Manual configuration of a Heartbeat cluster is often the most effective way of creating a reliable cluster that meets specific needs. Because of the extensive configurability of Heartbeat and the range of needs it can meet, it is not possible to document every possible scenario. To introduce several concepts of the Heartbeat configuration and demonstrate basic procedures, consider a real world example of an NFS file server. The goal is to create an NFS server that can be built with very low-cost parts and is as redundant as possible. For this, set up the following cluster:
Two machines that have redundant hardware
Data is mirrored on the disks of those machines with drbd
Only one machine at a time accesses and exports the data
Assign a special IP address to the computer for exporting the file system
Before starting with the cluster configuration, set up two nodes as described in Section 2.0, Installation and Setup. In addition to the system installation, both should have a data partition of the same size to setup drbd.
The configuration splits into two main parts. First, all the resources must be configured. After this, create a set of constraints that define the starting rules for those resources.
All the configuration data is written in XML. For convenience, the example relies on snippets that may be loaded into the cluster configuration individually.