Ian Davey wrote: > > I'll think about it some more and try and file a more useful bug report. > It'll be difficult to reproduce though I'd imagine. How about setting up a unix machine with inetd configured to run "cat >/dev/null" or some other non-terminating process on a connection to port 80 (or any >other port, if you make the port explicit in the URL)? This should at least provide a >first-pass simulation of a hung server, although it may not have the exact same >behavior as the hung server you were dealing with. Stuart.
