freebsd-skq/sbin/gvinum/gvinum.8

447 lines
13 KiB
Groff
Raw Normal View History

.\" Copyright (c) 2005 Chris Jones
.\" All rights reserved.
.\"
.\" This software was developed for the FreeBSD Project by Chris Jones
.\" thanks to the support of Google's Summer of Code program and
.\" mentoring by Lukas Ertl.
.\"
.\" Redistribution and use in source and binary forms, with or without
.\" modification, are permitted provided that the following conditions
.\" are met:
.\" 1. Redistributions of source code must retain the above copyright
.\" notice, this list of conditions and the following disclaimer.
.\" 2. Redistributions in binary form must reproduce the above copyright
.\" notice, this list of conditions and the following disclaimer in the
.\" documentation and/or other materials provided with the distribution.
.\"
.\" THIS SOFTWARE IS PROVIDED BY AUTHOR AND CONTRIBUTORS ``AS IS'' AND
.\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
.\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
.\" ARE DISCLAIMED. IN NO EVENT SHALL AUTHOR OR CONTRIBUTORS BE LIABLE
.\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
.\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
.\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
.\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
.\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
.\" SUCH DAMAGE.
.\"
.\" $FreeBSD$
.\"
.Dd May 6, 2014
.Dt GVINUM 8
.Os
.Sh NAME
.Nm gvinum
.Nd Logical Volume Manager control program
.Sh SYNOPSIS
.Nm
.Op Ar command
.Op Fl options
.Sh COMMANDS
.Bl -tag -width indent
Import the gvinum work that have been done during and after Summer of Code 2007. The work have been under testing and fixing since then, and it is mature enough to be put into HEAD for further testing. A lot have changed in this time, and here are the most important: - Gvinum now uses one single workerthread instead of one thread for each volume and each plex. The reason for this is that the previous scheme was very complex, and was the cause of many of the bugs discovered in gvinum. Instead, gvinum now uses one worker thread with an event queue, quite similar to what used in gmirror. - The rebuild/grow/initialize/parity check routines no longer runs in separate threads, but are run as regular I/O requests with special flags. This made it easier to support mounted growing and parity rebuild. - Support for growing striped and raid5-plexes, meaning that one can extend the volumes for these plex types in addition to the concat type. Also works while the volume is mounted. - Implementation of many of the missing commands from the old vinum: attach/detach, start (was partially implemented), stop (was partially implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes with one plex of these organizations). - The parity check and rebuild no longer goes between userland/kernel, meaning that the gvinum command will not stay and wait forever for the rebuild to finish. You can instead watch the status with the list command. - Many problems with gvinum have been reported since 5.x, and some has been hard to fix due to the complicated architecture. Hopefully, it should be more stable and better handle edge cases that previously made gvinum crash. - Failed drives no longer disappears entirely, but now leave behind a dummy drive that makes sure the original state is not forgotten in case the system is rebooted between drive failures/swaps. - Update manpage to reflect new commands and extend it with some examples. Sponsored by: Google Summer of Code 2007 Mentored by: le Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
.It Ic attach Ar plex volume Op Cm rename
.It Ic attach Ar subdisk plex Oo Ar offset Oc Op Cm rename
Attach a plex to a volume, or a subdisk to a plex.
Import the gvinum work that have been done during and after Summer of Code 2007. The work have been under testing and fixing since then, and it is mature enough to be put into HEAD for further testing. A lot have changed in this time, and here are the most important: - Gvinum now uses one single workerthread instead of one thread for each volume and each plex. The reason for this is that the previous scheme was very complex, and was the cause of many of the bugs discovered in gvinum. Instead, gvinum now uses one worker thread with an event queue, quite similar to what used in gmirror. - The rebuild/grow/initialize/parity check routines no longer runs in separate threads, but are run as regular I/O requests with special flags. This made it easier to support mounted growing and parity rebuild. - Support for growing striped and raid5-plexes, meaning that one can extend the volumes for these plex types in addition to the concat type. Also works while the volume is mounted. - Implementation of many of the missing commands from the old vinum: attach/detach, start (was partially implemented), stop (was partially implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes with one plex of these organizations). - The parity check and rebuild no longer goes between userland/kernel, meaning that the gvinum command will not stay and wait forever for the rebuild to finish. You can instead watch the status with the list command. - Many problems with gvinum have been reported since 5.x, and some has been hard to fix due to the complicated architecture. Hopefully, it should be more stable and better handle edge cases that previously made gvinum crash. - Failed drives no longer disappears entirely, but now leave behind a dummy drive that makes sure the original state is not forgotten in case the system is rebooted between drive failures/swaps. - Update manpage to reflect new commands and extend it with some examples. Sponsored by: Google Summer of Code 2007 Mentored by: le Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
If offset is specified, the subdisk will be attached to the given offset within
the plex.
If rename is specified, the subdisk or plex will change name according to the
object it attaches to.
2006-09-30 11:02:17 +00:00
.It Ic checkparity Oo Fl f Oc Ar plex
Check the parity blocks of a RAID-5 plex.
The parity check will start at the
beginning of the plex if the
.Fl f
flag is specified, or otherwise at the location of the parity check pointer,
2006-09-30 11:02:17 +00:00
the first location at which plex's parity is incorrect.
All subdisks in the
plex must be up for a parity check.
Import the gvinum work that have been done during and after Summer of Code 2007. The work have been under testing and fixing since then, and it is mature enough to be put into HEAD for further testing. A lot have changed in this time, and here are the most important: - Gvinum now uses one single workerthread instead of one thread for each volume and each plex. The reason for this is that the previous scheme was very complex, and was the cause of many of the bugs discovered in gvinum. Instead, gvinum now uses one worker thread with an event queue, quite similar to what used in gmirror. - The rebuild/grow/initialize/parity check routines no longer runs in separate threads, but are run as regular I/O requests with special flags. This made it easier to support mounted growing and parity rebuild. - Support for growing striped and raid5-plexes, meaning that one can extend the volumes for these plex types in addition to the concat type. Also works while the volume is mounted. - Implementation of many of the missing commands from the old vinum: attach/detach, start (was partially implemented), stop (was partially implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes with one plex of these organizations). - The parity check and rebuild no longer goes between userland/kernel, meaning that the gvinum command will not stay and wait forever for the rebuild to finish. You can instead watch the status with the list command. - Many problems with gvinum have been reported since 5.x, and some has been hard to fix due to the complicated architecture. Hopefully, it should be more stable and better handle edge cases that previously made gvinum crash. - Failed drives no longer disappears entirely, but now leave behind a dummy drive that makes sure the original state is not forgotten in case the system is rebooted between drive failures/swaps. - Update manpage to reflect new commands and extend it with some examples. Sponsored by: Google Summer of Code 2007 Mentored by: le Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
.It Ic concat Oo Fl fv Oc Oo Fl n Ar name Oc Ar drives
Create a concatenated volume from the specified drives.
If no name is specified, a unique name will be set by
.Ic gvinum .
