4 Create a Single-Member Cluster

After installing and configuring the Tru64 UNIX system, follow the directions in this chapter to make this system the first member of the cluster. Table 4-1 lists the tasks in order and references necessary information.

Note

If you are using Fibre Channel storagesets for the first member boot disk or the quorum disk, read the Fibre Channel chapter in the Cluster Hardware Configuration manual.

Table 4-1: Create Single-Member Cluster Tasks

Task	See
Gather the information needed to create a cluster.	Chapter 2
Install and configure Tru64 UNIX.	Chapter 3
Register the TruCluster Server license.	Section 4.1
Load the TruCluster Server subsets.	Section 4.2
Apply patches, if necessary.	Section 4.3
Run the `clu_create` command.	Section 4.4
Boot the first member's cluster boot disk.	Section 4.4 and Section 4.5
Make on-disk backup copies of important configuration files	Section 4.6
Perform a full backup of the single-member cluster.	Tru64 UNIX System Administration manual

4.1 Register the TruCluster Server Software License

The TruCluster Server kit includes a license Product Authorization Key (PAK). Use this PAK when registering a TruCluster Server license. (Section 1.5 describes the TruCluster Server license.) If you do not have a PAK, call your customer service representative.

For information on installing a license PAK, see the Tru64 UNIX Software License Management manual, lmf(8), and lmfsetup(8).

4.2 Load the TruCluster Server Subsets

To load the TruCluster Server kit, follow these steps:

Log in as superuser (root).

Mount the device or directory containing the TruCluster Server Software Version 5.1B kit.

Enter the setld -l command specifying the directory where the kit is located. For example, if you mount the CD-ROM on /mnt:
```
# setld -l /mnt/TruCluster/kit
 
```
You can choose one of the following subset installation options:
- All mandatory subsets only
- All mandatory and selected optional subsets
- All mandatory and all optional subsets
We recommend that you choose the "All mandatory and all optional subsets" option.
After you select an option, the installation procedure verifies that there is sufficient file system space before copying the subsets onto your system.

4.3 Apply Patches

Apply any patch kits you want to install on the system that will become the initial cluster member. When patching, both base operating system patches and cluster patches must be applied. This occurs only when the TruCluster Server subsets are present on the system.

If the system had patches applied before the cluster subsets were loaded, after loading the cluster subsets, you must patch the system again. We recommend you install the latest Tru64 UNIX and TruCluster Server patches.

Apply the patches before running the clu_create command.

For information about patching, see Patch Kit Installation Instructions. You can download patches from http://ftp.support.compaq.com/patches/.new/unix.shtml

4.4 Run the clu_create Command

The /usr/sbin/clu_create command creates the first member of the cluster from the Tru64 UNIX system.

The command checks that free space available on the /usr file system disk is at least three times the size of vmunix (typically, this product totals less than 100 MB). If less disk space is available, a message is displayed warning the user that there may not be enough space to build a customized cluster kernel. Even if the space is insufficient and a kernel cannot be built, the clu_create command should still complete successfully. You can then boot the cluster genvmunix from the boot disk and attempt to build a customized kernel on the single-member cluster.

Note

The clu_create command uses vdump and vrestore to populate the clusterwide root (/), /usr, and /var file systems. If any of these file systems on the Tru64 UNIX system has a network file system (NFS) file system mounted on it, and if that file system's NFS server is down, vdump will hang when trying to dump the file system. (If you run the automount daemon, remember that it mounts and unmounts NFS file systems, and the same potential for hangs exists.)
Before running clu_create, either unmount all NFS file systems or verify that they are accessible. Alternatively, you can reboot the Tru64 UNIX system before running clu_create to clean up any stale mounts. For systems that run the automount daemon, you can disable automounting by running the /sbin/init.d/nfsmount stop command before running clu_create.

Run the /usr/sbin/clu_create command. The command prompts for the information needed to create a single-member cluster. Answer the prompts using the information from the checklists in Appendix A. The command also provides online help for each question. To display the relevant help message, enter help or a question mark, ?, at a prompt.

