How to configure a firewall for domains and trusts

Summary

This article describes how to configure a firewall for domains and trusts.

Note: Not all the ports that are listed in the tables here are required in all scenarios. For example, if the firewall separates members and DCs, you don't have to open the FRS or DFSR ports. Also, if you know that no clients use LDAP with SSL/TLS, you don't have to open ports 636 and 3269.

More Information

To establish a domain trust or a security channel across a firewall, the following ports must be opened. Be aware that there may be hosts functioning with both client and server roles on both sides of the firewall. Therefore, ports rules may have to be mirrored.

Windows NT

Click here to show/hide solution
Or, you can establish a trust through the Point-to-Point Tunneling Protocol (PPTP) compulsory tunnel. This limits the number of ports that the firewall has to open. For PPTP, the following ports must be enabled.  
Client PortsServer PortProtocol
1024-65535/TCP1723/TCPPPTP
In addition, you would have to enable IP PROTOCOL 47 (GRE).

Note When you add permissions to a resource on a trusting domain for users in a trusted domain, there are some differences between the Windows 2000 and Windows NT 4.0 behavior. If the computer cannot display a list of the remote domain's users, consider the following behavior:
  • Windows NT 4.0 tries to resolve manually-typed names by contacting the PDC for the remote user's domain (UDP 138). If that communication fails, a Windows NT 4.0-based computer contacts its own PDC, and then asks for resolution of the name.
  • Windows 2000 and Windows Server 2003 also try to contact the remote user's PDC for resolution over UDP 138. However, they do not rely on using their own PDC. Make sure that all Windows 2000-based member servers and Windows Server 2003-based member servers that will be granting access to resources have UDP 138 connectivity to the remote PDC.
Properties

Article ID: 179442 - Last Review: Aug 11, 2016 - Revision: 1

Feedback