Administering Disks

The following sections describe the Solaris disk-naming conventions, commands for finding disk information (du, prtvtoc), and ways to repair or replace a bad disk.

Disk-Naming Conventions

Solaris disks have both block and raw (character) device files. The device name is the same, regardless of whether the command requires the block or raw device file.

Each type of device file has its own subdirectory in /dev: /dev/dsk (the block interface) or /dev/rdsk (the raw interface).

Some commands, such as mount, use the block interface device name from the /dev/dsk directory to specify the disk device. Other commands, such as newfs, require the raw interface device name from the /dev/rdsk directory to specify the disk device.

Raw device interfaces transfer only small amounts of data at a time. To use a raw device, you specify the device in the /dev/rdsk subdirectory. The r in rdisk stands for raw.

Block device interfaces include a buffer from which large blocks of data are read at one time. When a command requires the block device interface, you specify the /dev/dsk subdirectory.

The device name you use to identify a specific disk with either type of interface depends on the controller type: bus-oriented (SCSI) or direct. You refer to a disk device by specifying the subdirectory to which it is symbolically linked (either /dev/dsk or /dev/rdsk) followed by a string identifying the particular controller, disk, and slice.

/dev/[r]dsk/cwtxdysz

cw is the controller number, tx is the target number, dy is the drive number, and sz is the slice identifier. For SCSI drives, the drive number is the equivalent of the Logical Unit Number (LUN) of the drive.

Table 69 shows which interface to use for a few frequently used disk and file system commands.

Table 69. Device Interface Type for Some Frequently Used Commands

Command	Interface Type	Example
df(1M)	Block	df /dev/dsk/c0t1d0s0
fsck	Raw	fsck -p /dev/rdsk/c0t0d0s0
mount(1M)	Block	mount /dev/dsk/c1t0d0s7 /export/home
newfs(1M)	Raw	newfs /dev/rdsk/c0t0d1s1
prtvtoc(1M)	Raw	prtvtoc /dev/rdsk/c0t0d0s2

Using Disks with Bus Controllers

Figure 15 shows the device-naming convention for SPARC disks with bus controllers.

Figure 16 shows naming conventions for IA disks with SCSI controllers.

Each file system on a disk is assigned to a slice—a group of cylinders set aside for use by that file system. To specify a slice (partition) on a disk with a bus controller (either SCSI or IPI), use a device name with these conventions: /dev/dsk/cWtXdYsZ (the block interface) or /dev/rdsk/cWtXdYsZ (the raw interface).

Use the following guidelines to determine the values for the device file name.

• If you have only one controller on your system, W is always 0.

• For SCSI controllers, X is the target address set by the switch on the back of or inside the unit.

• Y is the Logical Unit Number (LUN) of the drive attached to the target. If the disk has an embedded controller, Y is always 0.

• Z is the slice (partition) number, with a value ranging from 0 to 7. Slice 2 is known as the backup slice and should never be used directly. Table 70 shows conventional assignments of slice (partition) numbers for the disk on which root is found.

Table 70. Conventional Assignments of Slices for Boot Disk

Slice	File	System Use
0	root	Operating system.
1	swap	Virtual memory space.
2	-	Entire disk.
3–5		Available for use according to your administrative policy.
6	/usr	Executable programs, program libraries, and documentation.

Table 71 shows some examples of raw device names for disks with bus-oriented controllers.

Table 71. Examples of Device Names for Disks with Bus-Oriented Controllers

Device Name	Description
/dev/rdsk/c0t0d0s0	Raw interface to the first slice (root) on the first LUN (disk) at the first SCSI target address on the first controller.
/dev/rdsk/c0t0d0s2	Raw interface to the third LUN (which represents the whole disk) on the first disk at the first SCSI target address on the first controller.
/dev/rdsk/c0t1d0s6	Raw interface to seventh (/usr) slice on the first LUN (disk) at the second SCSI target address on the first controller.

Using Disks with Direct Controllers

Disks with direct controllers do not have a target entry as part of the device name. To specify a slice (partition) on a disk with a direct controller, use a device name with the following conventions: /dev/dsk/cXdYsZ (the block interface) or /dev/rdsk/cXdY s Z (the raw interface).

Figure 17 shows the naming convention for SPARC-based disks with direct controllers. If you have only one controller on your system, X is always 0. Slice 2 is a guide to the size of the entire disk.

