get_addresses_by_lb_name).