On Friday, August 30, 2002, Peter Palmreuther wrote... MDP>> I think that the client side Socket connection is left open and MDP>> until it is flagged as closed TB won't open a new one.
> But how to tell? 'netstat' didn't show me an open socket neither on > my client nor on my server when I wan into that trouble. Netstat might not show an internal component having an open connection if Windows has terminated the connection, and the component didn't acknowledge it. This will still flag the socket in TB! as open, but windows will record it as closed. That'd make a bug in the component itself not properly acknowledging, and switching it's active state to false. (Or that global boolean theory I proposed). [snip] MDP>> That would be an interesting exercise. > And it was done. If I shut down the daemon listening to port 110 the normal > TCP mechanisms stepped in and the connection was terminated with an error > message, as it is supposed to be. This wasn't the kind of connection termination I was thinking of. The daemon in this case drops the connections safely, and reports valid terminations... I was meaning more along the lines of unplugging the PC (if you don't mind doing that kind of thing) from the switch/dsl/modem/etc. But even in my mind, I have a feeling this won't force the error to occur because Windows will report a it couldn't connect properly, or the connection terminated. > I can only guess this is an error in socket component used in The > Bat!, but of course I can't prove it. It could be a bug in the socket component itself (the one they use) which is not accepting certain types of disconnects. -- Jonathan Angliss ([EMAIL PROTECTED]) ________________________________________________ Current version is 1.61 | "Using TBUDL" information: http://www.silverstones.com/thebat/TBUDLInfo.html