4 Upgrade Installation

The TruCluster upgrade procedures assume that you are performing an update installation of the DIGITAL UNIX operating system, and thereby preserving existing TruCluster configuration and database files. If you perform a full installation of the DIGITAL UNIX operating system, you must perform a full installation of the TruCluster product.

Note

Some versions of TruCluster products are supported on more than one version of the DIGITAL UNIX operating system. If you are updating only the operating system, see Section 4.6.

Table 4-1 summarizes the upgrade installation tasks. It lists the tasks in order, indicates the TruCluster products to which each task applies, and provides pointers to necessary information. The column headings R and S refer to rolling and simultaneous upgrades. The section titles in this chapter also use the R convention to indicate when a section applies only to a rolling upgrade.

Both full and upgrade TruCluster installations use the same installation script. Because both installation types share several steps, Table 4-1 often refers to sections in Chapter 3.

Table 4-1: Upgrade Installation Tasks

Task Production Server Available Server See:

R S R S

Read the introductory chapters. X X X X Chapters 1 and 2

Read the rolling upgrade restrictions. X X Section 4.1

Perform preinstallation tasks. X X X X Table 2-1

Load the TruCluster kit. The installation procedure starts automatically when you load the kit. X X X X Section 3.1

Specify the IP name and address (CLUSTER_NET) of the cluster interconnect (PS) or the member network interface (AS). X X X X Section 2.1 and Section 3.2

Specify an ASE identifier (ASE_ID). X X Section 3.3

Decide whether to run the ASE logger daemon on this system. X X X X Section 3.4

Initialize the ASE database. X X X X Section 4.2

Select a kernel configuration file. X X X X Section 3.6

Build and install a new kernel. X X X X Section 3.8

Add host entries to /etc/hosts. X X X X Section 3.9

Enable new DLM interfaces and reboot the system. X Section 3.10

Reboot the system. X X X Section 3.11

Return the member to the ASE. X X Section 4.3

After all systems are upgraded, if using a saved ASE database, return the member to the ASE. X X Section 4.3

After all systems are upgraded, if not using a saved ASE database, populate the member list in the ASE database. X X Section 3.12

Specify a tie-breaker disk (CNX_DISK) (only for a two-system, virtual-hub cluster). X X Section 3.14

After all systems are upgraded, enable new DLM interfaces and reboot each system. X Section 4.4

After all systems are installed, create consistent device special files for an ASE highly available tape service. X X X X Section 3.13

After all systems are upgraded, restart ASE services. X X Section 4.3

After all systems are upgraded, use asemgr to turn on new features. X X Section 4.5

After all systems are upgraded, run clu_ivp to verify the installation. X X X X Chapter 5

Task	Production Server	Available Server	See:
Read the introductory chapters.	X	X	X	X	Chapters 1 and 2
Read the rolling upgrade restrictions.	X		X		Section 4.1
Perform preinstallation tasks.	X	X	X	X	Table 2-1
Load the TruCluster kit. The installation procedure starts automatically when you load the kit.	X	X	X	X	Section 3.1
Specify the IP name and address (`CLUSTER_NET)` of the cluster interconnect (PS) or the member network interface (AS).	X	X	X	X	Section 2.1 and Section 3.2
Specify an ASE identifier (`ASE_ID`).	X	X			Section 3.3
Decide whether to run the ASE logger daemon on this system.	X	X	X	X	Section 3.4
Initialize the ASE database.	X	X	X	X	Section 4.2
Select a kernel configuration file.	X	X	X	X	Section 3.6
Build and install a new kernel.	X	X	X	X	Section 3.8
Add host entries to `/etc/hosts`.	X	X	X	X	Section 3.9
Enable new DLM interfaces and reboot the system.		X			Section 3.10
Reboot the system.	X		X	X	Section 3.11
Return the member to the ASE.	X		X		Section 4.3
After all systems are upgraded, if using a saved ASE database, return the member to the ASE.		X		X	Section 4.3
After all systems are upgraded, if not using a saved ASE database, populate the member list in the ASE database.		X		X	Section 3.12
Specify a tie-breaker disk (`CNX_DISK`) (only for a two-system, virtual-hub cluster).	X	X			Section 3.14
After all systems are upgraded, enable new DLM interfaces and reboot each system.	X				Section 4.4
After all systems are installed, create consistent device special files for an ASE highly available tape service.	X	X	X	X	Section 3.13
After all systems are upgraded, restart ASE services.		X		X	Section 4.3
After all systems are upgraded, use `asemgr` to turn on new features.	X		X		Section 4.5
After all systems are upgraded, run `clu_ivp` to verify the installation.	X	X	X	X	Chapter 5

4.1 Rolling Upgrade Restrictions (PS-R, AS-R)

Before performing a rolling upgrade, note the following restrictions:

As a general note, do not attempt to use new features until all members are upgraded. For example, do not enable new DLM interfaces (Section 4.4) until the entire cluster is upgraded.

The extended UIDs/GIDs feature is disabled by default in the DIGITAL UNIX operating system. See the DIGITAL UNIX documentation for information on enabling and using this feature.

