That is the appropriate work around. There must be ACEs on the outside interface to allow the traffic in. The ACEs must references the IP addresses handed out to clients as the source. The destination depends on the configuration.
If you have nat 0 configured to prevent translation for traffic going from the internal server to the VPN pool addresses, then the ACEs should reference the private IP of server as the destination. This is recommended.
You could also setup a static statement for the server to be available on the outside interface by its inside address, but this would be available for all hosts on the outside interface and not just the VPN clients.
To test your configuration, make sure that logging is set to debuging for the buffer and show the log while trying to ping the server in question. It will tell you exactly what is wrong.
