te
Copyright (c) 2003, Sun Microsystems, Inc. All Rights Reserved
The contents of this file are subject to the terms of the Common Development and Distribution License (the "License"). You may not use this file except in compliance with the License.
You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE or http://www.opensolaris.org/os/licensing. See the License for the specific language governing permissions and limitations under the License.
When distributing Covered Code, include this CDDL HEADER in each file and include the License file at usr/src/OPENSOLARIS.LICENSE. If applicable, add the following below this CDDL HEADER, with the fields enclosed by brackets "[]" replaced with your own identifying information: Portions Copyright [yyyy] [name of copyright owner]
MD 7D "Aug 29, 2003"
NAME
md - user configurable pseudo device driver
DESCRIPTION

md is a user configurable pseudo device driver that provides disk concatenation, striping, mirroring, RAID5 metadevices, trans metadevices, and hot spare utilities. Trans devices are no longer supported and have been replaced by UFS logging. See mount_ufs(1M).

The block devices access the disk using the system's normal buffering mechanism and are read and written without regard to physical disk records. There is also a ``raw'' device which provides for direct transmission between the disk and the user's read or write buffer. A single read or write call usually results in one I/O operation; raw I/O is therefore considerably more efficient when many bytes are transmitted. The names of the block devices are found in /dev/md/dsk; the names of the raw devices are found in /dev/md/rdsk. Metadevices have the appearance of whole disks; there are no slices (partitions).

I/O requests (such as lseek(2)) to the metadevices must have an offset that is a multiple of 512 bytes (DEV_BSIZE), or the driver returns an EINVAL error. If the transfer length is not a multiple of 512 bytes, the tranfer count is rounded up by the driver.

The md pseudo device drivers support all disk devices on all Solaris 2.4 or later Solaris systems.

IOCTLS

This section provides a list of the ioctls supported by the metadisk driver.

The following ioctls are valid when issued to the raw metadevice, such as /dev/md/rdsk/d0. See dkio(7I) for additional information. DKIOCGGEOM

This ioctl is used to get the disk geometry. The metadisk driver fills in the dkg_nhead, dkg_nsect, dkg_rpm, dkg_write_reinstruct and dkg_read_reinstruct from the first component of the metadevice (at metainit time). dkg_ncyl is calculated using the size of the metadevice (reported by metastat) divided by (dkg_nhead * dkg_nsect). The total size is always a multiple of (dkg_nhead * dkg_nsect). If the first component of a metadevice does not start on cylinder number 0, then the dkg_ncyl is increased by one cylinder; because DKIOCGVTOC reports the metadevice as starting on cylinder 1. The side effect here is that it looks like cylinder 0 is not being used, but all the arithmetic works out correctly. If the metadevice is not set up, then ENXIO is returned.

DKIOCINFO

When issued to the administrative device or metadevice, this ioctl sets dki_unit to the unit number of the metadevice, dki_ctype to a value of DKC_MD, and dki_partition to 0, because there are no slices.

DKIOCGVTOC

This ioctl returns the current vtoc. If one has not been written, then a default vtoc is returned. v_nparts is always 1. v_part[0].p_start is 0 if the first component of the metadevice starts on cylinder 0. Otherwise, the p_start field is the starting sector of cylinder 1. v_part[0].p_size is the same as the total size reported by metastat.

DKIOCSVTOC

This ioctl stores the vtoc in the metadevice state database so it is persistent across reboots.

DIAGNOSTICS
"Notice Log Messages"

The informative log messages include:

md: dnum: Hotspared device dev with dev

The first device name listed has been hot spare replaced with the second device name listed.

md: dnum: Hotspared device dev(num,num) with dev(num,num)

The first device number listed has been hot spare replaced with the second device number listed.

md: Could not load misc /dev

The named misc module is not loadable. It is possibly missing, or something else has been copied over it.

md: dnum: no mem for property dev

Memory could not be allocated in the prop_op entry point.

md: db: Parsing error on 'dev'

Set command in /kernel/drv/md.conf for the mddb.bootlist <number> is not in the correct format. metadb -p can be run to put the correct set commands into the /kernel/drv/md.conf file.

md: dnum: dev(num,num) needs maintenance
md: dnum: dev needs maintenance

An I/O or open error has occurred on a device within a mirror causing a component in the mirror to change to the Maintenance state.

md: dnum: dev(num,num) last erred md: dnum: dev last erred

An I/O or open error has occurred on a device within a mirror and the data is not replicated elsewhere in the mirror. This is causing the component in the mirror to change to the Last Erred state.

"Warning Log Messages"

The warning log messages include:

md: State database is stale

This error message comes when there are not enough usable replicas for the state database to be able to update records in the database. All accesses to the metadevice driver will fail. To fix this problem, more replicas need to be added or inaccessible replicas need to be deleted.

md: dnum: read error on devmd: dnum: write error on dev

A read or write error has occurred on the specified submirror, at the specified device name. This happens if any read or write errors occur on a submirror.

md: dnum: read error on dev(num,num)md: dnum: write error on dev(
num,num)

A read or write error has occurred on the specified submirror, at the specified device number. This happens if any read or write errors occur on a submirror.

md: State database commit failed
md: State database delete failed

These messages occur when there have been device errors on components where the state database replicas reside. These errors only occur when more than half of the replicas have had device errors returned to them. For instance, if you have three components with state database replicas and two of the components report errors, then these errors may occur. The state database commit or delete is retried periodically. If a replica is added, then the commit or delete will finish and the system will be operational. Otherwise the system will timeout and panic.

md: dnum: Cannot load dev driver

Underlying named driver module is not loadable (for example, sd, id, xy, or a third-party driver). This could indicate that the driver module has been removed.

md: Open error of hotspare devmd: Open error of hotspare dev(num,num)

Named hotspare is not openable, or underlying driver is not loadable.

"Panic Log Messages"

The panic log messages include:

md: dnum: Unknown close typemd: dnum: Unknown open type

Metadevice is being opened/closed with an unknown open type (OTYP).

md: State database problem

Failed metadevice state database commit or delete has been retried the default 100 times.

FILES
/dev/md/dsk/dn

block device (where n is the device number)

/dev/md/rdsk/dn

raw device (where n is the device number)

/dev/md/setname/dsk/dn

block device (where setname is the name of the diskset and n is the device number)

/dev/md/setname/rdsk/dn

raw device (where setname is the name of the diskset and n is the device number)

/dev/md/admin

administrative device

/kernel/drv/md

driver module

/kernel/drv/md.conf

driver configuration file

/kernel/misc/md_stripe

stripe driver misc module

/kernel/misc/md_mirror

mirror driver misc module

/kernel/misc/md_hotspares

hotspares driver misc module

/kernel/misc/md_trans

metatrans driver for UFS logging

/kernel/misc/md_raid

RAID5 driver misc module

SEE ALSO

mdmonitord(1M), metaclear(1M), metadb(1M), metadetach(1M), metahs(1M), metainit(1M), metaoffline(1M), metaonline(1M), metaparam(1M), metarecover(1M), metarename(1M), metareplace(1M), metaroot(1M), metassist(1M), metaset(1M), metastat(1M), metasync(1M), metattach(1M), md.cf(4), md.tab(4), attributes(5),

Solaris Volume Manager Administration Guide

NOTES

Trans metadevices have been replaced by UFS logging. Existing trans devices are not logging--they pass data directly through to the underlying device. See mount_ufs(1M) for more information about UFS logging.