4.4.1 Choosing the Cluster Interconnect

The clu_create command prompts for the type of cluster interconnect (LAN or Memory Channel), offering Memory Channel as a default when a Memory Channel adapter is installed:

If you specify Memory Channel as the cluster interconnect type, clu_create offers a default physical cluster interconnect interface of mchan0.

If you specify LAN, the clu_create command prompts you for a physical cluster interconnect device name. You have the following options:
- Specify the device name of a single network interface, such as tu0, ee0, or alt0.
- Iteratively specify the device names of multiple network interfaces. The clu_create command allows you to specify that these interfaces be configured in a NetRAIN virtual interface.
Note

If you specify the device name of an existing NetRAIN device (for example, one defined in the /etc/rc.config file), the clu_create command prompts you to confirm that you want to redefine this NetRAIN device as a physical cluster interconnect device. If you respond "yes," the clu_create command removes the definition of the NetRAIN device from the /etc/rc.config file and defines the device as the cluster interconnect device in the ics_ll_tcp stanza of the /etc/sysconfigtab file.

The clu_create command then creates an IP name for the physical cluster interconnect device of the form membermember-ID-icstcp0, and by default offers an IP address of 10.1.0.member-ID for this device.

See Table C-2 for a list of /etc/sysconfigtab attributes written by the clu_create command to define the cluster interconnect.

4.4.2 Tasks Performed by clu_create

The clu_create command performs the following tasks:

Sets up the clusterwide root (/), /usr, and /var file systems, and the first member's boot disk.

Configures a quorum disk (optional).

Builds a kernel with cluster components.
If the kernel build succeeds, clu_create copies the new kernel to the first member's boot disk. If the kernel build fails, clu_create displays warning messages but continues creating this first member. (You can boot the cluster genvmunix from the boot disk and attempt to build a kernel on the single-member cluster.)

Sets boot-related console variables: bootdef_dev and boot_reset; creates and sets boot_dev.
If clu_create can set the variables and if the kernel build was successful, clu_create offers to reboot the system for you. If you choose not to reboot the system at this time, use the information in Section 4.5 when you are ready to boot the system as a single-member cluster.
If clu_create cannot set the variables, halt the system, set the variables to the values specified in Section 2.7, and use the information in Section 4.5 to boot the system as a single-member cluster.

Note

If the first member's boot disk is accessed through HS controllers that are connected to dual SCSI or Fibre Channel buses and configured for multiple-bus failover, or the system is an AlphaServer 8200 or 8400, halt the system and see Section 2.7 for information on setting the bootdef_dev console variable.

The clu_create command writes a log file of the installation to /cluster/admin/clu_create.log. The log file contains all installation prompts, responses, and messages. Examine this log file for errors before booting the system as a single-member cluster. (Section D.1 contains a sample clu_create log file.)

4.5 Boot the System as a Single-Member Cluster

If you decided to boot the system yourself, examine the following items in the clu_create log file, /cluster/admin/clu_create.log, before booting the system as a single-member cluster:

Verify that the kernel for the new member built properly and was copied to the boot partition for this member. Look for the following kinds of messages in the clu_create.log file:

*** PERFORMING KERNEL BUILD ***
	Working....Tue May  8 15:54:11 EDT 2001
 
The new kernel is /sys/PEPICELLI/vmunix
  Finished running the doconfig program.
 
  The kernel build was successful and the new kernel
   has been copied to this member's boot disk.
  Restoring kernel build configuration.

