Excellent question, and one which I have been experimenting and learning about as recently as last month.. I am saving this ultra-cool technique for part IIII, just because it is so radically bad@ss.

the right direction

 
function close_abort(){
  global $fp;
  @ fclose($fp);
}
 
register_shutdown_function(close_abort); // run function on connection close/reset
ignore_user_abort(true); // run script in background
set_time_limit(0); // run script forever
 
// open errorlog.txt in append mode
$fp = @ fopen('errorlog.txt', 'a+');
 
do{
  sleep(15);  // pauses script execution for 15 seconds
  echo "\n."; // to keep connection alive by sending data
 
  ob_start(); // start a buffer to capture output
  phpinfo();  // print php info to output
  $emess=ob_get_clean(); // save output into emess var
 
  fwrite($fp, $emess);   // write emess value to errorlog.txt
  break;  // outie
} while(true);
 
@ fclose($fp); // close errorlog.txt filepointer
 
// display emess for debugging
//header('Content-type: text/plain');
//echo $emess; 
 

Thank a lot for this interesting article. Question: how do you do error management in this technique? Specifically, what happens if some sort of error happens in the processing of background.php? You’ve just displayed a “Thank you” message, and then afterwards, you cannot inform the user that an error occurred.
Or, can you?