Resilient puppet infrastructure in the cloud (ENG)

Everyone can setup a puppetserver on a cloud instance. Is this puppet in the cloud? No. This is just a puppetserver in the cloud.

But what could be puppet in the cloud? One aspect of the cloud not typically covered by Puppet usage patterns is the coming and going of servers. This affects both agents and the puppet infrastructure itself. How can we effectively mange the classes assigned to spontaneously deployed instances? What happens if puppetserver or puppetdb instances vanish or scale up? Not all of this is documented in an easily understandable way.

This talk presents a way to create a resilient puppet server infrastructure in any cloud leveraging features of your cloud provider. We tackle typical obstacles and present a way to move from role assignment to a fact based role selection. All the introduced concepts are independent of cloud providers and can also be implemented on premises.