Typically each hunchentoot acceptor listens on a single port, and typically serves a single domain. A typical configuration is to have a frontend webserver like Nginx to route requests to the appropriate port.
This is hard to manage when the number of domains keep increasing (which is typical for small companies trying to iterate on different products). hunchentoot-multi-acceptor simplifies the configurations of such systems.
As an example, I personally run tdrhq.com (my personal website) and jipr.io (a project I'm working on.) These have two different hunchentoot acceptors:
(defvar *tdrhq-acceptor* (make-instance 'hunchentoot:easy-acceptor :port nil :name 'tdrhq)) (defvar *jipr-acceptor* (make-instance 'hunchentoot:easy-acceptor :port nil :name 'jipr))
These definitions could be in different packages and systems, it
doesn't matter. Also the
:port argument is useful when developing
locally and want to test only one of the websites, but it won't be
used by multi-acceptor.
Now, we have a multi-acceptor setup as follows:
(defparameter *multi-acceptor* (make-instance 'hunchentoot-multi-acceptor:multi-acceptor :port 4001 :name 'multi-acceptor)) (hunchentoot-multi-acceptor:add-acceptor *multi-acceptor* "tdrhq.com" *tdrhq-acceptor*) (hunchentoot-multi-acceptor:add-acceptor *multi-acceptor* "www.jipr.io" *jipr-acceptor*) (hunchentoot:start *multi-acceptor*)
At this point, you can configure your Nginx frontend for both websites
to point to
localhost:4001. Make sure the "Host" header is
appropriately set. (
proxy_set_header Host $host:$server_port). After
that hunchentoot-multi-acceptor will take care of routing to the
appropriate acceptor. Only one port will be opened.
Apache License, Version 2.0