Before we can use LVM, DRBD, or XFS on our servers, we must take some preliminary steps. We've never encountered a Linux system that is optimized for this kind of advanced usage directly after installation. In this recipe, we will modify several configuration files and even reboot the server.

We're trying to put each system in a standard state that we'll use for all future database servers. This means that LVM needs to ignore some devices to prevent disrupting DRBD, the initial RAM disks during boot should reflect this same allocation, and device performance shouldn't be lost between abstraction layers. We also need all of the tools that we'll use throughout this chapter.

This recipe will guarantee that these criteria are true, so be prepared!