Notes from the NH Ruby / Rails Group, 16 Nov 2009: RVM and EC2
Ted Roche
tedroche at tedroche.com
Wed Nov 18 14:04:45 EST 2009
Brian Turnbull and Nick Plante each had excellent presentations for the
New Hampshire Ruby Rails Group's first meeting on the third Monday of
the month, our new "regular meeting day." (We won't be having a December
meeting, however; have a happy Kwanzaa, Hanukkah, Christmas, Saturnalia,
Winter Solstice or non-denominational generic shopping holiday of your
choice.)
Brian was up first, and talked about rvm, the Ruby Version Manager
(slides here: http://www.slideshare.net/bturnbull/rvm-nhruby-nov-2009).
RVM doesn't just allow you to have multiple versions of ruby on your
machine; you've always been able to do that with side-by-side installs
or linking or aliases or other kludges. RVM can be found at
http://rvm.beginrescueend.com and can be installed as a Gem, from the
Git repository or as a tarball. It's not Ruby code, actually, but shell
scripts, and it allows not just switching of the current active version
of ruby, but much more interestingly, the entire set of gems associated
with a version! Multiple gemsets can be assigned an alias that lets you
group your gems by function or project or client. RVM also provides a
facility to dump the list of gems from one version or gemset and load it
into another, allowing you to test out the migration of an existing
system to a new version with much less hassle. Finally, rvm also lets
you invoke multiple versions serially, passing them a command like "rvm
1.86, 1.8.7, 1.9.1 rake spec" where rvm will run all your tests on
multiple versions and report results, even in JSON if specified. This
could be the core of a really simple Continuous Integration (CI) server
in a single line of code!
Brian cautioned that rvm is early in its development cycle (current as
of last night was 0.0.79) and changes are coming fast and furious, but
it's worth taking some time to keep track of this very interesting
development!
Brian also mentioned that rvm has some extensions to work with Bundler.
Nick mentioned that Bundler, a project from Yehuda Katz (core
contributor to both Ruby and jQuery) , is slated to become the default
means of managing gems in Rails 3.0. Here's a good link to learn more:
http://yehudakatz.com/2009/11/03/using-the-new-gem-bundler-today/
Nick Plante was up next, with his presentation on EC2 for Rails
Development (http://zapnap.github.com/presentations/ec2-rubber/#0). Nick
talked about the evolution of deployment of Rails (or Rack) solutions,
moving from very painful to much easier with the introduction of
Capistrano (http://capify.org/). He reviewed how simple it was to deploy
an application with Capistrano, and then talked about the power of the
Amazon Web Services (AWS) http://aws.amazon.com/ and Cloud Computing in
general. Capistrano is focused on general deployment to a server, but
the new add-on, Rubber (http://github.com/wr0ngway/rubber), extends
that to work with elastic computing resources like AWS' Elastic Compute
Cloud (EC2). We walked through the basics of the
install-configure-deploy scenario and then "Insert Demo Here." - he did
it. Defying the Demo Gods (who had already tortured Brian a bit), Nick
grabbed a copy of RedMine (http://www.redmine.org/) from GitHub, did the
couple of minor configuration items needed to make it work locally
(creating the session key, copying the sample database config to the
live one, setting it to use SQLite3 locally and running db:migrate) and
then set it up for deployment: first invoking Capistrano (doing some
configuring) and then Vulcanizing (how else would you make rubber) the
installation, then deploying it to a brand-new AWS instance. The steps
flew by lightening-fast: the Amazon instance was up and running a minute
after being invoked, then took a couple of minutes to install and
configure Apache, Passenger, MySQL, the Redmine application, and sftp
the development directory files to the instance. Within a few minutes,
Nick could browse the machine and see the running application as well as
ssh into the box and make any changes. There are innumerable options, of
course, but they are all accessible by reviewing the configuration files
and the source to tune it to your particular needs.
While we waited for the installation to complete, I brought up some of
the customer's concerns and practitioner's insights from last Thursday's
MonadLUG meeting on Cloud Computing and we talked about how those
applied. I expressed concerns about how persistent data storage worked,
since the cloud images were considered ephemeral. Nick pointed out that
the instance, while running could have a local database and it could be
in a master-slave replication arrangement with either another instance
(within the AWS, there aren't bandwidth charges) or a hosted or local
machine. Also mentioned was the Amazon Relation Database Service (RDS),
which can provide a MySQL database instance with backups. So, there are
lots of options.
Thanks to Brian and Nick for excellent presentations as well as
organizing the meeting and bringing the pizza, to Tim and the folks at
RMC Research for providing the excellent facilities and to all eleven
who attended and participated in the meeting.
Remember, no Ruby meeting meeting in December. Stay tuned for a January
announcement.
--
Ted Roche
Ted Roche& Associates, LLC
http://www.tedroche.com
More information about the gnhlug-discuss
mailing list