HP StorageWorks XP48 Disk Array - Manual (Page 76)

Working With OS-Based Constraints 76
Supplementary note: To manually specify the setting, specify /etc/fstab .
Setting value: 2
Default value: 0 (Do not check the file system at startup)
# Depending on the cluster management software, if a standby configuration is used and this setting is
specified so that the file system is checked at startup, it may take time to switch the server when an error
occurs. To avoid this situation, the setting must be changed so that the file system check is not performed
at startup. For details, see the documentation for the relevant cluster management software.
4-6-5-2 Notes on Managing Volumes in Smaller Units
In Provisioning Manager, the unit of volume management is an LU or a volume group. If you need to manage
volumes in smaller units for purposes such as specifying volumes, physical partitions, and usage order, use an
OS command or volume manager command to manually create, expand, or delete a file system or device file.
4-6-5-3 Notes on Creating Many Small Files
When creating many small size files (for example, files of several tens of kilobytes), reducing the fragment size
of the file system can save the disk space but the duration of file system operations might increase.
The following value is specified by default as the file system fragment size:
• When the file system size is no more than 4 TB
1,024 bytes
• When the file system size is more than 4 TB but no more than 8 TB
2,048 bytes
• When the file system size is more than 8 TB but no more than 16 TB
4,096 bytes
• When the file system size is more than 16 TB but no more than 32 TB
8,192 bytes
When you want to create a file system with a fragment size different from the default, manually change the
setting for the fragment size, and then create the file system.
4-6-5-4 Notes on Handling File Systems and Device Files as Cluster Resources
To handle a file system and device files managed by Provisioning Manager as resources of the cluster
management software (Serviceguard), you must perform the following operations:
• Disable the volume group (use the -a n option of the vgchange command).
• Specify the setting for disabling the volume group at host startup (specify 0 for AUTO_VG_ACTIVATE in
the /etc/lvmrc file).
• Specify the volume groups that will be not be used as cluster resources (define these volume groups in
the custom_vg_activation function in the /etc/lvmrc file).
• Unmount the file system (use the umount command).
• Specify the settings for unmounting the file system at host startup. (edit /etc/fstab, by deleting or
commenting out the description in the file system to be used as cluster resource.)
• Mark the volume groups in a cluster as part of the high availability cluster (use the -c y option of the
vgchange command).
If you do not perform these operations, the file system might become damaged or the system may stop. For
details about the settings, see the relevant OS documentation.
4-6-5-5 Notes on Manipulating File Systems and Device Files Used as Cluster Resources
When you delete or expand a file system or device file used as a cluster resource, perform the following
operations before and after deletion or expansion. For operation details, see the documentation for the relevant
OS.
When deleting a file system or device file
Before you start deletion, perform the following on the host:
1. Stop the cluster.
If you continue operation without stopping the cluster, the cluster environment might become damaged.
You must stop the cluster before proceeding further.