I've just found the problem - its the firewall. If I disable the firewall, then everything works fine from both local and remote hosts
> Hi > Using Xfree86 4.2.0 with latest cygwin, I see the following problem whcih > may relate to the xdmcp connection problems reported by several people > recently, although I am not using xdmcp. > > I set up the DISPLAY variable: > export DISPLAY=ntbox:0 > > I start the X server with: > XWin -ac -noreset& > > then a test client with: > xclock& > > the xclock appears as expected. Then I try a second client: > xeyes > > This fails to appear, eventually timing out with a "could not connect" > error. > This behaviour is consistent, whether the clients are started on the same > machine as the server, or on some remote linux box, for any clients (xclock > & xeyes are for example only). Note that, on the local NT box, if I use > "DISPLAY=localhost:0" or "DISPLAY=:0" then there is no problem and the > clients all start OK. > > It is as if the first tcp connection is accepted, but then the server stops > listening for further tcp requests. If that is the case, then it would > explain exactly the "black screen" syndrome experienced by the xdmcp users - > the first client is probably an "xsetroot -solid black" which succeeds, but > then all further requests are rejected. > > Can anyone else reproduce this problem? Is it an error in my local X > configuration, or is there anything else I can change to fix it? Steven