Quote:> There is a proxy server that I need to have all of my web servers'
> HTTP (specifically port 80 and 443) traffic to go through (both
> inbound and outbound). How do I go about doing this?
1) You set up a proxy machine. You can use squid for this but other
programs are available as well. (This is usually called an 'inverse
proxy' because the proxy is on the server end rather than the client.)
2) You configure the proxy to refer requests to your server(s).
3) If necessary for security, you configure your servers to not accept
requests from sources other than the proxies.
4) Rather than advertising the address of the server to the public,
you advertise the address of the proxy. So if the web server handles
'www.example.com', you make 'www.example.com' resolve to the IP
address of the proxy.
5) You fix any problems this causes the web server. For example, if it
advertises its own host name or IP address (say in links on the
pages), it needs to be changed to advertise the proxy. Logging, host-
based authentication, or IP source based authentication, blacklisting,
or rate-limiting may break as all connections are seen to come from
the proxy's address.
This assumes you want a non-transparent proxy, which is generally
simpler to setup and doesn't require any special network layout.
> I have a mixture of Linux and Windows web servers that need to use the
> proxy. All of them are in different networks (different hosts).
> When any person makes the http request to a web page for one of these
> web servers' domains, I need the request to go through the proxy. The
> proxy will pass the request to the web server. The web server will
> generate the response and send the response to the proxy and the proxy
> will pass the response on to the end user.
Squid can do this.