During a rolling upgrade, you should not run the Cluster Monitor (cmon) utility until all cluster members are upgraded. If you must use the Cluster Monitor utility during a rolling upgrade, apply Patch TCR141-006 to all Version 1.4 (Available Server) or Version 1.4A (Available Server or Production Server) member systems before starting the rolling upgrade. Follow the installation instructions that accompany the patch.

Note

If you are installing this patch on TruCluster Available Server Software Version 1.4, copy the patched tractd file to /usr/opt/TCR140/sbin/tractd. The TCR141 subdirectory included in the target file path in the patch's installation instructions is correct only for a Version 1.4A installation.

During a rolling upgrade, do not create or modify an ASE service that uses the Logical Storage Manager (LSM) on a system that is running DIGITAL UNIX Version 4.0D. Creating or modifying LSM services on a system running DIGITAL UNIX Version 4.0D may result in these services not being properly relocated. After upgrading all cluster members to DIGITAL UNIX Version 4.0D, you can create or modify LSM services on any member.

When you delete an ASE service that uses Advanced File System (AdvFS) filesets or LSM volumes, the storage configuration information is retained on the system. Therefore, you should delete AdvFS or LSM services on a system that has already been upgraded to DIGITAL UNIX Version 4.0D to avoid the possibility of overwriting the information when installing the operating system during the rolling upgrade procedure.

During a rolling upgrade, you cannot use the audit feature on file system objects (for example, files or directories) that are located on shared devices.

The MEMORY CHANNEL API library will not operate in a cluster in which member systems are running different versions of the TruCluster software.
When a member boots into a Production Server cluster, an attempt is made to initialize the MEMORY CHANNEL API (because the installation procedure sets IMC_AUTO_INIT=1 in /etc/rc.config). This initialization is transparent when all members of the cluster are running the same version of the Production Server software. However, when a Production Server Version 1.5 system boots into a cluster where one or more members are running earlier versions, there is a MEMORY CHANNEL API version incompatibility.
In this case, the initialization procedure generates imc_init: and mcs_IMC_init: error messages indicating a version incompatibility, and the MEMORY CHANNEL API initialization fails. Because the Production Server does not rely on the MEMORY CHANNEL API for cluster operations, you can usually ignore these messages. To initialize the MEMORY CHANNEL API after all members are upgraded to Version 1.5, run the /usr/sbin/imc_init command on each member.
However, this initialization behavior has some important consequences for any highly available services that use the MEMORY CHANNEL API.

Note

The following restriction applies only to customers who have created a highly available service that uses the MEMORY CHANNEL API library routines.

The rolling upgrade process consists of booting newly installed Version 1.5 members into an existing cluster. When each of these members is booted, its MEMORY CHANNEL API library initialization fails as noted earlier, leaving only the remaining Version 1.4A members capable of running any highly available services that use the MEMORY CHANNEL API library routines. At some point in the rolling upgrade, you must stop these services in order to complete the upgrade of the last Version 1.4A system to Version 1.5.
To minimize the downtime of these services, perform the following procedure:
1. Use the asemgr utility to relocate any services using the MEMORY CHANNEL API library routines to a Version 1.4A system.
2. Upgrade all other systems to Version 1.5 of the Production Server.
3. On the remaining Version 1.4A system, stop all services that use the MEMORY CHANNEL API library.
4. Halt the Version 1.4A system.
5. Run the /usr/sbin/imc_init command on all Version 1.5 systems.
6. Use the asemgr utility on a Version 1.5 system to restart the service.
7. Upgrade the remaining Version 1.4A system to Version 1.5.

The upgrade procedure is not reversible. To return to an earlier version of the operating system and TruCluster software after DIGITAL UNIX Version 4.0D and TruCluster software are installed, you must deinstall the TruCluster software subsets, and then reinstall the operating system and the earlier version of TruCluster software.

Note

DIGITAL recommends that you complete the rolling upgrade procedure as quickly as possible and minimize any hardware and software changes during the upgrade. For example, if you want to upgrade the member systems and also remove a member system, delete the member system before you perform the upgrade. If you want to add a member system, perform the upgrade and then add the new member.

4.2 Initialize the ASE Database (PS, AS)

If you did not save the ASE database when deinstalling the TruCluster subsets, the installation procedure automatically creates and initializes a new ASE database, /var/ase/config/asecdb. However, if the installation procedure detects an existing ASE database, it asks whether you want to use this database:

An old ASE database file has been found. Do you want to use this (y/n):

If you answer n, the installation procedure deletes the database and creates a new one.

For a rolling upgrade, answer n. You want the existing ASE to repopulate this system when it is returned to the ASE.

For a simultaneous upgrade, answer y if you saved the ASE database when deinstalling the TruCluster subsets and want to reuse that database. After all systems are upgraded, you must run the asemgr utility to reinitialize the member list before the ASE services in the saved database can be started. The following message is displayed to remind you to reinitialize the member list before starting ASE services:
```
Note: You will need to run asemgr to reinitialize the member list before the
saved ASE services can be started.
```

4.3 Return the System to Its ASE (PS, AS)

To return the member to the ASE, follow the steps for your type of upgrade.

