newarray(1M)

NAME

newarray — configure a disk array

SYNOPSIS

newarray [-N Config_Name | -rRAID_Level] [Options] device_file

DESCRIPTION

newarray, a front-end program for the utility cfl (see cfl(1M)), facilitates the configuration of Hewlett-Packard SCSI disk arrays. It is the recommended utility for all array configuration. Array configuration maps a set of one or more physical disk mechanisms in an array to a set of one or more logical disks, addressable by HP-UX. Logical disks are addressed through device files. Each logical disk in an array (also known as a LUN, for Logical UNit), has its own device file. A logical disk can consist of a single physical disk, a portion of a single physical disk, multiple physical disks, or portions of multiple physical disks. For additional information about possible array configurations, see the array configuration table contained in the file /etc/hpC2400/arraytab, and arraytab(4).

Supported configurations for the array device are pre-defined in the array configuration table, located in file /etc/hpC2400/arraytab.

newarray can configure a complete set of logical partitions for an array in one operation. Due to the inter-dependency of logical partitions, this is the recommended method for configuration. A single logical partition can be added to an array configuration using an entry from the array configuration table by using the -L option.

device_file is a character device file that specifies the I/O address, and driver to use when configuring the disk array. The way that this file is used by newarray is system dependent. See dependencies below. Logical partitions in an array are independently addressable by using the appropriate device file to address the logical unit assigned to a partition.

Prior to configuring the array (except with the -L option ), all currently configured logical partitions are removed from the configuration.

To simplify array configuration newarray obtains much of the necessary information directly from the array device, and its attached disk mechanisms. The array model number, and the number of available physical disks available, is determined by querying the device. This information is used to locate the appropriate configuration entry in the array configuration table. Optional parameters can be used to override the default, and inquiry values.

The preferred configuration method is to use the -N option to specify a configuration by name. The name determines which configuration newarray uses from the array configuration table. Configuration parameters are obtained from the named configuration entry. Parameters of the chosen configuration can be overridden using options to newarray, or by creating and using a custom configuration entry in the array configuration table. See the WARNINGS section of this manpage.

Because the array controller type, and disk mechanism types are used in addition to the configuration name to select an entry from the array configuration table the configuration name does not have to be unique within the array configuration table. However, the combination of configuration name, array controller type, and disk mechanism types must be unique within the array configuration table. During configuration, the array controller type, and disk mechanism types are obtained by querying the devices.

The -r option specifies an operating mode, rather than specifying a configuration by name. The -d option, which specifies the size of a disk group, is often used with the -r option. If -d is not used, newarray selects the configuration in the array configuration table that most closely matches the disks in the array.

When the configuration parameters have been determined, newarray calls cfl.

If the -V option is used, newarray prints its actions, and the parameters it passes to cfl to configure the array (see cfl(1M)).

Array Configuration

newarray obtains its configuration values from the array configuration table. If not specified there, default values are provided by cfl (see cfl(1M)). Configuration values can be overridden by newarray options.

Options

-L unit addr

Configures a single LUN from the specified configuration. The -L option is useful for adding disks to an array without changing the existing configuration. Because the order in which LUN's are configured determines the physical mapping on the disks within the array, be very careful when using the -L option.

-N config_name

The name of the configuration to be used, as specified in the configuration file /etc/hpC2400/arraytab. See arraytab(4).

-V

Display the parameters of array configuration, and the utility commands issued as part of the configuration process.

-b block_size

The size in bytes of the LUN block. Must be an integral number of the physical disk mechanism sector size. Currently supported values are 512, 1024, 2048, and 4096.

-c capacity

The size in blocks of the LUN. A value of 0 defaults to the largest capacity available. If the LUN type is set to sub-LUN, the capacity is the available capacity of the composite drive group or 2 GByte if the 2 GByte flag is set, which ever is smaller. See -f option.

-d group_size

Physical drive group will contain this number of disks in the logical partition configuration.

-f flags

Configuration flags. There are 16 flags, represented by a 16 bit hexadecimal number. Currently only four of the flags are defined. The flag definitions and their default value are:

Bit 0 off: Not used.
Bit 1 on: Disable auto reconstruction. When set (on), disables the automatic detection, and initiation of failed disk data reconstruction.
Bit 2 off: Not used.
Bit 3 off: Not used.
Bit 4 on: When set (on), enables AEN (automatic event notification) polling.
Bit 5 on: When set (on), enables read parity verification.
Bit 6 on: When set (on), enables write with parity verification.
Bit 7 off: Not used.
Bit 8 off: Mode Sense default pages. Bit 8 and Bit 9 concurrently set is reserved.
Bit 9 off: Mode Sense current pages. Bit 8 and Bit 9 concurrently set is reserved.
Bit 10-15 off: Not used.

-g group_name

Use physical drive group configuration with label GroupName (in array configuration table) for this LUN configuration.

-i seg0_size

The size in bytes of the first segment LUN. This allows this area to be set to a size different than the remainder of the disk, an area typically used as the boot block for some systems. This must be a integral number of the block-size. If there are no special requirements, this parameter should be set to 0.

-k recon_size

Reconstruction size. The -k option specifies (in LUN blocks) the amount of data to be reconstructed in a single operation during reconstruction of a redundant drive configuration. Larger values provide more efficient (faster) reconstruction, but hold off the servicing of I/O requests. Smaller values allow quicker servicing of I/O requests, but with less efficient (slower) reconstruction.

