I'm writing a server application in python. I understand that most
servers fork off a process for every incoming connection and then
renegotiate on a different port to free up the well-known server port. I
would rather do a multithread model than a forked process model because of
shared memory needs. I've heard that forking the process "automatically"
renegotiates the port somehow, but that doesn't make much sense to me.
More importantly, since I don't want to fork at all, how does one switch
to a different port? Is this a manual process? It seems it would be slow
to have so much round-trip communication before even establishing a
connection, and none of my network manuals say anything more than this
sort of thing "happens." Any tips?
----------------------------------
Nathan Clegg
nathan@islanddata.com
servers fork off a process for every incoming connection and then
renegotiate on a different port to free up the well-known server port. I
would rather do a multithread model than a forked process model because of
shared memory needs. I've heard that forking the process "automatically"
renegotiates the port somehow, but that doesn't make much sense to me.
More importantly, since I don't want to fork at all, how does one switch
to a different port? Is this a manual process? It seems it would be slow
to have so much round-trip communication before even establishing a
connection, and none of my network manuals say anything more than this
sort of thing "happens." Any tips?
----------------------------------
Nathan Clegg
nathan@islanddata.com