After moving a Linux server serving as a wireless gateway/router to another ISP, I noticed a problem with NNTP. When attempting to communicate NNTP through the Linux server, there are long delays that often result in timeouts. For example, when using the newsreader pan, I get the following errors in the log file, in reverse chronological order: Connect Failure NNTP handshake failed: Error reading from socket. Handshake failed: Error reading from socket. NNTP handshake failed: Error reading from socket. Timed out waiting to read from the server. The Linux server is fed with a T1 from a different source. Before changing over, there was no problem with NNTP. The only other difference that would seem to matter is the reconfiguration of the server. I only took some services off line (qmail for example) and reconfigured the routing. All other traffic through the server works well, with the exception of dhcp which I haven't resolved yet. I suspect a QoS control issue from upstream. Is it possible that NNTP requests are sitting in a queue somewhere waiting for a the "right" time to dequeue? If so, would there be something I could do with the TOS that might help? Or Am I on the wrong track? TIA, David Koski david@KosmosIsland.com
