sdipath(1M)

Synopsis

Description

Currently, disks are the only devices supported by multipath I/O.

Multipath I/O is implemented as a device driver in the operating system kernel and as such is able to detect and discover all known paths for devices on system boot. By default the multipath I/O (MPIO) driver will attempt to use all paths and load balance I/O across controllers.

The sdipath command is the administrative interface to managing multipath I/O. There are operations provided for listing paths that include status and performance information, and operations for controlling which paths are active so you can tune your system for better performance or higher availability, as well as repair paths that have failed.

As the system boots, physical disks are discovered and brought online. Each disk has a stamp associated with it that is used to determine if this is the first occurrence of this disk. If it is the first occurrence, a device node is created in /dev/rdsk and /dev/dsk, then an entry is made into the resource manager database. All subsequent occurrences of this disk will only get an entry in the resmgr(1M) database under an ``sdol'' entry.

The pathname associated with each path to the disk is created by appending the resource manager key value to the word path. Thus if a resource manager entry is made for a path having a resmgr key of 29, the name of this path would be ``path29''.

Options

-o

Specifies the operation to be performed. These can be as follows:

list

This operation is used to list information about I/O paths known to the MPIO driver. The following information is displayed:

• pathname

• path state

• path locale

• number of read requests for path

• number of write requests for path

The path state can be either Active, Inactive, or Failed.

If active, the path is currently available for I/O. In this state, the MPIO driver can send I/O through this path to the device.

If inactive, the path is not available for I/O. For example, a path may be inactive if it is connected to the inactive port of a standalone RAID device, such as a CLARiION disk array.

If failed, either the driver marked the path as failed after being unable to perform I/O on the path, or the system administrator marked the path as failed manually. Paths that are in the failed state will remain failed until the administrator issues a repair operation after the problem has been corrected.

fail

This operation is used to stop the I/O flow through a specified path. It is sometimes useful to be able to turn off the I/O through a specific path for either performance, availability, or maintenance reasons. When this operation is performed, the MPIO driver sets a flag to indicate that the specified path was failed.

repair

This operation is used to restart the I/O flow through a path that has been in one of the failed states. This operation should be used to activate paths that were failed by the administrator, as well as reactivating paths that have been physically repaired after a system failure that resulted in the path being marked failed.

-d device

device can be specified in one of three ways. You can use the cCbBtTdLp0 notation, specify a full path to the device node, or specify the alias name for the device in the /etc/device.tab file.

-q

Quiet mode. Produce an unformatted listing with no headers and fields delimited by a single space.

-v

Verbose mode. Produce a formatted listing with headers and aligned columns. (This option is the default.)

Usage

To list information about all paths on the system:

sdipath -o list

To list information about all paths associated with disk3:

sdipath -o list -d disk3

To fail path51 of device c0b0t0d0p0:

sdipath -o fail -d c0b0t0d0p0 path51

To repair path52 of a device whose alias is disk1:

sdipath -o repair -d disk1 path52