The issue is 3 different messages

with back to back to back runs

As in, I literally ran them one after another and got completely different messages

To clarify that part specifically, in case it wasn't clear already: that sounds to me like your puppetserver host has three or more jruby instances (

max-active-instances

in

puppetserver.conf

), or perhaps you have multiple puppetservers behind a load balancer, and those instances each have different data cached.