Did you ever get this working? I am very interested in using fail2ban with my discourse instance as well.