Building an OpenStack cloud at production scales involves deploying and configuring a complex software stack on lots of hardware, usually using Puppet/Chef or other highly-automated systems to drive the process. If you want the resulting cloud to be robust and easy to scale, you need to know the hardware stack is ready: that BIOS settings and firmware levels are correct and that all base level interactions behave as expected. If this is not done right the first time, as the number of target systems grows, the painful process of manual checking and correcting quickly becomes unmanageable.
How do you figure this stuff out? How do you detect and maybe even fix these issues systematically, so that your deployments are trouble free? What happens between unboxing and Puppet/Chef? Luckily, there are tools to help!
This talk will discuss tools and techniques for pre-deployment hardware issue-detection and repair. We ll look at several species of tools, including:
- Bare-metal deployment systems like OpenStack ironic and Ubuntu MAAS, which incorporate device discovery and checkout features.
- What other solutions are coming to solve this problem?
You ll leave with a good understanding of potential issues, a workable method, and tools that will not only help you master pre-flight hardware checkout, but also facilitate different flavors of OpenStack cloud deployment.