Not sure if this is a big help to you, but you can use the return code of the child processes to communicate with the parent process. I did that in this case:

$pm->run_on_finish(
    sub {
        my ($pid, $exit_code, $ident) = @_;
        $connected_server_count += $exit_code;
        $progress = ($count / $total_server_count) * 100;
        printf "Progress: %4.1f\%\r",$progress;
        $count++;
    }
);
[download]

Example of a bad (0) return:

$pm->finish(0); # Couldn't connect to this one, skip to next server in
+ main loop
[download]

Example of a good (1) return:

    # Close ssh connection to this server
$ssh->close();
$pm->finish(1);
[download]

The point being that $count kept track of how many servers were processed ($connected_server_count is used elsewhere) but it based that count on the return code. In this code, a server returned 0 if it could not connect, but returned 1 if it could, so I just summed those values. I've not played around with it, but I assume you can map out any numeric code to suit your purposes, perhaps the number of days, etc.