9    Cluster Installation and Administration

This chapter provides an overview of cluster installation (Section 9.1) and administration (Section 9.2).

9.1    Installation

TruCluster Server Version 5.1A supports three upgrade paths:

1. A full installation of TruCluster Server Version 5.1A.

2. A rolling upgrade from TruCluster Server Version 5.1. A rolling upgrade is a software upgrade of a cluster that is performed while the cluster is in operation. One member at a time is rolled and returned to operation while the cluster transparently maintains a mixed-version environment for the Tru64 UNIX base operating system, cluster, and Worldwide Language Support (WLS) software.

   The rolling upgrade procedure is used for three major tasks:

   1. Rolling from the prior version of the Tru64 UNIX base operating system and cluster software to the current version.

   2. Rolling patch kits into the cluster.

   3. Rolling New Hardware Delivery (NHD) kits into the Version 5.1A cluster.

   Rolling in a patch kit or an NHD kit uses the same procedure as rolling in a new release of the base operating system and cluster software. The clu_upgrade command controls rolling upgrades. See clu_upgrade(8) for a description of the clu_upgrade command.

3. Three upgrade procedures for those upgrading from the TruCluster Production Server Software or TruCluster Available Server Software Version 1.5 or Version 1.6 products. Two of these options use scripts that are specifically designed to facilitate the migration of storage from the old cluster (rz* style device names) to the new cluster (dsk* style device names). The Cluster Installation manual also describes an upgrade path for TruCluster Memory Channel Software products that have little or no shared storage.

One major difference when creating a TruCluster Server Version 5.x cluster is that you install Tru64 UNIX on only one system in the cluster. Because CFS creates shared clusterwide file systems, after a cluster is created, additional members boot into the cluster and have access to these files. (Before TruCluster Server Version 5.0, you had to install the base operating system on all cluster members, and there were no clusterwide file systems.)

For TruCluster Server, the initial creation of a cluster, the adding of members, and the removing of members are accomplished through three interactive installation scripts: clu_create, clu_add_member, and clu_delete_member. The scripts provide online help and write log files to the /cluster/admin directory.

The following list outlines the steps needed to install and create a new TruCluster Server cluster:

1. Using the information in the Cluster Hardware Configuration manual, configure the system and storage hardware and firmware.

2. Using AdvFS file systems, install Tru64 UNIX on a private disk on the system that will become the first cluster member.

3. Configure the Tru64 UNIX system, including network and time services. Load and configure the applications that you plan to use in the cluster.

4. Load the TruCluster Server license and software.

   Notes

   Each cluster member must have both a Tru64 UNIX license and a TruCluster Server license.

   If there are any patch or NHD kits available, you can install them after loading the cluster software but before running clu_create. Installing them before running clu_create means that you do not have to roll them into the cluster later.

5. Run the clu_create command to create the boot disk for the first cluster member, and to create and populate the clusterwide root (/), /usr, and /var AdvFS file systems.

6. Halt the Tru64 UNIX system and boot the disk containing the first member's cluster boot partition. As the system boots, it forms a single-member cluster and mounts the clusterwide root (/), /usr, and /var file systems.

7. Log in to the single-member cluster and run the clu_add_member command to add members to the cluster. Boot each new member before adding the next.

See the Cluster Installation manual for more information on installing TruCluster Server.

9.2    Administration

Having a clusterwide file namespace greatly simplifies cluster management. A cluster has just one copy of most system configuration files. For example, a cluster is managed as a single security domain through one /etc/group file and one /etc/passwd file.

User access to files is independent of which node a user is logged in on, and which node is serving the file. File permissions and access control lists (ACLs) are uniform across the cluster.

Audit logs are kept in a common location; each member's host name is appended to its log files to avoid confusion when tracking audit events.

In most cases, the fact that you are administering a cluster rather than a single system becomes apparent because of the occasional need to manage one of the following aspects of a TruCluster Server environment. In the following list, each area of administration is followed by one or more of the cluster-specific commands used to manage or monitor it. (You can use the SysMan Menu and SysMan Station GUIs to perform most command-line functions; you must use the cluster installation commands to install a cluster.)

• Cluster creation, which supports installing the initial cluster member, adding and deleting members, and querying the cluster configuration (clu_create, clu_add_member, clu_delete_member, and clu_check_config).

• Cluster application availability (CAA), which lets you define and manage highly available applications (caa_profile, caa_register, caa_unregister, caa_start, caa_stop, caa_relocate, and caa_stat).

• Cluster aliases, which provide a single system view from the network (cluamgr).

• Cluster quorum and votes, which determine what constitutes a valid cluster and membership in that cluster, and thereby control access to cluster resources (clu_quorum).

• Optional load-balancing of CFS servers (cfsmgr).

• Optional load-balancing of the device request dispatcher subsystem (drdmgr).

In addition to the previous items, there are some command-level exceptions to the Single System Image (SSI) model. (SSI means that, when possible, the cluster appears to the user like a single computer system.) For example, when you execute the wall command, the message is sent only to users who are logged in on the cluster member where the command executes. To send a message to all users who are logged in on all cluster members, use the wall -c command. The same logic applies to the shutdown command; you can shut down an individual member or the entire cluster.

See the Cluster Administration manual for more information on configuring and managing a TruCluster Server cluster.