Implement graceful shutdown for clustering

I am not sure exactly what needs to happen here, I think all that is required is for the node to leave the cluster before suspend starts.