Import the gvinum work that have been done during and after Summer of Code 2007. The work have been under testing and fixing since then, and it is mature enough to be put into HEAD for further testing. A lot have changed in this time, and here are the most important: - Gvinum now uses one single workerthread instead of one thread for each volume and each plex. The reason for this is that the previous scheme was very complex, and was the cause of many of the bugs discovered in gvinum. Instead, gvinum now uses one worker thread with an event queue, quite similar to what used in gmirror. - The rebuild/grow/initialize/parity check routines no longer runs in separate threads, but are run as regular I/O requests with special flags. This made it easier to support mounted growing and parity rebuild. - Support for growing striped and raid5-plexes, meaning that one can extend the volumes for these plex types in addition to the concat type. Also works while the volume is mounted. - Implementation of many of the missing commands from the old vinum: attach/detach, start (was partially implemented), stop (was partially implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes with one plex of these organizations). - The parity check and rebuild no longer goes between userland/kernel, meaning that the gvinum command will not stay and wait forever for the rebuild to finish. You can instead watch the status with the list command. - Many problems with gvinum have been reported since 5.x, and some has been hard to fix due to the complicated architecture. Hopefully, it should be more stable and better handle edge cases that previously made gvinum crash. - Failed drives no longer disappears entirely, but now leave behind a dummy drive that makes sure the original state is not forgotten in case the system is rebooted between drive failures/swaps. - Update manpage to reflect new commands and extend it with some examples. Sponsored by: Google Summer of Code 2007 Mentored by: le Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
.It Ic create Oo Fl f Oc Op Ar description-file
Create a volume as described in
.Ar description-file .
If no
.Ar description-file
provided, opens an editor and provides the current
.Nm
configuration for editing.
Import the gvinum work that have been done during and after Summer of Code 2007. The work have been under testing and fixing since then, and it is mature enough to be put into HEAD for further testing. A lot have changed in this time, and here are the most important: - Gvinum now uses one single workerthread instead of one thread for each volume and each plex. The reason for this is that the previous scheme was very complex, and was the cause of many of the bugs discovered in gvinum. Instead, gvinum now uses one worker thread with an event queue, quite similar to what used in gmirror. - The rebuild/grow/initialize/parity check routines no longer runs in separate threads, but are run as regular I/O requests with special flags. This made it easier to support mounted growing and parity rebuild. - Support for growing striped and raid5-plexes, meaning that one can extend the volumes for these plex types in addition to the concat type. Also works while the volume is mounted. - Implementation of many of the missing commands from the old vinum: attach/detach, start (was partially implemented), stop (was partially implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes with one plex of these organizations). - The parity check and rebuild no longer goes between userland/kernel, meaning that the gvinum command will not stay and wait forever for the rebuild to finish. You can instead watch the status with the list command. - Many problems with gvinum have been reported since 5.x, and some has been hard to fix due to the complicated architecture. Hopefully, it should be more stable and better handle edge cases that previously made gvinum crash. - Failed drives no longer disappears entirely, but now leave behind a dummy drive that makes sure the original state is not forgotten in case the system is rebooted between drive failures/swaps. - Update manpage to reflect new commands and extend it with some examples. Sponsored by: Google Summer of Code 2007 Mentored by: le Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
The
.Fl f
flag will make gvinum ignore any errors regarding creating objects that already
exists.
However, in contrast to vinum, objects that are not properly named in the
.Ar description-file
will not be created when the
.Fl f
flag is given.
.It Ic detach Oo Fl f Oc Op Ar plex | subdisk
Detach a plex or subdisk from the volume or plex to which it is
attached.
.It Ic grow Ar plex device
Grow a plex by creating a gvinum drive and subdisk on device and attach it to
the plex.
If required by the plex organization, it will be put into the growable state.
2006-09-30 11:02:17 +00:00
.It Ic help
Provides a synopsis of
.Nm
commands and arguments.
2006-09-30 11:02:17 +00:00
.It Ic l | list Oo Fl rvV Oc Op Ar volume | plex | subdisk
.It Ic ld Oo Fl rvV Oc Op Ar drive ...
.It Ic ls Oo Fl rvV Oc Op Ar subdisk ...
.It Ic lp Oo Fl rvV Oc Op Ar plex ...
.It Ic lv Oo Fl rvV Oc Op Ar volume ...
List information about the relevant object(s).
The
.Fl r
flag provides recursive display, showing each object's subordinate objects in
2006-09-30 11:02:17 +00:00
proper relation.
The
.Fl v
and
.Fl V
flags provide progressively more detailed output.
Import the gvinum work that have been done during and after Summer of Code 2007. The work have been under testing and fixing since then, and it is mature enough to be put into HEAD for further testing. A lot have changed in this time, and here are the most important: - Gvinum now uses one single workerthread instead of one thread for each volume and each plex. The reason for this is that the previous scheme was very complex, and was the cause of many of the bugs discovered in gvinum. Instead, gvinum now uses one worker thread with an event queue, quite similar to what used in gmirror. - The rebuild/grow/initialize/parity check routines no longer runs in separate threads, but are run as regular I/O requests with special flags. This made it easier to support mounted growing and parity rebuild. - Support for growing striped and raid5-plexes, meaning that one can extend the volumes for these plex types in addition to the concat type. Also works while the volume is mounted. - Implementation of many of the missing commands from the old vinum: attach/detach, start (was partially implemented), stop (was partially implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes with one plex of these organizations). - The parity check and rebuild no longer goes between userland/kernel, meaning that the gvinum command will not stay and wait forever for the rebuild to finish. You can instead watch the status with the list command. - Many problems with gvinum have been reported since 5.x, and some has been hard to fix due to the complicated architecture. Hopefully, it should be more stable and better handle edge cases that previously made gvinum crash. - Failed drives no longer disappears entirely, but now leave behind a dummy drive that makes sure the original state is not forgotten in case the system is rebooted between drive failures/swaps. - Update manpage to reflect new commands and extend it with some examples. Sponsored by: Google Summer of Code 2007 Mentored by: le Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
.It Ic mirror Oo Fl fsv Oc Oo Fl n Ar name Oc Ar drives
Create a mirrored volume from the specified drives.
Import the gvinum work that have been done during and after Summer of Code 2007. The work have been under testing and fixing since then, and it is mature enough to be put into HEAD for further testing. A lot have changed in this time, and here are the most important: - Gvinum now uses one single workerthread instead of one thread for each volume and each plex. The reason for this is that the previous scheme was very complex, and was the cause of many of the bugs discovered in gvinum. Instead, gvinum now uses one worker thread with an event queue, quite similar to what used in gmirror. - The rebuild/grow/initialize/parity check routines no longer runs in separate threads, but are run as regular I/O requests with special flags. This made it easier to support mounted growing and parity rebuild. - Support for growing striped and raid5-plexes, meaning that one can extend the volumes for these plex types in addition to the concat type. Also works while the volume is mounted. - Implementation of many of the missing commands from the old vinum: attach/detach, start (was partially implemented), stop (was partially implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes with one plex of these organizations). - The parity check and rebuild no longer goes between userland/kernel, meaning that the gvinum command will not stay and wait forever for the rebuild to finish. You can instead watch the status with the list command. - Many problems with gvinum have been reported since 5.x, and some has been hard to fix due to the complicated architecture. Hopefully, it should be more stable and better handle edge cases that previously made gvinum crash. - Failed drives no longer disappears entirely, but now leave behind a dummy drive that makes sure the original state is not forgotten in case the system is rebooted between drive failures/swaps. - Update manpage to reflect new commands and extend it with some examples. Sponsored by: Google Summer of Code 2007 Mentored by: le Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
It requires at least a multiple of 2 drives.
