It turns out that it was just pure coincidence that my change was made on the voice_network rules and that the root cause of this was a gateway being down on the secondary device and XMLRPC triggering the filter reload flushed states as this is what I have enabled.