Puppet: configuration management made easy

By Sam Varghese
Tuesday, 28 April 2009 07:06

Technology

Like any open source project, Puppet is reasonably well documented but there are some areas where documentation is written by developers and occasionally a bit light on, says Turnbull. "I started adding documentation, and next thing I know, Luke Kanies, the primary developer of Puppet, was saying, 'look, there's this wiki over here, can you look after that?'"

As time went on, Turnbull found himself managing a lot of tickets and a lot of the release-related things. He was nominated as the release manager for the project. "And beyond that, it occurred to me that this would be a viable topic to write about, so I wrote a book about it as well," he says.

Puppet can be used to manage small and large enterprise assets. It is principally UNIX-based and runs on GNU/Linux, OSX, HPUX, AIX, Solaris and all the BSDs. It is designed to  be able to automate the process of configuring a host.

"It doesn't do PXE boot or anything like that - the next step after you've done your initial install of your host is to add a Puppet client and then to configure your hosts," says Turnbull. "It does this through a master in the environment which contains all your  configuration and it knows the particular host and the particular questions posed, so it can answer them in a particular way.

"For example, on web servers or print servers, each of your hosts has a client on it. The client connects to the master and says 'I'm host number 1, tell me what my configuration should be'. And the master says, 'host number 1, you're a web server, so you need to have Apache, you need to have Squid, you need to have all the relevant packages, files, configuration files, users', and it instantiates that configuration on the host."

CONTINUED