Figure 18 shows the naming convention for IA-based disks with direct controllers.

Use slice 2 (s2) to specify the entire Solaris fdisk partition.

Setting Up Disk Slices

Files are stored within file systems. Each disk slice is treated as a separate disk drive both by the operating system and by the system administrator. When setting up slices, be aware of the following constraints.

• Any disk slice can have a file system written on it.

• No file system can span multiple slices.

You set up slices differently on SPARC and IA platforms, as described in Table 72.

Table 72. Slice Differences on Platforms

SPARC Platform	IA Platform
Entire disk is used for Solaris environment.	Disk is divided into fdisk partitions, one per operating environment.
Disk is divided into eight slices, numbered 0–7.	The Solaris fdisk partition is divided into 10 slices, numbered 0–9.

SPARC Disk Slices

For SPARC-based systems, Sun recommends that you define only three disk slices, root (/), swap, and the backup slice (slice 2). Table 73 describes additional slice conventions that you can use if you choose.

Table 73. SPARC Disk Slice Conventions

Slice	File System	Client/Server	Description
0	root	Both	Hold files and directories that make up the operating system.
1	swap	Both	Provide virtual memory or swap space.
2		Both	By convention, refer to the entire disk. The entire disk is defined automatically by the format command and the Solaris installation programs. Do not change the size of this slice.
5	/opt	Both	Hold applications software added to a system. If a slice is not allocated for this file system during installation, the /opt directory is merged into the slice that contains the root (/) file system.
6	/usr	Both	Hold operating system commands—also known as executables—designed to be run by users. This slice also holds documentation, system programs such as init and syslogd, and library routines. If a slice is not allocated for this file system during installation, the /usr directory is merged into the slice that contains the root (/) file system.

IA Disk Slices

On IA-based systems, you divide disks into fdisk partitions. Each fdisk partition is a section of the disk reserved for a particular operating environment. For a Solaris fdisk partition, you can define 10 slices, numbered from 0 through 9 and assign each to a conventional use. The uses for slices 0–2 and 5–7 are the same as on Solaris systems, described in Table 73. Table 74 describes slices 8 and 9.

Table 74. IA Conventions for Slices 8 and 9

Slice	File System	Client/Server	Description
8	—	Both	Contain the boot slice information at the beginning of the Solaris partition that enables Solaris to boot from the hard disk.
9	—	Both	Provide an area reserved for alternative disk blocks. Slice 9 is known as the alternative sector slice.

Determining Which Slices to Use

When you set up file systems for a disk, you choose not only the size of each slice but also which slices to use. Your decisions depend on the configuration of the system and the software you want to install on the disk. System types are defined by how they access the root (/) and /usr file systems, including the swap area. For example, stand-alone and server systems mount these file systems from a local disk; other clients mount the file system remotely.

In previous releases, you could set up the following five system configurations.

• Servers.

• Diskless clients.

• Stand-alone systems.

• Dataless clients.

• Solstice AutoClient systems.

With the Solaris 8 release, the system configurations are simplified to servers, stand-alone systems, and JavaStations. The JavaStation is a client designed for zero administration. It optimizes Java technology and takes full advantage of the network to deliver everything from Java applications and services to complete, integrated system and network management. You do no local administration for a JavaStation. The server handles booting, administration, and data storage.

Table 75 summarizes the three system types.

Table 75. System Configurations and Slice Requirements

System Type	Local File Systems	Local Swap	Remote File Systems	Network Use
Server	root (/)	Yes	None	High
	/usr
	/home
	/opt
	/export/home
	/export/root
Stand-alone	root(/)	Yes	None	Low
	/usr
	/export/home
JavaStation	None	No	/home	High

Disks and SMC

SMC provides two tools in the Storage category, described in Table 76, that you can use to display and format disks.

Table 76. SMC Disks Tools in the Storage Category

Tool	Description
Disks	Display disk information, display partition information, partition disks, copy disk layouts from one disk to a disk of the same size and manufacturer, create fdisk partitions, and change the active fdisk partition on an IA computer.
Enhanced Storage	Create and manage RAID0 volumes, including stripes and concatenation; RAID1 volumes, including mirrors; RAID5 volumes; soft partitions; hot spare pools; disksets; and the state configuration database (the metadb). New in SMC 2.1.