Verify that the boot-related console variables are set to boot the cluster boot disk for the first member, not the Tru64 UNIX boot disk. In the log, look for the following kinds of messages:

Updating console variables
  Setting console variable 'bootdef_dev' to dsk10
  Setting console variable 'boot_dev' to dsk10
  Setting console variable 'boot_reset' to ON
  Saving console variables to non-volatile storage

If the new kernel is in place and the console variables are set correctly, reboot the system as a single-member cluster from its newly created cluster member boot disk. For example:
```
# shutdown -r now
 
```

If the kernel build did not succeed or the console variables could not be set, halt the system:
```
# shutdown -h now
 
```
Perform the following procedure:
1. If clu_create could not set the console variables, set the console variables according to the values specified in Section 2.7. (The remaining steps assume that the console variables are set correctly.)
2. If the kernel build succeeded, boot vmunix from the first member's cluster boot disk (make sure that bootdef_dev is set to the first member's cluster boot disk, not to the Tru64 UNIX disk):
```
>>> boot
 
```
3. If the kernel build did not succeed or you could not boot vmunix from the first member's boot disk, make sure that bootdef_dev is set to the first member's cluster boot disk (not to the Tru64 UNIX disk) and boot genvmunix:
```
>>> boot -file genvmunix
 
```
  When the system reaches multiuser mode, log in and attempt to build a kernel. If the build succeeds, copy (cp) the new kernel from /sys/HOSTNAME/vmunix to /vmunix. (If you move (mv) the kernel to /vmumix, you will overwrite the /vmunix context-dependent symbolic link (CDSL).) Then reboot the system so it will be running on its customized kernel.
```
# doconfig -c HOSTNAME
# cp /sys/HOSTNAME/vmunix /vmunix
# shutdown -r now
 
```

If you assigned a reserved network IP address to the default cluster alias, see Section 2.3.1 for information on advertising that alias address.

If you plan to use LSM, see Section 2.5.2.

When you boot this node as a single-member cluster, some access-related files, like /etc/ftpusers, are shared by all cluster members while other files, like /etc/securettys, are replaced by CDSLs that point to member-specific files. The reason is that files like ftpusers deal with user accounts (which are clusterwide entities), while files like securettys deal with member-specific information, in this case tty devices.

When first booted as a cluster member, the system runs the clu_check_config command to examine the configuration of several important cluster subsystems. Look at the clu_check_config log files in the /cluster/admin directory to verify that these subsystems are configured properly and operating correctly. If you discover any problems, read clu_check_config(8) so you know what tests the command performs. You can then run the command in verbose mode to display more information about why a subsystem failed the initial test. See the Tru64 UNIX System Administration and the TruCluster Server Cluster Administration manuals for information on configuring subsystems.

The following group of commands are useful for taking a quick look at the initial configuration of the cluster and some of its major subsystems:

# clu_get_info -full | more
# clu_quorum
# cfsmgr | more
# drdmgr `ls /etc/fdmns/* | grep dsk | sed 's/[a-z]$//' |\
   uniq | sort` | more
# cluamgr -s DEFAULTALIAS
# caa_stat

4.6 Make On-Disk Backup Copies of Important Configuration Files

Because cluster members rely on the information in the following files, we recommend that, after booting the first member of the cluster, you make on-disk copies of these files in case of inadvertent modification. For member-specific files, the examples assume that the member ID of the first member is 1 (memberid=1).

/etc/sysconfigtab.cluster:

# cp /etc/sysconfigtab.cluster /etc/sysconfigtab.cluster.sav

/etc/rc.config.common:

# cp /etc/rc.config.common /etc/rc.config.common.sav

/etc/sysconfigtab - This file is a CDSL whose target is:
```
../cluster/members/{memb}/boot_partition/etc/sysconfigtab
 
```
To make a backup copy, change directory to the first member's boot_partition/etc directory and make a copy of its sysconfigtab file.
For example:
```
# cd /cluster/members/member1/boot_partition/etc
# cp sysconfigtab sysconfigtab.sav
 
```

/etc/rc.config - This file is a CDSL whose target is:
```
../cluster/members/{memb}/etc/rc.config
 
```
To make a backup copy, change directory to the first member's etc directory and make a copy of its rc.config file. For example:
```
# cd /cluster/members/member1/etc
# cp rc.config rc.config.sav
 
```

In addition, we recommend that you perform a full backup of the single-member cluster.