If no name is specified, a unique name will be set by gvinum.
If the
Import the gvinum work that have been done during and after Summer of Code 2007. The work have been under testing and fixing since then, and it is mature enough to be put into HEAD for further testing. A lot have changed in this time, and here are the most important: - Gvinum now uses one single workerthread instead of one thread for each volume and each plex. The reason for this is that the previous scheme was very complex, and was the cause of many of the bugs discovered in gvinum. Instead, gvinum now uses one worker thread with an event queue, quite similar to what used in gmirror. - The rebuild/grow/initialize/parity check routines no longer runs in separate threads, but are run as regular I/O requests with special flags. This made it easier to support mounted growing and parity rebuild. - Support for growing striped and raid5-plexes, meaning that one can extend the volumes for these plex types in addition to the concat type. Also works while the volume is mounted. - Implementation of many of the missing commands from the old vinum: attach/detach, start (was partially implemented), stop (was partially implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes with one plex of these organizations). - The parity check and rebuild no longer goes between userland/kernel, meaning that the gvinum command will not stay and wait forever for the rebuild to finish. You can instead watch the status with the list command. - Many problems with gvinum have been reported since 5.x, and some has been hard to fix due to the complicated architecture. Hopefully, it should be more stable and better handle edge cases that previously made gvinum crash. - Failed drives no longer disappears entirely, but now leave behind a dummy drive that makes sure the original state is not forgotten in case the system is rebooted between drive failures/swaps. - Update manpage to reflect new commands and extend it with some examples. Sponsored by: Google Summer of Code 2007 Mentored by: le Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
.Fl s
flag is specified, a striped mirror will be created, and thus requires a
multiple of 4 drives.
2006-09-30 11:02:17 +00:00
.It Ic move | mv Fl f Ar drive subdisk Op Ar ...
Move the subdisk(s) to the specified drive.
The
.Fl f
flag is required, as all data on the indicated subdisk(s) will be destroyed as
2006-09-30 11:02:17 +00:00
part of the move.
This can currently only be done when the subdisk is
not being accessed.
.Pp
Import the gvinum work that have been done during and after Summer of Code 2007. The work have been under testing and fixing since then, and it is mature enough to be put into HEAD for further testing. A lot have changed in this time, and here are the most important: - Gvinum now uses one single workerthread instead of one thread for each volume and each plex. The reason for this is that the previous scheme was very complex, and was the cause of many of the bugs discovered in gvinum. Instead, gvinum now uses one worker thread with an event queue, quite similar to what used in gmirror. - The rebuild/grow/initialize/parity check routines no longer runs in separate threads, but are run as regular I/O requests with special flags. This made it easier to support mounted growing and parity rebuild. - Support for growing striped and raid5-plexes, meaning that one can extend the volumes for these plex types in addition to the concat type. Also works while the volume is mounted. - Implementation of many of the missing commands from the old vinum: attach/detach, start (was partially implemented), stop (was partially implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes with one plex of these organizations). - The parity check and rebuild no longer goes between userland/kernel, meaning that the gvinum command will not stay and wait forever for the rebuild to finish. You can instead watch the status with the list command. - Many problems with gvinum have been reported since 5.x, and some has been hard to fix due to the complicated architecture. Hopefully, it should be more stable and better handle edge cases that previously made gvinum crash. - Failed drives no longer disappears entirely, but now leave behind a dummy drive that makes sure the original state is not forgotten in case the system is rebooted between drive failures/swaps. - Update manpage to reflect new commands and extend it with some examples. Sponsored by: Google Summer of Code 2007 Mentored by: le Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
If a single subdisk is moved, and it forms a part of a RAID-5 plex, the moved
subdisks will need to be set to the
.Dq stale
state, and the plex will require a
.Ic start
command.
If multiple subdisk(s) is moved, and form part of a RAID-5 plex, the
moved disk(s) will need to be set to the
2006-09-30 11:02:17 +00:00
.Dq up
state and the plex will require a
.Ic rebuildparity
Import the gvinum work that have been done during and after Summer of Code 2007. The work have been under testing and fixing since then, and it is mature enough to be put into HEAD for further testing. A lot have changed in this time, and here are the most important: - Gvinum now uses one single workerthread instead of one thread for each volume and each plex. The reason for this is that the previous scheme was very complex, and was the cause of many of the bugs discovered in gvinum. Instead, gvinum now uses one worker thread with an event queue, quite similar to what used in gmirror. - The rebuild/grow/initialize/parity check routines no longer runs in separate threads, but are run as regular I/O requests with special flags. This made it easier to support mounted growing and parity rebuild. - Support for growing striped and raid5-plexes, meaning that one can extend the volumes for these plex types in addition to the concat type. Also works while the volume is mounted. - Implementation of many of the missing commands from the old vinum: attach/detach, start (was partially implemented), stop (was partially implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes with one plex of these organizations). - The parity check and rebuild no longer goes between userland/kernel, meaning that the gvinum command will not stay and wait forever for the rebuild to finish. You can instead watch the status with the list command. - Many problems with gvinum have been reported since 5.x, and some has been hard to fix due to the complicated architecture. Hopefully, it should be more stable and better handle edge cases that previously made gvinum crash. - Failed drives no longer disappears entirely, but now leave behind a dummy drive that makes sure the original state is not forgotten in case the system is rebooted between drive failures/swaps. - Update manpage to reflect new commands and extend it with some examples. Sponsored by: Google Summer of Code 2007 Mentored by: le Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
command.
If the subdisk(s) form part of a plex that is mirrored with other
2006-09-30 11:02:17 +00:00
plexes, the plex will require restarting and will sync once restarted.
Moving
more than one subdisk in a RAID-5 plex or subdisks from both sides of a
2006-09-30 11:02:17 +00:00
mirrored plex volume will destroy data.
Note that parity rebuilds and syncing
must be started manually after a move.
2006-09-30 11:02:17 +00:00
.It Ic printconfig
Write a copy of the current configuration to standard output.
2006-09-30 11:02:17 +00:00
.It Ic quit
Exit
.Nm
2006-09-30 11:02:17 +00:00
when running in interactive mode.
Normally this would be done by entering the
EOF character.
Import the gvinum work that have been done during and after Summer of Code 2007. The work have been under testing and fixing since then, and it is mature enough to be put into HEAD for further testing. A lot have changed in this time, and here are the most important: - Gvinum now uses one single workerthread instead of one thread for each volume and each plex. The reason for this is that the previous scheme was very complex, and was the cause of many of the bugs discovered in gvinum. Instead, gvinum now uses one worker thread with an event queue, quite similar to what used in gmirror. - The rebuild/grow/initialize/parity check routines no longer runs in separate threads, but are run as regular I/O requests with special flags. This made it easier to support mounted growing and parity rebuild. - Support for growing striped and raid5-plexes, meaning that one can extend the volumes for these plex types in addition to the concat type. Also works while the volume is mounted. - Implementation of many of the missing commands from the old vinum: attach/detach, start (was partially implemented), stop (was partially implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes with one plex of these organizations). - The parity check and rebuild no longer goes between userland/kernel, meaning that the gvinum command will not stay and wait forever for the rebuild to finish. You can instead watch the status with the list command. - Many problems with gvinum have been reported since 5.x, and some has been hard to fix due to the complicated architecture. Hopefully, it should be more stable and better handle edge cases that previously made gvinum crash. - Failed drives no longer disappears entirely, but now leave behind a dummy drive that makes sure the original state is not forgotten in case the system is rebooted between drive failures/swaps. - Update manpage to reflect new commands and extend it with some examples. Sponsored by: Google Summer of Code 2007 Mentored by: le Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
.It Ic raid5 Oo Fl fv Oc Oo Fl s Ar stripesize Oc Oo Fl n Ar name Oc Ar drives
Create a RAID-5 volume from the specified drives.