Disk Use Check (du)

To find the number of 512-byte disk blocks used per file or directory, type du and press Return. When directories contain subdirectories, the subdirectories and their contents are included in the block count, as shown in the following example.

oak% du
2913     ./3.0templates
639      ./Art
347      ./Howto
1998     ./Clipart
607      ./Newtemplates
38       ./Modemstuff
2004     ./Config/Art
6593     ./Config
13280    .
oak%

The output is displayed in 512-byte blocks. To convert to megabytes, divide by 2048. In the preceding example, 13280/2048 = 6.48 Mbytes.

Starting with the Solaris 9 release, the du command has an -h option that you can use to display the output in a user-friendly format, as shown in the following example.

mopoke% du -h
   2K   ./.dt/sessionlogs
   1K   ./.dt/types/fp_dynamic
   2K   ./.dt/types
   1K   ./.dt/icons
   1K   ./.dt/appmanager
  19K   ./.dt/help/winsor-mopoke-0
  20K   ./.dt/help
  11K   ./.dt/sessions/current
  17K   ./.dt/sessions
   1K   ./.dt/tmp
   1K   ./.dt/Trash
   1K   ./.dt/Desktop
   1K   ./.dt/palettes
   2K   ./.dt/.Printers
  56K   ./.dt
   1K   ./.java/.userPrefs
   2K   ./.java
   2K   ./.solregis
  17K   ./.netscape/cache
   1K   ./.netscape/archive
   1K   ./.netscape/xover-cache/host-news
   2K   ./.netscape/xover-cache
 331K   ./.netscape
   1K   ./nsmail
 1.1G   .
mopoke%

Disk Information Check (prtvtoc)

Use the prtvtoc (print volume table of contents) command to display information about disk partitioning. If you use the standard slice-naming conventions, specifying slice 2 displays the contents of the entire disk.

Use the following steps to display information about disk partitioning.

paperbark% su
Password:
# prtvtoc /dev/rdsk/c0t0d0s2
* /dev/rdsk/c0t0d0s2 partition map
*
* Dimensions:
*     512 bytes/sector
*      80 sectors/track
*      19 tracks/cylinder
*    1520 sectors/cylinder
*    3500 cylinders
*    2733 accessible cylinders
*
* Flags:
*   1: unmountable
*  10: read-only
*
*                          First     Sector      Last
* Partition Tag Flags      Sector     Count      Sector  Mount Directory
       0     2   00      1048800   2865200     3913999   /
       1     3   01            0   1048800     1048799
       2     5   00            0   4154160     4154159
       7     8   00      3914000    240160     4154159   /export/home

Bad-Disk Repair

The following sections describe the steps for repairing a bad disk or reinstalling a new one.

Try Archiving the Files

If you can access the drive, do a ufsdump of all the file systems on the disk. See "Backing Up and Restoring File Systems" on page 211 for information on how to use the ufsdump command.

Try Copying Data from the Disk

If you cannot run ufsdump on the disk, find another disk of the same type, connect it to the system, and use either the dd or volcopy commands to copy the data from the bad disk. See the dd(1M) and volcopy(1M) manual pages for complete information on how to use these commands.

The dd command makes a literal (block) copy of a complete UFS file system to another file system or to a tape. By default, the dd command copies its standard input to its standard output.

You can specify a device name in place of the standard input, the standard output, or both. The following example copies contents of a diskette to a file in the /tmp directory.

oak% dd < /floppy/floppy0 > /tmp/output.file
2400+0 records in
2400+0 records out
oak%

The dd command reports on the number of blocks it reads and writes. The number after the + is a count of the partial blocks that were copied.

The dd command syntax is different from most other commands. You specify options as keyword=value pairs, where keyword is the option you want to set and value is the argument for that option. For example, you can replace the standard input and output with the following syntax.

dd if=input-file of=output-file

The following example uses the keyword=value pairs instead of the redirect symbols in the previous example.

oak% dd if=/floppy/floppy0 of=/tmp/output.file

Use the following steps to clone a disk with the dd command.

oak% su
Password
# dd < /floppy/floppy0 > /tmp/output.file
# boot
(Boot messages)
# dd if=/dev/dsk/c0t0d0s2 of=/dev/dsk/c0t2d0s2 bs=100k
# fsck /dev/rdsk/c0t2d0s2
# mount /dev/dsk/c0t2d0s2 /mnt
# cd /mnt/etc
# vi vfstab
(Modify entries for the new disk)
# cd /
# umount /mnt
# init 0
(Shutdown messages)
# boot disk2 -s
(Boot messages)
# sys-unconfig
# boot disk2

Try Repairing Any Bad Blocks

If the disk has bad blocks, you may be able to repair them with the format command. See the format(1M) manual page for more information.

