F    Manual Storage Configuration When Upgrading

This appendix contains information for sites that are upgrading from TruCluster Software Version 1.5 or Version 1.6 to TruCluster Server Version 5.1B but that either cannot or choose not to use the Option 2 or Option 3 storage mapping and configuration scripts described in Chapter 8.

In general and where possible, we recommend that you use the procedures and scripts described in Chapter 8. However, if your storage topology, system configurations, or site policy make it impossible to do so, this appendix describes how to manually gather storage configuration information in the Available Server Environment (ASE) and how to manually configure storage on the new Tru64 UNIX system or single-member cluster. You are responsible for mapping old-style (rz*) device names to new-style (dsk*) device names.

You must create your own upgrade procedure. Read all of Chapter 8 and this appendix, decide which upgrade option is a reasonable starting point for your upgrade, and then modify that option's procedure.

F.1    Manually Gathering Device and Storage Configuration Information

This section replaces the steps in Option 2 and Option 3 that use the clu_migrate_save script to gather information about the current ASE configuration and storage environment.

First, read Section 8.3.2 and create an up-to-date configuration map for the current ASE.

In Chapter 8, the clu_migrate_save script captures the current shared storage configuration including LSM and AdvFS configuration information. The clu_migrate_configure script reads the gathered information and configures storage on the new Tru64 UNIX system or single-member cluster. However, if you plan to manually configure storage on the new cluster after connecting the physical devices, you will not run clu_migrate_configure and therefore need to manually gather storage configuration information for the members of the ASE.

Note

In addition to manually gathering information, we recommend running clu_migrate_save on the current ASE members. Other than gathering storage configuration information, clu_migrate_save writes each device's special file name (rz*) to the label: field of that device's disk label. (The script also saves the the original device label as described in Chapter 8; you can restore the original labels after upgrading.)

We also recommend running clu_migrate_configure -x on the new Tru64 UNIX system or TruCluster Server cluster. The clu_migrate_configure -x command does not configure storage; it lists the commands it would run if invoked without the -x option and displays a mapping of old-style rz* device names to new-style (dsk*) device names. However, you must have run clu_migrate_save on the ASE members in order for clu_migrate_configure -x to provide this mapping.

If you do not use the scripts to help map old-style device names to new-style device names, exercise extreme care when manually mapping device names. You must know the physical location of each device, and be able to use this knowledge, and utilities like hwmgr and scu, to determine which dsk* name is assigned to each device during the upgrade. Determining which dsk* the new system or cluster assigns to a device previously known as rz* is not a trivial task; it is the primary reason for providing the scripts. Although you do not have to use the migration scripts to configure storage on the new Tru64 UNIX system or TruCluster Server cluster, they are highly recommended for making sure that you know which rz* device is now known as dsk*.

On each member of the ASE:

F.2    Manually Configuring Storage on the New Tru64 UNIX System or TruCluster Server Cluster

This section replaces the steps in Option 2 and Option 3 that use the clu_migrate_configure script to configure storage on the new Tru64 UNIX system or single-member cluster.

Note

If you did not run clu_migrate_save on the ASE members, you cannot use clu_migrate_configure -x to display device-name mappings. Before continuing, perform a manual mapping of all old-style rz* device names to the new-style dsk* device names. In the following procedure, substitute the results of your own mapping for those provided by clu_migrate_configure -x.

If you ran clu_migrate_save on the ASE members, on the Tru64 UNIX system or the single-member cluster, run clu_migrate_configure -x: 

# /usr/opt/TruCluster/tools/migrate/clu_migrate_configure -x
 

The clu_migrate_configure -x command displays a mapping of old-style device names to new-style device names. Use this information when configuring the storage that was controlled by the ASE.

The following steps provide some guidance when configuring storage:

  1. If you are not using the clu_migrate_save and clu_migrate_configure -x commands to map device names, manually map old-style (rz*) device names to new-style (dsk) device names. Use the scu command on the new system to help map devices to their new device names. For example: 

    # scu -f /dev/rdisk/dsk5g show device | grep -E "Vendor|Product|Firmware"
          Vendor Identification: DEC
         Product Identification: RZ26L    (C) DEC
        Firmware Revision Level: 440C
 
# scu -f /dev/rdisk/dsk5g show inq page serial | grep "Product Serial"
Product Serial Number: PCB=420240831056(ZG40831056 ?); \
HDA=0000000042181869
 
# scu -f /dev/rdisk/dsk5g show nexus
Device: RZ26L, Bus: 1, Target: 4, Lun: 0, Type: Direct Access
 

    Using the scu information that you collected from the ASE members, create a map of old-style device names to new-style device names. The hwmgr command is also a useful tool when manually mapping device names.

    The scu examples in Section F.1, show that this device was known as rz28 in the ASE. The bus numbers in the show nexus output are not the same. Because ase_fix_config was run in the ASE, bus numbers are not the same on both systems, and are not a reliable piece of information for mapping devices.

  2. Using the LSM information from the saved asecdb database, the output from sys_check -all, and the device mapping table that you created manually or with clu_migrate_configure -x, configure each device and LSM disk group.

    Note

    The LSM configuration information on disk has changed format. If you need to revert to the ASE from this point on, you will need to retore the LSM information when you import on an ASE system.

    For every new LSM device, enter: 

    # voldisk define device
# voldisk online device
 

    For every disk group, enter: 

    # voldg import disk_group
 

    For every volume, enter: 

    # volume -g disk_group start volume
 

    For example: 

    # voldisk define dsk5g
# voldisk online dsk5g
# voldg import toolsdg
lsm:voldg: WARNING: Volume vol01: \
Temporarily renumbered due to conflict
# volume -g toolsdg start vol01
# volume -g toolsdg start vol02
 

    You can ignore the warnings.

  3. To prepare LSM to update names following the next reboot, enter the lsmupdate_setup command: 

    # /sbin/lsm.d/bin/lsmupdate_setup
 

  4. Using the AdvFS information from the saved asecdb database, the output from sys_check -all, and the device mapping table that you manually created or with clu_migrate_configure -x, manually re-create the AdvFS domains that were on the ASE, mapping the old-style rz device names to new-style dsk device names and creating the appropriate /etc/fdmns entries.

    For example: 

    # mkdir /etc/fdmns/data1_domain
# cd /etc/fdmns/data1_domain
# ln -s /dev/disk/dsk6g
 
# mkdir /etc/fdmns/tools_dmn
# cd /etc/fdmns/tools_dmn
# ln -s /dev/vol/toolsdg/vol01 toolsdg.vol01
 

    For each domain, enter the showfsets domain command and verify that the filesets are correct for the domain.

  5. Mount the domains. For each domain, run the showfdmn domain command.

  6. Add file systems to the /etc/fstab file. Also update any other configuration files that contain storage information, such as /etc/exports.

  7. Reboot the system to configure LSM with the new device names: 

    # shutdown -r now
 

  8. Enter the following LSM commands to examine the LSM configuration: 

    # voldisk list
# volprint -thA