If no name is specified, a unique name will be set by
.Ic gvinum .
Import the gvinum work that have been done during and after Summer of Code 2007. The work have been under testing and fixing since then, and it is mature enough to be put into HEAD for further testing. A lot have changed in this time, and here are the most important: - Gvinum now uses one single workerthread instead of one thread for each volume and each plex. The reason for this is that the previous scheme was very complex, and was the cause of many of the bugs discovered in gvinum. Instead, gvinum now uses one worker thread with an event queue, quite similar to what used in gmirror. - The rebuild/grow/initialize/parity check routines no longer runs in separate threads, but are run as regular I/O requests with special flags. This made it easier to support mounted growing and parity rebuild. - Support for growing striped and raid5-plexes, meaning that one can extend the volumes for these plex types in addition to the concat type. Also works while the volume is mounted. - Implementation of many of the missing commands from the old vinum: attach/detach, start (was partially implemented), stop (was partially implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes with one plex of these organizations). - The parity check and rebuild no longer goes between userland/kernel, meaning that the gvinum command will not stay and wait forever for the rebuild to finish. You can instead watch the status with the list command. - Many problems with gvinum have been reported since 5.x, and some has been hard to fix due to the complicated architecture. Hopefully, it should be more stable and better handle edge cases that previously made gvinum crash. - Failed drives no longer disappears entirely, but now leave behind a dummy drive that makes sure the original state is not forgotten in case the system is rebooted between drive failures/swaps. - Update manpage to reflect new commands and extend it with some examples. Sponsored by: Google Summer of Code 2007 Mentored by: le Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
This organization requires at least three drives.
2006-09-30 11:02:17 +00:00
.It Ic rename Oo Fl r Oc Ar drive | subdisk | plex | volume newname
Change the name of the specified object.
The
.Fl r
flag will recursively rename subordinate objects.
.Pp
Note that device nodes will not be renamed until
.Nm
is restarted.
2006-09-30 11:02:17 +00:00
.It Ic rebuildparity Oo Fl f Oc Ar plex
Rebuild the parity blocks of a RAID-5 plex.
The parity rebuild will start at
the beginning of the plex if the
.Fl f
flag is specified, or otherwise at the location of the parity check pointer.
All subdisks in the plex must be up for a parity check.
.It Ic resetconfig Oo Fl f Oc
Reset the complete
.Nm
configuration.
2006-09-30 11:02:17 +00:00
.It Ic rm Oo Fl r Oc Ar volume | plex | subdisk
Remove an object and, if
.Fl r
is specified, its subordinate objects.
2006-09-30 11:02:17 +00:00
.It Ic saveconfig
Save
.Nm
configuration to disk after configuration failures.
2006-09-30 11:02:17 +00:00
.It Ic setstate Oo Fl f Oc Ar state volume | plex | subdisk | drive
Set state without influencing other objects, for diagnostic purposes
2006-09-30 11:02:17 +00:00
only.
The
.Fl f
flag forces state changes regardless of whether they are legal.
2006-09-30 11:02:17 +00:00
.It Ic start
Read configuration from all vinum drives.
2006-09-30 11:02:17 +00:00
.It Ic start Oo Fl S Ar size Oc Ar volume | plex | subdisk
Allow the system to access the objects.
Import the gvinum work that have been done during and after Summer of Code 2007. The work have been under testing and fixing since then, and it is mature enough to be put into HEAD for further testing. A lot have changed in this time, and here are the most important: - Gvinum now uses one single workerthread instead of one thread for each volume and each plex. The reason for this is that the previous scheme was very complex, and was the cause of many of the bugs discovered in gvinum. Instead, gvinum now uses one worker thread with an event queue, quite similar to what used in gmirror. - The rebuild/grow/initialize/parity check routines no longer runs in separate threads, but are run as regular I/O requests with special flags. This made it easier to support mounted growing and parity rebuild. - Support for growing striped and raid5-plexes, meaning that one can extend the volumes for these plex types in addition to the concat type. Also works while the volume is mounted. - Implementation of many of the missing commands from the old vinum: attach/detach, start (was partially implemented), stop (was partially implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes with one plex of these organizations). - The parity check and rebuild no longer goes between userland/kernel, meaning that the gvinum command will not stay and wait forever for the rebuild to finish. You can instead watch the status with the list command. - Many problems with gvinum have been reported since 5.x, and some has been hard to fix due to the complicated architecture. Hopefully, it should be more stable and better handle edge cases that previously made gvinum crash. - Failed drives no longer disappears entirely, but now leave behind a dummy drive that makes sure the original state is not forgotten in case the system is rebooted between drive failures/swaps. - Update manpage to reflect new commands and extend it with some examples. Sponsored by: Google Summer of Code 2007 Mentored by: le Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
If necessary, plexes will be synced and rebuilt.
If a subdisk was added to a running RAID-5 or striped plex, gvinum will
expand into this subdisk and grow the whole RAID-5 array.
This can be done without unmounting your filesystem.
2006-09-30 11:02:17 +00:00
The
.Fl S
flag is currently ignored.
Import the gvinum work that have been done during and after Summer of Code 2007. The work have been under testing and fixing since then, and it is mature enough to be put into HEAD for further testing. A lot have changed in this time, and here are the most important: - Gvinum now uses one single workerthread instead of one thread for each volume and each plex. The reason for this is that the previous scheme was very complex, and was the cause of many of the bugs discovered in gvinum. Instead, gvinum now uses one worker thread with an event queue, quite similar to what used in gmirror. - The rebuild/grow/initialize/parity check routines no longer runs in separate threads, but are run as regular I/O requests with special flags. This made it easier to support mounted growing and parity rebuild. - Support for growing striped and raid5-plexes, meaning that one can extend the volumes for these plex types in addition to the concat type. Also works while the volume is mounted. - Implementation of many of the missing commands from the old vinum: attach/detach, start (was partially implemented), stop (was partially implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes with one plex of these organizations). - The parity check and rebuild no longer goes between userland/kernel, meaning that the gvinum command will not stay and wait forever for the rebuild to finish. You can instead watch the status with the list command. - Many problems with gvinum have been reported since 5.x, and some has been hard to fix due to the complicated architecture. Hopefully, it should be more stable and better handle edge cases that previously made gvinum crash. - Failed drives no longer disappears entirely, but now leave behind a dummy drive that makes sure the original state is not forgotten in case the system is rebooted between drive failures/swaps. - Update manpage to reflect new commands and extend it with some examples. Sponsored by: Google Summer of Code 2007 Mentored by: le Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
.It Ic stop Oo Fl f Oc Op Ar volume | plex | subdisk
Terminate access to the objects, or stop
.Nm
if no parameters are specified.
.It Ic stripe Oo Fl fv Oc Oo Fl n Ar name Oc Ar drives
Create a striped volume from the specified drives.
