Correct way to include a Distributed workflow in a module

General Usage
1

I want to produce a private package that fully manages and runs a series of distributed processes on private workers, but I’m seeing unstable code loading & scope issues. The kind of approach below errors most of the time with On worker 12: UndefVarError: #104#111 not defined type messages (which seems to be a code loading scope issue with the pmap anonymous function) but sometimes it works…

The general idea is this…
Usage:

using Foo
Foo.launch() #add private remote machines to the workers and load code on each
Foo.rundist() #run distributed processes, while monitoring workers locally 

module Foo
using Distributed

export dosteps

function launch()
    remote_hosts = get_dns_remote_machines()
    addprocs(remote_hosts)
    @everwhere @eval using Foo
end

function dosteps(i, remchan) #The function that does the actual work
    #do something long that produces a dict with info
    put!(remchan, info_dict)
    #do something else long that produces a dict with info
    put!(remchan info_dict)
end

const EXIT = Ref{Bool}(false)
function rundist()
    infochan = RemoteChannel(()->Channel{Dict{Symbol,Any}}(100))
    
    @async begin #an async local worker monitoring process
        while !EXIT[]
            d = take!(infochaan)
            message = # form a messsage from dict d
            println(message)
            wait(Timer(1))
        end
    end

    try
        pmap(i->dosteps(i, infochan), 1:10)
    catch e
        EXIT[] = true
        throw(e)
    end
    EXIT[] = true
end

end #module