From: Steve B. <st...@pr...> - 2006-11-07 15:42:07
|
Thank you both for your swift and lucid responses. My situation is that I use tclhttpd to provide encrypted scripts to run on a (software) thin client. Different instances of the server supply tailored scripts, and talk to different MetaKit databases, for different user populations. Lots of custom code, extracted from databases on the server, put together on an individual basis for different users. But overall, the data rate is low. At present I do it by aliasing IP addresses onto the same physical server, but if there were a way of handling it without using up so much address space I would try to use it. So this is not a standard web server use at all, more a supplier of special applications. Kind regards, -- Steve Blinkhorn <st...@pr...> >>>Jan Kandziora said: > Am Dienstag, 7. November 2006 16:14 schrieb Steve Blinkhorn: > > > > What about multiple software servers, i.e. instances of tclhttpd, each > > responding to a different virtual domain but using the same port > > number, although on just one IP address? And when did this feature > > arrive in tclhttpd? (I am a long-term user, BTW). > > > Two or more processes cannot listen (to all) on the same port number on the > same machine. It's just the way TCP/IP works. > > To achieve something like you may think of, you have to build a "proxy" > listener, which dispatches the accepted connections to slave processes. > Hence, this usually (virtual domains, not just parallizing or load-balancing ) > involves the investigation of the "Host:" HTTP header field, which is the > only way to check which server should get the connection. > > So your question sounds pretty pointless to me. > |