How does distribute clean up Distributed without shutting down parallel processes

How does distribute clean up Distributed without shutting down parallel processes
Because it takes time to start the loop, and obviously the second run is faster than the first, I cycle through the parallel process and don’t want to quit and start the parallel process again.
First, the process 3 variable is called in the first loop

remotecall_fetch(varinfo, 3)
   name size summary
   ––––––––––––––––––––––––––––
   Base Module
   Core Module
   Distributed 1.454 GiB Module
   Main Module
The second loop calls process 3 variables
remotecall_fetch(varinfo, 5)
   name size summary
––––––––––––––––––––––––––––––––––––––––––––––
   Base Module
   Core Module
   Distributed 2.907 GiB Module
   Main Module
   drr3d_atuo 0 bytes drr3d_atuo (generic function with 1 method)

Use

clear! (:Distributed) 

will cause an error, prompting
invalid redefinition of constant Distributed!