to add to this, distribute your worker processes onto multiple machines. That way if you lose a machine, your queue will still process jobs (albeit slightly cri

to add to this, distribute your worker processes onto multiple machines. That way if you lose a machine, your queue will still process jobs (albeit slightly crippled).