Hello!
I am working on a small Clojure web application, and while still in the
early stages of development, I have been thinking, among other things,
about a deployment plan.
The production environment for the webapp will be a CentOS server (not
virtual), the webapp will be sitting behind an Apache gateway on the
same machine, the webapp will be deployed in a single instance (no load
balancing involved), and there will be no ssl support. These pieces of
information are probably irrelevant to what will follow, but I thought
I'd include them just in case, and maybe they can be summarized as: I
will not be using Heroku or Google app engine or Amazon.
I want to leverage the REPL (not Clojurescript browser REPL, but Clojure
server-side REPL) on prod for two things:
a. being able to check the status of the webapp by connecting to the
REPL and querying things, derefing atoms, etc.
b. hot-deploying small fixes without restarting the webapp by redefining
not-^:dynamic vars through the REPL
I am very excited about a. and b., but also somewhat unsure about what
the implications of using them are.
1. Some web servers, including Tomcat, have the functionality to unload
a webapp if it is getting no requests for a while, and reload it when a
request comes in. If I want to use such a web server, I need to make
sure this functionality is switched off, or otherwise it will interfere
with b., as live var redefinitions will be lost upon webapp reload, correct?
2. I suppose that the REPL is running in a separate thread. I feel
uncomfortable with having this thread started by the webapp, especially
having in mind that it is a long-running thread. The web server is
already managing a thread pool for handling incoming requests. Ideally,
this thread pool should also be the one responsible for starting the
REPL thread. Is there a web server that offers this functionality more
or less out of the box?
3. When developing a webapp, it is common to first start a REPL, and
then, from inside this REPL, start a web server. Would doing the same
thing for production deployment solve my concern 2.? I guess it should,
because then the REPL will be the "main" thread, and not some
long-running thread started by the webapp without the knowledge of the
web server's thread pool manager. Also, while this way of deploying
webapps is very different from how it is typically done in the Java
world, I find it beneficial, as one would typically want to do a bit of
a. immediately after startup, and have a more transparent startup
overall. Or are there any indications that this way of starting up is
not appropriate for prod?
Thank you!
Deyan
--
You received this message because you are subscribed to the Google
Groups "Clojure" group.
To post to this group, send email to clojure@googlegroups.com
Note that posts from new members are moderated - please be patient with your
first post.
To unsubscribe from this group, send email to
clojure+unsubscr...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/clojure?hl=en
---
You received this message because you are subscribed to the Google Groups "Clojure" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to clojure+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
