Hacker News Reader

Comment by dan_manges | Hacker News Reader

You can use Puppet without doing all of that. At Braintree, we use Capistrano to upload a tarball of our Puppet scripts and run the puppet command.

sudo puppet --templatedir $HOME/puppet/templates --factpath $HOME/puppet/facts puppet/puppet.pp

View on HN

Let me endorse both of the above posts. Having now spent many months tinkering with Puppet, I wish to god I had fully understood the value of not using any of Puppet's server features but had instead just shipped the manifests via some other method and applied them locally.

If you need a super-performant version of that strategy (which you probably don't) try googling up Twitter's "murder" project. (Not as violent as it sounds! It's the crow kind of murder.)

Until I read this thread, I didn't even think of that as an option -- everything in the Puppet documentation, and the community, and in what they offer as far as training courses, indicates that Puppet's client-and-server are the light and the way to rightness.

I'll have to take a second look at the Puppet manifest system coupled with Capistrano (which I really like), but it still irks me that there isn't one tool to handle configuration management across a large number of servers.

Any solid, good Howto's; we're using puppet as well and had the same issues. We use Capistrano + Puppet + some custom scripts; we would like to get rid of the latter.

Is there any howto for setting up a completely clean/empty Linux (Ubuntu / Debian) with RVM as Ruby installer? But thanks for this; at least now I know there is a solution :)

Reply on news.ycombinator.com