If no name is specified, a unique name will be set by
.Ic gvinum .
This organization requires at least two drives.
.El
.Sh DESCRIPTION
The
.Nm
utility communicates with the kernel component of the GVinum logical volume
2006-09-30 11:02:17 +00:00
manager.
It is designed either for interactive use, when started without
command line arguments, or to execute a single command if the command is
2006-09-30 11:02:17 +00:00
supplied on the command line.
In interactive mode,
.Nm
maintains a command line history.
.Sh OPTIONS
2006-09-30 11:02:17 +00:00
The
.Nm
commands may be followed by an option.
.Bl -tag -width indent
.It Fl f
The
.Fl f
.Pq Dq force
2006-09-30 11:02:17 +00:00
option overrides safety checks.
It should be used with extreme caution.
This
option is required in order to use the
.Ic move
command.
.It Fl r
The
.Fl r
.Pq Dq recursive
2006-09-30 11:02:17 +00:00
option applies the command recursively to subordinate objects.
For example, in
conjunction with the
.Ic lv
command, the
.Fl r
option will also show information about the plexes and subdisks belonging to
2006-09-30 11:02:17 +00:00
the volume.
It is also used by the
.Ic rename
command to indicate that subordinate objects such as subdisks should be renamed
to match the object(s) specified and by the
.Ic rm
command to delete plexes belonging to a volume and so on.
.It Fl v
The
.Fl v
.Pq Dq verbose
option provides more detailed output.
.It Fl V
The
.Fl V
2006-09-30 11:02:17 +00:00
.Pq Dq "very verbose"
option provides even more detailed output than
.Fl v .
.El
2006-09-30 11:02:17 +00:00
.Sh ENVIRONMENT
.Bl -tag -width ".Ev EDITOR"
.It Ev EDITOR
The name of the editor to use for editing configuration files, by
default
.Xr vi 1
is invoked.
.El
.Sh FILES
2006-09-30 11:02:17 +00:00
.Bl -tag -width ".Pa /dev/gvinum/plex"
.It Pa /dev/gvinum
directory with device nodes for
.Nm
objects
.El
Import the gvinum work that have been done during and after Summer of Code 2007. The work have been under testing and fixing since then, and it is mature enough to be put into HEAD for further testing. A lot have changed in this time, and here are the most important: - Gvinum now uses one single workerthread instead of one thread for each volume and each plex. The reason for this is that the previous scheme was very complex, and was the cause of many of the bugs discovered in gvinum. Instead, gvinum now uses one worker thread with an event queue, quite similar to what used in gmirror. - The rebuild/grow/initialize/parity check routines no longer runs in separate threads, but are run as regular I/O requests with special flags. This made it easier to support mounted growing and parity rebuild. - Support for growing striped and raid5-plexes, meaning that one can extend the volumes for these plex types in addition to the concat type. Also works while the volume is mounted. - Implementation of many of the missing commands from the old vinum: attach/detach, start (was partially implemented), stop (was partially implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes with one plex of these organizations). - The parity check and rebuild no longer goes between userland/kernel, meaning that the gvinum command will not stay and wait forever for the rebuild to finish. You can instead watch the status with the list command. - Many problems with gvinum have been reported since 5.x, and some has been hard to fix due to the complicated architecture. Hopefully, it should be more stable and better handle edge cases that previously made gvinum crash. - Failed drives no longer disappears entirely, but now leave behind a dummy drive that makes sure the original state is not forgotten in case the system is rebooted between drive failures/swaps. - Update manpage to reflect new commands and extend it with some examples. Sponsored by: Google Summer of Code 2007 Mentored by: le Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
.Sh EXAMPLES
To create a mirror on disks /dev/ada1 and /dev/ada2, create a filesystem,
mount, unmount and then stop
.Ic gvinum :
Import the gvinum work that have been done during and after Summer of Code 2007. The work have been under testing and fixing since then, and it is mature enough to be put into HEAD for further testing. A lot have changed in this time, and here are the most important: - Gvinum now uses one single workerthread instead of one thread for each volume and each plex. The reason for this is that the previous scheme was very complex, and was the cause of many of the bugs discovered in gvinum. Instead, gvinum now uses one worker thread with an event queue, quite similar to what used in gmirror. - The rebuild/grow/initialize/parity check routines no longer runs in separate threads, but are run as regular I/O requests with special flags. This made it easier to support mounted growing and parity rebuild. - Support for growing striped and raid5-plexes, meaning that one can extend the volumes for these plex types in addition to the concat type. Also works while the volume is mounted. - Implementation of many of the missing commands from the old vinum: attach/detach, start (was partially implemented), stop (was partially implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes with one plex of these organizations). - The parity check and rebuild no longer goes between userland/kernel, meaning that the gvinum command will not stay and wait forever for the rebuild to finish. You can instead watch the status with the list command. - Many problems with gvinum have been reported since 5.x, and some has been hard to fix due to the complicated architecture. Hopefully, it should be more stable and better handle edge cases that previously made gvinum crash. - Failed drives no longer disappears entirely, but now leave behind a dummy drive that makes sure the original state is not forgotten in case the system is rebooted between drive failures/swaps. - Update manpage to reflect new commands and extend it with some examples. Sponsored by: Google Summer of Code 2007 Mentored by: le Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
.Pp
.Dl "gvinum mirror /dev/ada1 /dev/ada2"
Import the gvinum work that have been done during and after Summer of Code 2007. The work have been under testing and fixing since then, and it is mature enough to be put into HEAD for further testing. A lot have changed in this time, and here are the most important: - Gvinum now uses one single workerthread instead of one thread for each volume and each plex. The reason for this is that the previous scheme was very complex, and was the cause of many of the bugs discovered in gvinum. Instead, gvinum now uses one worker thread with an event queue, quite similar to what used in gmirror. - The rebuild/grow/initialize/parity check routines no longer runs in separate threads, but are run as regular I/O requests with special flags. This made it easier to support mounted growing and parity rebuild. - Support for growing striped and raid5-plexes, meaning that one can extend the volumes for these plex types in addition to the concat type. Also works while the volume is mounted. - Implementation of many of the missing commands from the old vinum: attach/detach, start (was partially implemented), stop (was partially implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes with one plex of these organizations). - The parity check and rebuild no longer goes between userland/kernel, meaning that the gvinum command will not stay and wait forever for the rebuild to finish. You can instead watch the status with the list command. - Many problems with gvinum have been reported since 5.x, and some has been hard to fix due to the complicated architecture. Hopefully, it should be more stable and better handle edge cases that previously made gvinum crash. - Failed drives no longer disappears entirely, but now leave behind a dummy drive that makes sure the original state is not forgotten in case the system is rebooted between drive failures/swaps. - Update manpage to reflect new commands and extend it with some examples. Sponsored by: Google Summer of Code 2007 Mentored by: le Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
.Dl "newfs /dev/gvinum/gvinumvolume0"
.Dl "mount /dev/gvinum/gvinumvolume0 /mnt"
Import the gvinum work that have been done during and after Summer of Code 2007. The work have been under testing and fixing since then, and it is mature enough to be put into HEAD for further testing. A lot have changed in this time, and here are the most important: - Gvinum now uses one single workerthread instead of one thread for each volume and each plex. The reason for this is that the previous scheme was very complex, and was the cause of many of the bugs discovered in gvinum. Instead, gvinum now uses one worker thread with an event queue, quite similar to what used in gmirror. - The rebuild/grow/initialize/parity check routines no longer runs in separate threads, but are run as regular I/O requests with special flags. This made it easier to support mounted growing and parity rebuild. - Support for growing striped and raid5-plexes, meaning that one can extend the volumes for these plex types in addition to the concat type. Also works while the volume is mounted. - Implementation of many of the missing commands from the old vinum: attach/detach, start (was partially implemented), stop (was partially implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes with one plex of these organizations). - The parity check and rebuild no longer goes between userland/kernel, meaning that the gvinum command will not stay and wait forever for the rebuild to finish. You can instead watch the status with the list command. - Many problems with gvinum have been reported since 5.x, and some has been hard to fix due to the complicated architecture. Hopefully, it should be more stable and better handle edge cases that previously made gvinum crash. - Failed drives no longer disappears entirely, but now leave behind a dummy drive that makes sure the original state is not forgotten in case the system is rebooted between drive failures/swaps. - Update manpage to reflect new commands and extend it with some examples. Sponsored by: Google Summer of Code 2007 Mentored by: le Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
.Dl "..."