-l recon_freq

Reconstruction frequency. The -l option specifies (in tenths of a second) the time period between reconstruction of disk segments in a redundant drive configuration. Small time periods cause the array to consume most of its time reconstructing data, but allow the reconstruction to complete more quickly. Large time periods allocate more time to I/O processing, but require longer reconstruction times.

-r raid_level

The RAID (redundancy level) to apply to the disks in the array. Valid entries for raid_level are RAID_0, RAID_1, RAID_3, and RAID_5. Some RAID levels require specific physical drive configurations. See also the -g option.

-s seg_size

The number of bytes of a contiguous segment of the logical address space residing on a single physical disk. This affects how many physical disks are involved in a single I/O request. If I/O requests are mostly random, single-block requests, set this value to the integral number of the LUN block size that minimizes the number of disks necessary to service most I/O requests. A larger size will allocate more time to I/O processing.

-t LUN_type

LUNs can be configured as regular LUNs (reg), or sub-LUNs (sub). A regular LUN utilizes all the available capacity of a disk group, or limits the LUN configuration to 2 GBytes if the 2 GByte limiter is set. If a regular LUN configuration is used, the -c option is ignored. A sub-LUN allows logical partitioning of the disk group capacity into a maximum of eight LUNs. Valid values for LUN_type are "reg " and "sub ".

Custom Configurations

You can create array configurations that might be better suited to a particular application by using newarray's command line parameters to override default values, or by creating special entries in the array configuration table in the file /etc/hpC2400/arraytab. Before you do, see cautionary notes in the WARNINGS section of this manpage.

RETURN VALUES

newarray will return the following values:

0: Successful completion.
-1: Command failed (an error occurred).

ERRORS

newarray: device busy 

To ensure that newarray does not modify a disk array that is being used by another process, newarray attempts to obtain exclusive access to the disk array. If the disk array is already opened by another process (for example, LVM — the Logical Volume Manager), a ``device busy'' error message is returned by the driver. To eliminate the ``device busy'' condition, determine what process has the device open. In the case of LVM, it is necessary to deactivate the volume group containing the array before configuring the array (see vgchange(1M)).

EXAMPLES:

The following examples use configurations contained in /etc/hpC2400/arraytab.

Raid Level Specification

To configure an HP C2425D with 5 internal disks to a five drive RAID level 0 configuration (on Series 700 computer):

newarray -rRAID_0 /dev/rdsk/c2t3d0 

To configure an HP C2425D with 5 internal disks to a one drive RAID level 0 configuration (on Series 700):

newarray -rRAID_0 -d1 /dev/rdsk/c2t3d0 

Name Specification

To configure an HP C2430D with five disks connected on SCSI channel 3 (on a Series 800) using the configuration "Raid_3_5d" in /etc/hpC2400/arraytab:

newarray -NRaid_3_5d /dev/rdsk/c2t3d0 

WARNINGS

We strongly recommend that you use the array configurations that are specified, and delivered by Hewlett-Packard, in the file /etc/hpC2400/arraytab. These configurations have been tested and certified for proper use on Hewlett-Packard computer systems. Custom configurations cannot be warranted for proper operation.

Configuring a disk array causes the loss of user data on the array.

When using the -L option, physical media is assigned to the logical unit in the order in which the logical units are configured. Existing logical unit configurations are NOT removed prior to configuration with this option. The use of this option is not recommended at this time.

DEPENDENCIES

File System Considerations

The disk array maps the address space of one or more physical disk mechanisms onto logical "disk" partitions. The parameters defined in the configuration, together with the data access patterns of the user's application, determine the operating characteristics of the logical disk. Some configurations create multiple logical partitions, that share a set of physical disks. I/O traffic to each of the logical partitions affects performance, due to the common physical disk resources. The file system or application using the "logical" disk may require or assume certain characteristics. For optimal system performance it is necessary that the file system configuration and application be compatible with the array configuration.

Your choice of segment size directly affects the performance of the disk array. Choose this parameter in concert with the choice of the parameters used when building the file system on the device. In general, the segment size determines how much data from a single I/O will be stored on a single disk within the array. A smaller value will involve more of the disks with the I/O, whereas a larger value will involve fewer disks. If input/output operations tend to be very long, the involvement of multiple disks may hasten the completion of each I/O. In this case the access time is the same as a single disk, but the disk data transfer time is shared across the set of disks. If input/output operations are short, the access time will dominate relative to the disk data transfer time, and more input/output operations may be processed in parallel by involving fewer disks in each I/O. In all cases the relative locality of data and the access pattern will affect the performance. For highly sequential data, it may be advantageous to locate the data for a single I/O on a single disk, to take advantage of read-ahead caching within each disk.

Configurations for the HP C2430 disk array should enable the automatic data reconstruction LUN flag as part of the configuration specification.

Supported Array Products:

The HP C2425 and HP C2427 disk arrays are only supported on Series 700 systems running HP-UX version 9.0X.

The HP C2430 disk array is supported on Series 700 and 800 systems running HP-UX versions 9.0X and 10.0X.

AUTHOR

newarray was developed by HP.