> Well, urlgrabber (the Package) has a "timeout" option which may be > getting used from yum and not from the "urlgrabber" executable. For a > proper comparison, we should make sure that the function is getting > called the same way. I suspect that's why you're seeing the > difference, though. > > Now, assuming that's it, the timeout error is just that: a timeout > error. This is probably not a bug in the code, although it could be. > URLGrabber has been Thoroughly tested with NAT. My home desktop is > behind a NAT linux box, so I routinely use it that way. I see that > the timeout option doesn't appear in the unit tests, though, so it's > possible that there's a conflict that got missed. > > Anyway, are we sure that the timeout isn't a "legitimate" timeout? I bet I know what it is. I bet the session is remaining open and being killed by the netgear for inactivity/use. Matthew, try putting: net.ipv4.tcp_keepalive_time = 300 in the users /etc/sysctl.conf file and restarting their machine. see if the problem goes away. -sv