.Dl "unmount /mnt"
.Dl "gvinum stop"
.Pp
To create a striped mirror on disks /dev/ada1 /dev/ada2 /dev/ada3 and
/dev/ada4 named "data" and create a filesystem:
Import the gvinum work that have been done during and after Summer of Code 2007. The work have been under testing and fixing since then, and it is mature enough to be put into HEAD for further testing. A lot have changed in this time, and here are the most important: - Gvinum now uses one single workerthread instead of one thread for each volume and each plex. The reason for this is that the previous scheme was very complex, and was the cause of many of the bugs discovered in gvinum. Instead, gvinum now uses one worker thread with an event queue, quite similar to what used in gmirror. - The rebuild/grow/initialize/parity check routines no longer runs in separate threads, but are run as regular I/O requests with special flags. This made it easier to support mounted growing and parity rebuild. - Support for growing striped and raid5-plexes, meaning that one can extend the volumes for these plex types in addition to the concat type. Also works while the volume is mounted. - Implementation of many of the missing commands from the old vinum: attach/detach, start (was partially implemented), stop (was partially implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes with one plex of these organizations). - The parity check and rebuild no longer goes between userland/kernel, meaning that the gvinum command will not stay and wait forever for the rebuild to finish. You can instead watch the status with the list command. - Many problems with gvinum have been reported since 5.x, and some has been hard to fix due to the complicated architecture. Hopefully, it should be more stable and better handle edge cases that previously made gvinum crash. - Failed drives no longer disappears entirely, but now leave behind a dummy drive that makes sure the original state is not forgotten in case the system is rebooted between drive failures/swaps. - Update manpage to reflect new commands and extend it with some examples. Sponsored by: Google Summer of Code 2007 Mentored by: le Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
.Pp
.Dl "gvinum mirror -s -n data /dev/ada1 /dev/ada2 /dev/ada3 /dev/ada4"
Import the gvinum work that have been done during and after Summer of Code 2007. The work have been under testing and fixing since then, and it is mature enough to be put into HEAD for further testing. A lot have changed in this time, and here are the most important: - Gvinum now uses one single workerthread instead of one thread for each volume and each plex. The reason for this is that the previous scheme was very complex, and was the cause of many of the bugs discovered in gvinum. Instead, gvinum now uses one worker thread with an event queue, quite similar to what used in gmirror. - The rebuild/grow/initialize/parity check routines no longer runs in separate threads, but are run as regular I/O requests with special flags. This made it easier to support mounted growing and parity rebuild. - Support for growing striped and raid5-plexes, meaning that one can extend the volumes for these plex types in addition to the concat type. Also works while the volume is mounted. - Implementation of many of the missing commands from the old vinum: attach/detach, start (was partially implemented), stop (was partially implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes with one plex of these organizations). - The parity check and rebuild no longer goes between userland/kernel, meaning that the gvinum command will not stay and wait forever for the rebuild to finish. You can instead watch the status with the list command. - Many problems with gvinum have been reported since 5.x, and some has been hard to fix due to the complicated architecture. Hopefully, it should be more stable and better handle edge cases that previously made gvinum crash. - Failed drives no longer disappears entirely, but now leave behind a dummy drive that makes sure the original state is not forgotten in case the system is rebooted between drive failures/swaps. - Update manpage to reflect new commands and extend it with some examples. Sponsored by: Google Summer of Code 2007 Mentored by: le Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
.Dl "newfs /dev/gvinum/data"
.Pp
To create a raid5 array on disks /dev/ada1 /dev/ada2 and /dev/ada3,
with stripesize 493k you can use the raid5 command:
Import the gvinum work that have been done during and after Summer of Code 2007. The work have been under testing and fixing since then, and it is mature enough to be put into HEAD for further testing. A lot have changed in this time, and here are the most important: - Gvinum now uses one single workerthread instead of one thread for each volume and each plex. The reason for this is that the previous scheme was very complex, and was the cause of many of the bugs discovered in gvinum. Instead, gvinum now uses one worker thread with an event queue, quite similar to what used in gmirror. - The rebuild/grow/initialize/parity check routines no longer runs in separate threads, but are run as regular I/O requests with special flags. This made it easier to support mounted growing and parity rebuild. - Support for growing striped and raid5-plexes, meaning that one can extend the volumes for these plex types in addition to the concat type. Also works while the volume is mounted. - Implementation of many of the missing commands from the old vinum: attach/detach, start (was partially implemented), stop (was partially implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes with one plex of these organizations). - The parity check and rebuild no longer goes between userland/kernel, meaning that the gvinum command will not stay and wait forever for the rebuild to finish. You can instead watch the status with the list command. - Many problems with gvinum have been reported since 5.x, and some has been hard to fix due to the complicated architecture. Hopefully, it should be more stable and better handle edge cases that previously made gvinum crash. - Failed drives no longer disappears entirely, but now leave behind a dummy drive that makes sure the original state is not forgotten in case the system is rebooted between drive failures/swaps. - Update manpage to reflect new commands and extend it with some examples. Sponsored by: Google Summer of Code 2007 Mentored by: le Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
.Pp
.Dl "gvinum raid5 -s 493k /dev/ada1 /dev/ada2 /dev/ada3"
Import the gvinum work that have been done during and after Summer of Code 2007. The work have been under testing and fixing since then, and it is mature enough to be put into HEAD for further testing. A lot have changed in this time, and here are the most important: - Gvinum now uses one single workerthread instead of one thread for each volume and each plex. The reason for this is that the previous scheme was very complex, and was the cause of many of the bugs discovered in gvinum. Instead, gvinum now uses one worker thread with an event queue, quite similar to what used in gmirror. - The rebuild/grow/initialize/parity check routines no longer runs in separate threads, but are run as regular I/O requests with special flags. This made it easier to support mounted growing and parity rebuild. - Support for growing striped and raid5-plexes, meaning that one can extend the volumes for these plex types in addition to the concat type. Also works while the volume is mounted. - Implementation of many of the missing commands from the old vinum: attach/detach, start (was partially implemented), stop (was partially implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes with one plex of these organizations). - The parity check and rebuild no longer goes between userland/kernel, meaning that the gvinum command will not stay and wait forever for the rebuild to finish. You can instead watch the status with the list command. - Many problems with gvinum have been reported since 5.x, and some has been hard to fix due to the complicated architecture. Hopefully, it should be more stable and better handle edge cases that previously made gvinum crash. - Failed drives no longer disappears entirely, but now leave behind a dummy drive that makes sure the original state is not forgotten in case the system is rebooted between drive failures/swaps. - Update manpage to reflect new commands and extend it with some examples. Sponsored by: Google Summer of Code 2007 Mentored by: le Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
.Pp
Then the volume will be created automatically.