Try Reformatting the Disk

If the disk is bad, reformatting it may fix the problem. Use the format command to reformat a disk. See the format(1M) manual page for more information.

Replacing the Bad Disk

If reformatting and repairing bad blocks do not work, replace the disk. See the disk installation manual for more information.

Adding Defect List, Format, Partition, and Label Disk (format)

Use the following steps to put a defect list on a new disk and to format, partition, and label it.

oak% su
Password:
# format
Searching for disks...done

AVAILABLE DISK SELECTIONS:
        0. sd0 at esp0 slave 24
           sd0: <SUN0207 cyl 1254 alt 2 hd 9 sec 36>
        1. sd2 at esp0 slave 16
           sd2: <SUN0207 cyl 1254 alt 2 hd 9 sec 36>
Specify disk (enter its number): 1
selecting c0t0d0
[disk formatted]
FORMAT MENU:
        disk       - select a disk
        type       - select (define) a disk type
        partition  - select (define) a partition table
        current    - describe the current disk
        format     - format and analyze the disk
        repair     - repair a defective sector
        label      - write label to the disk
        analyze    - surface analysis
        defect     - defect list management
        backup     - search for backup labels
        verify     - read and display labels
        save       - save new disk/partition definitions
        inquiry    - show vendor, product and revision
        volname    - set 8-character volume name
        quit
format > defect
defect > primary
Extracting primary defect list . . . Extraction complete.
Current Defect List updated, total of 30 defects.
defect > quit
format > format
format> partition
PARTITION MENU:
        0      - change '0' partition
        1      - change '1' partition
        2      - change '2' partition
        3      - change '3' partition
        4      - change '4' partition
        5      - change '5' partition
        6      - change '6' partition
        7      - change '7' partition
        select - select a predefined table
        modify - modify a predefined partition table
        name   - name the current table
        print  - display the current table
        label  - write partition map and label to the disk
        quit
partition> <partition the disk>
partition> label
partition> quit
format > quit
#

Remaking the File Systems (newfs)

A disk must be formatted, partitioned, and labeled before you can create UFS file systems on it. If you are re-creating an existing UFS file system, unmount the file system before performing the following steps.

The following example creates a file system on /dev/rdsk/c0t3d0s7.

oak% su
Password:
# newfs /dev/rdsk/c0t3d0s7
newfs: construct a new file system /dev/rdsk/c0t3d0s7 (y/n)? y
/dev/rdsk/c0t3d0s7:     163944 sectors in 506 cylinders of 9 tracks, 36
 sectors
       83.9MB in 32 cyl groups (16 c/g, 2.65MB/g, 1216 i/g)
super-block backups (for fsck -b #) at:
 32, 5264, 10496, 15728, 20960, 26192, 31424, 36656, 41888,
 47120, 52352, 57584, 62816, 68048, 73280, 78512, 82976, 88208,
 93440, 98672, 103904, 109136, 114368, 119600, 124832, 130064, 135296,
 140528, 145760, 150992, 156224, 161456,
#

Mounting the File System on a Temporary Mount Point (mount)

Type mount /dev/dsk/ cntndnsn /mnt and press Return. The file system is mounted on the /mnt temporary mount point. To mount the disk, specify the block device directory (/dev/dsk), not the raw device directory.

Restoring Files to the File System (ufsrestore)

Restore the contents of the latest full backup, and then restore subsequent incremental backups from lowest to highest level (ufsrestore), by using the following steps.

Unmounting the File System from Its Temporary Mount Point (umount)

Use the following steps to unmount the file system from its temporary mount point.

Checking the File System for Inconsistencies (fsck)

Type fsck /dev/rdsk/cntndnsn and press Return. The file system is checked for consistency.

Performing a Level 0 Backup of the Restored File System (ufsdump)

You always should do an immediate backup of a newly created file system because ufsrestore repositions the files and changes the inode allocation.

Use the following steps to perform a level 0 backup of the restored file system.

Mounting the File System at Its Permanent Mount Point (mount)

Type mount /dev/dsk/cntndnsn and press Return. The restored file system is mounted and available for use.