Learning Mcollective

The MCollective system administration toolset for parallel task execution is strategically intertwined with Puppet—so much so that Puppet Labs has encouraged users to deploy MCollective for ad-hoc control of their puppet hosts. Unfortunately, very little documentation on MCollective is available. Learning MCollective fills this critical gap.

With MCollective, you can build server orchestration or parallel job execution systems that enable real-time discovery of network resources. This book shows system administrators and operations engineers how to:

• Break free from complex naming conventions for hostnames as a means of identity, using a rich set of meta data provided by each machine
• Use meta data from Puppet, Chef, Facter, Ohai, or plugins you provide yourself
• Call remote agents, using simple command line tools
• Write custom reports about your infrastructure
• Manage packages, services, and other common components using agents
• Write simple RPC-style agents, clients, and web UIs in the easy-to-understand Ruby language

Jo Rhett is a network engineer and system administrator with more than 20 years of experience maintaining production CentOS/Red Hat, Debian, FreeBSD, Solaris and other Unix-based hosts and IP-based networks. He actively assists companies with IPv6 deployment.

Jo has been using, promoting, and enhancing configuration management systems for over 20 years. He implemented the complete package management functionality in Cfengine v2, before switching to Puppet for the last three years.