2.3. Setting Up the Nodes

2.3.1. Installing the Basic Cluster Hardware

Nodes must provide the CPU processing power and memory required by applications.

In addition, nodes must be able to accommodate the SCSI or Fibre Channel adapters, network interfaces, and serial ports that the hardware configuration requires. Systems have a limited number of pre-installed serial and network ports and PCI expansion slots. Table 2-10 helps determine how much capacity the node systems employed require.

Most systems come with at least one serial port. If a system has graphics display capability, it is possible to use the serial console port for a power switch connection. To expand your serial port capacity, use multi-port serial PCI cards. For multiple-node clusters, use a network power switch.

Also, ensure that local system disks are not on the same SCSI bus as the shared disks. For example, use two-channel SCSI adapters, such as the Adaptec 39160-series cards, and put the internal devices on one channel and the shared disks on the other channel. Using multiple SCSI cards is also possible.

Refer to the system documentation supplied by the vendor for detailed installation information. Refer to Appendix A Supplementary Hardware Information for hardware-specific information about using host bus adapters in a cluster.

2.3.2. Shared Storage considerations

In a cluster, shared disks can be used to store cluster service data. Because this storage must be available to all nodes running the cluster service configured to use the storage, it cannot be located on disks that depend on the availability of any one node.

There are some factors to consider when setting up shared disk storage in a cluster:

• It is recommended to use a clustered filesystem such as Red Hat GFS to configure Red Hat Cluster Manager storage resources, as it offers shared storage that is suited for high-availability cluster services. For more information about installing and configuring Red Hat GFS, refer to the Red Hat GFS Administrator's Guide.

• Whether you are using Red Hat GFS, local, or remote (NFS) storage, it is strongly recommended that you connect any storage systems or enclosures to redundant UPS systems for a highly-available source of power. Refer to Section 2.5.3 Configuring UPS Systems for more information.

• The use of software RAID or Logical Volume Management (LVM) for shared storage is not supported. This is because these products do not coordinate access to shared storage from multiple hosts. Software RAID or LVM may be used on non-shared storage on cluster nodes (for example, boot and system partitions, and other file systems which are not associated with any cluster services).

  An exception to this rule is CLVM, the daemon and library that supports clustering of LVM2. CLVM allows administrators to configure shared storage for use as a resource in cluster services when used in conjunction with the CMAN cluster manager and the Distributed Lock Manager (dlm) mechanism for prevention of simultaneous node access to data and possible corruption.

• For remote filesystems such as NFS, you may use gigabit Ethernet for improved bandwidth over 10/100 Ethernet connections. Consider redundant links or channel bonding for improved remote file system availability. Refer to Section 2.5.1 Configuring Ethernet Channel Bonding for more information.

• Multi-initiator SCSI configurations are not supported due to the difficulty in obtaining proper bus termination. Refer to Appendix A Supplementary Hardware Information for more information about configuring attached storage.

• A shared partition can be used by only one cluster service.

• Do not include any file systems used as a resource for a cluster service in the node's local /etc/fstab files, because the cluster software must control the mounting and unmounting of service file systems.

• For optimal performance of shared file systems, make sure to specify a 4 KB block size with the mke2fs -b command. A smaller block size can cause long fsck times. Refer to Section 2.5.3.2 Creating File Systems.

After setting up the shared disk storage hardware, partition the disks and create file systems on the partitions. Refer to Section 2.5.3.1 Partitioning Disks, and Section 2.5.3.2 Creating File Systems for more information on configuring disks.

2.3.3. Setting Up a Console Switch

Although a console switch is not required for cluster operation, it can be used to facilitate node management and eliminate the need for separate monitors, mouses, and keyboards for each cluster node. There are several types of console switches.

For example, a terminal server enables connection to serial consoles and management of many nodes from a remote location. For a low-cost alternative, use a KVM (keyboard, video, and mouse) switch, which enables multiple nodes to share one keyboard, monitor, and mouse. A KVM switch is suitable for configurations in which GUI access to perform system management tasks is preferred.

Set up the console switch according to the documentation provided by the vendor.

After the console switch has been set up, connect it to each cluster node. The cables used depend on the type of console switch. For example, a Cyclades terminal server uses RJ45 to DB9 crossover cables to connect a serial port on each node to the terminal server.

2.3.4. Setting Up a Network Switch or Hub

A network switch or hub, although not required for operating a two-node cluster, can be used to facilitate cluster and client system network operations. Clusters of more than two nodes require a switch or hub.

Set up a network switch or hub according to the documentation provided by the vendor.

After setting up the network switch or hub, connect it to each node by using conventional network cables. A terminal server, if used, is connected to the network switch or hub through a network cable.