Problem
TCP Connections over the WAN may take a very long time or timeout completely.
Cause
This can be caused by intermediate proxy servers or firewalls require reverse DNS authentication before passing TCP packets.
As an example, a SCO server is inside a firewall and configured to accept inbound telnet sessions.
The DNS is configured correctly (see TA 112741 to test) and any connection from the internal subnet occurs correctly, however a user trying to telnet from a home computer either is unable to connect, or sees a consistent delay (24-35 seconds) that is consistent with a DNS failure.
This can be caused because the intermediate firewall is configured to do reverse DNS lookups on all telnet connection attempts, and the home computer does not have a valid reverse DNS entry that is available to the firewall.
Depending on the firewall product and how it is configured it may either wait for a DNS timeout and then forward the packet, wait for a DNS failure (typically 60 seconds) and forward the packet, or completely refuse to forward the packet to the internal SCO server.
Solution
This must be resolved by changing the configuration of the intermediate firewall or proxy servers, and/or ensuring that all TCP clients have valid reverse DNS entries. The firewall or proxy server may be able to be configured to:
1. Check reverse DNS on all connections, or on connections by
protocol.2. Do reverse DNS lookups on all connection attempts for enhanced
logging.3. Block all connection attempts without valid reverse DNS.Any of these can cause a DNS failure before the client's packets ever reach the SCO server.
See also
TA112741 - What are the tools available to run diagnostics on my DNS configuration?
TA 112741