adaptive load balancing with gnu parallel

Question

Is there some way to run gnu parallel with a dynamically changing list of remote hosts? The dynamism isn't intermittent or irregular -- I'm looking for a way to use the Google compute engine autoscaling feature to smoothly scale up to a max number of hosts and have gnu parallel dispatch jobs as these hosts come alive. I guess I can create a fake job to trigger autoscaling to launch the multiple hosts and have them register themselves to some central host file.. Any ideas how best to manage this?

Answer 1

From man parallel:

   --slf filename
            File with sshlogins. The file consists of sshlogins on
            separate lines.
            : 
            If the sshloginfile is changed it will be re-read when a
            job finishes though at most once per second. This makes it
            possible to add and remove hosts while running.