Afterwards, you have to initialize the volume:
.Pp
.Dl "gvinum start myraid5vol"
.Pp
The initialization will start, and the states will be updated when it's
finished.
The list command will give you information about its progress.
.Pp
Imagine that one of the drives fails, and the output of 'printconfig' looks
something like this:
.Pp
.Dl "drive gvinumdrive1 device /dev/ada2"
Import the gvinum work that have been done during and after Summer of Code 2007. The work have been under testing and fixing since then, and it is mature enough to be put into HEAD for further testing. A lot have changed in this time, and here are the most important: - Gvinum now uses one single workerthread instead of one thread for each volume and each plex. The reason for this is that the previous scheme was very complex, and was the cause of many of the bugs discovered in gvinum. Instead, gvinum now uses one worker thread with an event queue, quite similar to what used in gmirror. - The rebuild/grow/initialize/parity check routines no longer runs in separate threads, but are run as regular I/O requests with special flags. This made it easier to support mounted growing and parity rebuild. - Support for growing striped and raid5-plexes, meaning that one can extend the volumes for these plex types in addition to the concat type. Also works while the volume is mounted. - Implementation of many of the missing commands from the old vinum: attach/detach, start (was partially implemented), stop (was partially implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes with one plex of these organizations). - The parity check and rebuild no longer goes between userland/kernel, meaning that the gvinum command will not stay and wait forever for the rebuild to finish. You can instead watch the status with the list command. - Many problems with gvinum have been reported since 5.x, and some has been hard to fix due to the complicated architecture. Hopefully, it should be more stable and better handle edge cases that previously made gvinum crash. - Failed drives no longer disappears entirely, but now leave behind a dummy drive that makes sure the original state is not forgotten in case the system is rebooted between drive failures/swaps. - Update manpage to reflect new commands and extend it with some examples. Sponsored by: Google Summer of Code 2007 Mentored by: le Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
.Dl "drive gvinumdrive2 device /dev/???"
.Dl "drive gvinumdrive0 device /dev/ada1"
Import the gvinum work that have been done during and after Summer of Code 2007. The work have been under testing and fixing since then, and it is mature enough to be put into HEAD for further testing. A lot have changed in this time, and here are the most important: - Gvinum now uses one single workerthread instead of one thread for each volume and each plex. The reason for this is that the previous scheme was very complex, and was the cause of many of the bugs discovered in gvinum. Instead, gvinum now uses one worker thread with an event queue, quite similar to what used in gmirror. - The rebuild/grow/initialize/parity check routines no longer runs in separate threads, but are run as regular I/O requests with special flags. This made it easier to support mounted growing and parity rebuild. - Support for growing striped and raid5-plexes, meaning that one can extend the volumes for these plex types in addition to the concat type. Also works while the volume is mounted. - Implementation of many of the missing commands from the old vinum: attach/detach, start (was partially implemented), stop (was partially implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes with one plex of these organizations). - The parity check and rebuild no longer goes between userland/kernel, meaning that the gvinum command will not stay and wait forever for the rebuild to finish. You can instead watch the status with the list command. - Many problems with gvinum have been reported since 5.x, and some has been hard to fix due to the complicated architecture. Hopefully, it should be more stable and better handle edge cases that previously made gvinum crash. - Failed drives no longer disappears entirely, but now leave behind a dummy drive that makes sure the original state is not forgotten in case the system is rebooted between drive failures/swaps. - Update manpage to reflect new commands and extend it with some examples. Sponsored by: Google Summer of Code 2007 Mentored by: le Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
.Dl "volume myraid5vol"
.Dl "plex name myraid5vol.p0 org raid5 986s vol myraid5vol"
.Dl "sd name myraid5vol.p0.s2 drive gvinumdrive2 len 32538s driveoffset 265s"
.Dl "plex myraid5vol.p0 plexoffset 1972s"
.Dl "sd name myraid5vol.p0.s1 drive gvinumdrive1 len 32538s driveoffset 265s"
.Dl "plex myraid5vol.p0 plexoffset 986s"
.Dl "sd name myraid5vol.p0.s0 drive gvinumdrive0 len 32538s driveoffset 265s"
.Dl "plex myraid5vol.p0 plexoffset 0s"
.Pp
Create a new drive with this configuration:
.Pp
.Dl "drive gdrive4 device /dev/ada4"
Import the gvinum work that have been done during and after Summer of Code 2007. The work have been under testing and fixing since then, and it is mature enough to be put into HEAD for further testing. A lot have changed in this time, and here are the most important: - Gvinum now uses one single workerthread instead of one thread for each volume and each plex. The reason for this is that the previous scheme was very complex, and was the cause of many of the bugs discovered in gvinum. Instead, gvinum now uses one worker thread with an event queue, quite similar to what used in gmirror. - The rebuild/grow/initialize/parity check routines no longer runs in separate threads, but are run as regular I/O requests with special flags. This made it easier to support mounted growing and parity rebuild. - Support for growing striped and raid5-plexes, meaning that one can extend the volumes for these plex types in addition to the concat type. Also works while the volume is mounted. - Implementation of many of the missing commands from the old vinum: attach/detach, start (was partially implemented), stop (was partially implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes with one plex of these organizations). - The parity check and rebuild no longer goes between userland/kernel, meaning that the gvinum command will not stay and wait forever for the rebuild to finish. You can instead watch the status with the list command. - Many problems with gvinum have been reported since 5.x, and some has been hard to fix due to the complicated architecture. Hopefully, it should be more stable and better handle edge cases that previously made gvinum crash. - Failed drives no longer disappears entirely, but now leave behind a dummy drive that makes sure the original state is not forgotten in case the system is rebooted between drive failures/swaps. - Update manpage to reflect new commands and extend it with some examples. Sponsored by: Google Summer of Code 2007 Mentored by: le Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
.Pp
Then move the stale subdisk to the new drive:
.Pp
.Dl "gvinum move gdrive4 myraid5vol.p0.s2"
.Pp
Then, initiate the rebuild:
.Pp
.Dl "gvinum start myraid5vol.p0"
.Pp
The plex will go up form degraded mode after the rebuild is finished.
The plex can still be used while the rebuild is in progress, although requests
might be delayed.
