On Fri, May 22, 2009 at 2:22 PM, Christopher Schultz <ch...@christopherschultz.net> wrote: > On 5/22/2009 2:01 PM, will trillich wrote: >> - we'd accidentally set up 'default gateway' on both >> >> This meant the system was confused as to where to send replies. > > I recently had a sever misconfiguration in a slightly different way: the > default route was set to use an eth device that wasn't physically > connected to anything, and the server's primary IP address was also > bound to that device. > > The symptoms were that no outgoing connections could be made > successfully, but that incoming connections worked just fine. > > This was on a Linux 2.4 kernel, which appears to route the replies > without a problem even with such an egregious misconfiguration: the > kernel appeared to be smart enough to send replies to the "right place" > regardless of the stupidity of the configuration. > > In what environment did this default route misconfiguration cause > unreliable responses?
Windo~1 2003 Enterprise. It was really weird. For hours at a time, all was well. Packets came, packets went. Pages were browsed. Then -- poof! -- no traffic at all. We were originally thinking it was another computer with the same IP address that had no server, getting all the traffic. Nope. Just too many default gateways. :) -- will trillich "Our only real economic security lies in our power to meet human needs." -- S.Covey, the 8th Habit --------------------------------------------------------------------- To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org For additional commands, e-mail: users-h...@tomcat.apache.org