For a rolling upgrade:

Run the asemgr utility on an existing ASE member system and add the upgraded system to the ASE.

If the upgraded member's name was removed from the member list in any ASE service's automatic service placement (ASP) policy when you deleted the system from the ASE, use the asemgr utility to stop the ASE service and add the member back to the ASP policy member list.

For a simultaneous upgrade using a saved copy of the ASE database:

Check that all systems are upgraded.

Run the asemgr utility on one member system only in each ASE to reinitialize the member list.

If you changed your configuration by removing a system from the ASE, and you are reusing an existing ASE database, the asemgr utility checks the ASP policies of services for the deleted member name. You are asked to confirm whether or not to remove the deleted member name from the service's ASP policy. If deleting the member name results in an empty list:
- A service with the Favored Member policy effectively becomes one with the Balanced policy.
- A service with the Restricted to Member policy cannot be started until its ASP policy is modified to specify an existing member.
- A service with the Balanced policy is not affected.

4.4 Enable New DLM Interfaces and Reboot (PS-R)

Complete a rolling upgrade of all systems before enabling the new distributed lock manager (DLM) interfaces. After all systems are upgraded, run the dlm_enable script on each system.

Caution

After the new DLM interfaces are enabled on a cluster member, the DLM will panic if any cluster member is running a version of TruCluster software prior to Version 1.5, or if a Version 1.5 member attempts to use these new interfaces to communicate with another Version 1.5 member that does not have the interfaces enabled. Once you enable the new features on one member, make sure to enable these features on all cluster members before running an application that uses the new DLM interfaces.

To enable the new DLM features and reboot, follow these steps on each system in turn:

Run the dlm_enable script:
```
# /usr/sbin/dlm_enable
```

The script displays:
```
Would you would like to reboot the system at this time (y/n)? [n]:
 
```
Enter y. The script then runs the shutdown -r now command.

4.5 Enable New ASE Features (PS-R, AS-R)

Complete a rolling upgrade of all systems before enabling new ASE features. After all system are upgraded, enable new ASE features on one member in each ASE.

Note

Because there are no software checks that prevent you from enabling new features before all systems are upgraded, make sure that all systems are upgraded to the latest version of the TruCluster product before continuing.

To enable new features, run asemgr and enable the features by selecting the f menu option, Enable ASE V1.5 functionality. Once you enable new features, this menu option is disabled (the f is no longer visible). The following example shows the menu option with the f option enabled:

# asemgr
 
        TruCluster Production Server (ASE)
 
                ASE Main Menu
 
    a)  Managing the ASE         
-->
    m)  Managing ASE Services    
-->
    s)  Obtaining ASE Status     
-->
 
    x)  Exit                             ?)  Help
 
Enter your choice: a
 
                Managing the ASE
 
    a)  Add a member
    d)  Delete a member
    n)  Modify the network configuration
    m)  Display the status of the members
    C)  Display the configuration of the ASE database
    l)  Set the logging level
    e)  Edit the error alert script
    t)  Test the error alert script
    f)  Enable ASE V1.5 functionality
 
    q)  Quit (back to the Main Menu)    
    x)  Exit                             ?)  Help
 
Enter your choice: f

4.6 How to Update the DIGITAL UNIX Operating System Without Upgrading the TruCluster Software (PS, AS)

If the following conditions are met, you can update the DIGITAL UNIX operating system without upgrading the installed version of the TruCluster product:

Both versions of the operating system support the installed version of the TruCluster product.

The newer version of the operating system supports update installation.

The DIGITAL UNIX Release Notes provide information on the supported update paths for the operating system and the supported versions of TruCluster products.

When you update the DIGITAL UNIX operating system on an existing TruCluster member system, but are not upgrading the TruCluster software, you do not need to deinstall the TruCluster software kit before proceeding with the base system update. Correspondingly, you do not need to reinstall the TruCluster software kit after you have completed the update of the operating system. The current ASE configuration database (/var/ase/config/asecdb) is automatically preserved during an update of the operating system.

Note

You must update the operating system on all cluster or ASE members. At the end of the update, all systems must be running the same version of the DIGITAL UNIX operating system.

To update the operating system underlying the TruCluster software, perform the following procedure on each member system:

Disable the TruCluster software by adding or editing the clubase stanza entry in /etc/sysconfigtab to include the following lines:
```
clubase:
           cluster_disable=1
```

Shut down and halt the system using a command similar to the following:
```
#shutdown -h +30 Please log out
```

From the console prompt, invoke the init command and boot /genvmunix to single-user mode; for example:
```
>>> init 
>>> boot -fi genvmunix -fl s
```

Update the operating system. See the DIGITAL UNIX Release Notes and the DIGITAL UNIX Installation Guide for information about updating the operating system. See Section 1.3 for information about the operating system subsets required for the TruCluster product.

Reenable the TruCluster software by setting the value of the cluster_disable attribute to 0 in the /etc/sysconfigtab file:
```
clubase:
           cluster_disable=0
```

Reboot the system.

The procedure is complete after you update the last member's operating system.