.Pp
Given the configuration as in the previous example, growing a RAID-5 or STRIPED
array is accomplished by using the grow command:
Import the gvinum work that have been done during and after Summer of Code 2007. The work have been under testing and fixing since then, and it is mature enough to be put into HEAD for further testing. A lot have changed in this time, and here are the most important: - Gvinum now uses one single workerthread instead of one thread for each volume and each plex. The reason for this is that the previous scheme was very complex, and was the cause of many of the bugs discovered in gvinum. Instead, gvinum now uses one worker thread with an event queue, quite similar to what used in gmirror. - The rebuild/grow/initialize/parity check routines no longer runs in separate threads, but are run as regular I/O requests with special flags. This made it easier to support mounted growing and parity rebuild. - Support for growing striped and raid5-plexes, meaning that one can extend the volumes for these plex types in addition to the concat type. Also works while the volume is mounted. - Implementation of many of the missing commands from the old vinum: attach/detach, start (was partially implemented), stop (was partially implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes with one plex of these organizations). - The parity check and rebuild no longer goes between userland/kernel, meaning that the gvinum command will not stay and wait forever for the rebuild to finish. You can instead watch the status with the list command. - Many problems with gvinum have been reported since 5.x, and some has been hard to fix due to the complicated architecture. Hopefully, it should be more stable and better handle edge cases that previously made gvinum crash. - Failed drives no longer disappears entirely, but now leave behind a dummy drive that makes sure the original state is not forgotten in case the system is rebooted between drive failures/swaps. - Update manpage to reflect new commands and extend it with some examples. Sponsored by: Google Summer of Code 2007 Mentored by: le Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
.Pp
.Dl "gvinum grow myraid5vol.p0 /dev/ada4"
Import the gvinum work that have been done during and after Summer of Code 2007. The work have been under testing and fixing since then, and it is mature enough to be put into HEAD for further testing. A lot have changed in this time, and here are the most important: - Gvinum now uses one single workerthread instead of one thread for each volume and each plex. The reason for this is that the previous scheme was very complex, and was the cause of many of the bugs discovered in gvinum. Instead, gvinum now uses one worker thread with an event queue, quite similar to what used in gmirror. - The rebuild/grow/initialize/parity check routines no longer runs in separate threads, but are run as regular I/O requests with special flags. This made it easier to support mounted growing and parity rebuild. - Support for growing striped and raid5-plexes, meaning that one can extend the volumes for these plex types in addition to the concat type. Also works while the volume is mounted. - Implementation of many of the missing commands from the old vinum: attach/detach, start (was partially implemented), stop (was partially implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes with one plex of these organizations). - The parity check and rebuild no longer goes between userland/kernel, meaning that the gvinum command will not stay and wait forever for the rebuild to finish. You can instead watch the status with the list command. - Many problems with gvinum have been reported since 5.x, and some has been hard to fix due to the complicated architecture. Hopefully, it should be more stable and better handle edge cases that previously made gvinum crash. - Failed drives no longer disappears entirely, but now leave behind a dummy drive that makes sure the original state is not forgotten in case the system is rebooted between drive failures/swaps. - Update manpage to reflect new commands and extend it with some examples. Sponsored by: Google Summer of Code 2007 Mentored by: le Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
.Pp
If everything went ok, the plex state should now be set to growable.
You can then start the growing with the
.Ic start
command:
.Pp
.Dl "gvinum start myraid5vol.p0"
.Pp
As with rebuilding, you can watch the progress using the
.Ic list
command.
.Pp
For a more advanced usage and detailed explanation of gvinum, the
handbook is recommended.
.Sh SEE ALSO
.Xr geom 4 ,
.Xr geom 8
.Sh HISTORY
The
.Nm
utility first appeared in
2006-09-30 11:02:17 +00:00
.Fx 5.3 .
The
.Nm vinum
utility, on which
.Nm
2006-09-30 11:02:17 +00:00
is based, was written by
.An "Greg Lehey" .
.Pp
2006-09-30 11:02:17 +00:00
The
.Nm
2006-09-30 11:02:17 +00:00
utility
was written by
.An "Lukas Ertl" .
The
.Ic move
and
.Ic rename
commands and
documentation were added by
.An "Chris Jones"
through the 2005 Google Summer
of Code program.
2013-03-14 18:55:41 +00:00
A partial rewrite of gvinum was done by
.An "Lukas Ertl"
and
.An "Ulf Lilleengen"
Import the gvinum work that have been done during and after Summer of Code 2007. The work have been under testing and fixing since then, and it is mature enough to be put into HEAD for further testing. A lot have changed in this time, and here are the most important: - Gvinum now uses one single workerthread instead of one thread for each volume and each plex. The reason for this is that the previous scheme was very complex, and was the cause of many of the bugs discovered in gvinum. Instead, gvinum now uses one worker thread with an event queue, quite similar to what used in gmirror. - The rebuild/grow/initialize/parity check routines no longer runs in separate threads, but are run as regular I/O requests with special flags. This made it easier to support mounted growing and parity rebuild. - Support for growing striped and raid5-plexes, meaning that one can extend the volumes for these plex types in addition to the concat type. Also works while the volume is mounted. - Implementation of many of the missing commands from the old vinum: attach/detach, start (was partially implemented), stop (was partially implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes with one plex of these organizations). - The parity check and rebuild no longer goes between userland/kernel, meaning that the gvinum command will not stay and wait forever for the rebuild to finish. You can instead watch the status with the list command. - Many problems with gvinum have been reported since 5.x, and some has been hard to fix due to the complicated architecture. Hopefully, it should be more stable and better handle edge cases that previously made gvinum crash. - Failed drives no longer disappears entirely, but now leave behind a dummy drive that makes sure the original state is not forgotten in case the system is rebooted between drive failures/swaps. - Update manpage to reflect new commands and extend it with some examples. Sponsored by: Google Summer of Code 2007 Mentored by: le Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
through the 2007 Google Summer of Code program.
The documentation have been updated to reflect the new functionality.
2006-09-30 11:02:17 +00:00
.Sh AUTHORS
.An Lukas Ertl Aq Mt le@FreeBSD.org
.An Chris Jones Aq Mt soc-cjones@FreeBSD.org
.An Ulf Lilleengen Aq Mt lulf@FreeBSD.org
.Sh BUGS
2006-09-30 11:02:17 +00:00
Currently,
.Nm
does not rename devices in
.Pa /dev/gvinum
until reloaded.
.Pp
The
.Fl S
initsize flag to
.Ic start
is ignored.
.Pp
Moving subdisks that are not part of a mirrored or RAID-5 volume will
2006-09-30 11:02:17 +00:00
destroy data.
It is perhaps a bug to permit this.
.Pp
Plexes in which subdisks have been moved do not automatically sync or
2006-09-30 11:02:17 +00:00
rebuild parity.
This may leave data unprotected and is perhaps unwise.
.Pp
2006-09-30 11:02:17 +00:00
Currently,
.Nm
does not yet fully implement all of the functions found in
.Nm vinum .
2006-09-30 11:02:17 +00:00
Specifically, the following commands from
.Nm vinum
2006-09-30 11:02:17 +00:00
are not supported:
.Bl -tag -width indent
.It Ic debug
Cause the volume manager to enter the kernel debugger.
.It Ic debug Ar flags
Set debugging flags.
.It Ic dumpconfig Op Ar drive ...
List the configuration information stored on the specified drives, or all
drives in the system if no drive names are specified.
2006-09-30 11:02:17 +00:00
.It Ic info Op Fl vV
List information about volume manager state.
.It Ic label Ar volume
Create a volume label.
2006-09-30 11:02:17 +00:00
.It Ic resetstats Oo Fl r Oc Op Ar volume | plex | subdisk
Reset statistics counters for the specified objects, or for all objects if none
are specified.
.It Ic setdaemon Op Ar value
Set daemon configuration.
.El