spacing

[lvm2.git] / man / lvm.conf.5.in

.TH LVM.CONF 5 "LVM TOOLS #VERSION#" "Sistina Software UK" \" -*- nroff -*-
.SH NAME
lvm.conf \- Configuration file for LVM2
.SH SYNOPSIS
.B #DEFAULT_SYS_DIR#/lvm.conf
.SH DESCRIPTION
lvm.conf is loaded during the initialisation phase of
\fBlvm\fP(8).  This file can in turn lead to other files
being loaded - settings read in later override earlier
settings.  File timestamps are checked between commands and if
any have changed, all the files are reloaded.
.LP
Use \fBlvm dumpconfig\fP to check what settings are in use.
.SH SYNTAX
.LP
This section describes the configuration file syntax.
.LP
Whitespace is not significant unless it is within quotes.
This provides a wide choice of acceptable indentation styles.
Comments begin with # and continue to the end of the line.
They are treated as whitespace.
.LP
Here is an informal grammar:
.TP
.BR file " = " value *
.br
A configuration file consists of a set of values.
.TP
.BR value " = " section " | " assignment
.br
A value can either be a new section, or an assignment.
.TP
.BR section " = " identifier " '" { "' " value "* '" } '
.br
A section is groups associated values together.
.br
It is denoted by a name and delimited by curly brackets.
.br
e.g.	backup {
.br
		...
.br
	}
.TP
.BR assignment " = " identifier " '" = "' ( " array " | " type " )"
.br
An assignment associates a type with an identifier.
.br
e.g.	max_archives = 42
.br
.TP
.BR array " =  '" [ "' ( " type " '" , "')* " type " '" ] "' | '" [ "' '" ] '
.br
Inhomogeneous arrays are supported.
.br
Elements must be separated by commas.
.br
An empty array is acceptable.
.TP
.BR type " = " integer " | " float " | " string
.BR integer " = [0-9]*"
.br
.BR float " = [0-9]*'" . '[0-9]*
.br
.B string \fR= '\fB"\fR'.*'\fB"\fR'
.IP
Strings must be enclosed in double quotes.

.SH SECTIONS
.LP
The sections that may be present in the file are:
.TP
\fBdevices\fP \(em Device settings
.IP
\fBdir\fP \(em Directory in which to create volume group device nodes.
Defaults to "/dev".  Commands also accept this as a prefix on volume
group names.
.IP
\fBscan\fP \(em List of directories to scan recursively for
LVM physical volumes.
Devices in directories outside this hierarchy will be ignored.
Defaults to "/dev".
.IP
\fBpreferred_names\fP \(em List of patterns compared in turn against
all the pathnames referencing the same device in in the scanned directories.
The pathname that matches the earliest pattern in the list is the
one used in any output.  As an example, if device-mapper multipathing
is used, the following will select multipath device names:
.br
\fBdevices { preferred_names = [ "^/dev/mapper/mpath" ] }\fP
.IP
\fBfilter\fP \(em List of patterns to apply to devices found by a scan.
Patterns are regular expressions delimited by any character and preceded
by \fBa\fP (for accept) or \fBr\fP (for reject).  The list is traversed
in order, and the first regex that matches determines if the device
will be accepted or rejected (ignored).  Devices that don't match
any patterns are accepted. If you want to reject patterns that
don't match, end the list with "r/.*/".
If there are several names for the same device (e.g. symbolic links
in /dev), if the first matching pattern in the list for any of the names is an
\fBa\fP pattern, the device is accepted; otherwise if the first matching
pattern in the list for any of the names is an \fBr\fP pattern it is rejected;
otherwise it is accepted.  As an example, to ignore /dev/cdrom you could use:
.br
\fBdevices { filter=["r|cdrom|"] }\fP
.IP
\fBcache_dir\fP \(em Persistent filter cache file directory.
Defaults to "#DEFAULT_CACHE_DIR#".
.IP
\fBwrite_cache_state\fP \(em Set to 0 to disable the writing out of the
persistent filter cache file when \fBlvm\fP exits.
Defaults to 1.
.IP
\fBtypes\fP \(em List of pairs of additional acceptable block device types
found in /proc/devices together with maximum (non-zero) number of
partitions (normally 16).  By default, LVM2 supports ide, sd, md, loop,
dasd, dac960, nbd, ida, cciss, ubd, ataraid, drbd, power2, i2o_block
and iseries/vd.  Block devices with major
numbers of different types are ignored by LVM2.
Example: \fBtypes = ["fd", 16]\fP.
To create physical volumes on device-mapper volumes
created outside LVM2, perhaps encrypted ones from \fBcryptsetup\fP,
you'll need \fBtypes = ["device-mapper", 16]\fP.  But if you do this,
be careful to avoid recursion within LVM2.  The figure for number
of partitions is not currently used in LVM2 - and might never be.
.IP
\fBsysfs_scan\fP \(em If set to 1 and your kernel supports sysfs and
it is mounted, sysfs will be used as a quick way of filtering out
block devices that are not present.
.IP
\fBmd_component_detection\fP \(em If set to 1, LVM2 will ignore devices
used as components of software RAID (md) devices by looking for md
superblocks. This doesn't always work satisfactorily e.g. if a device
has been reused without wiping the md superblocks first.
.IP
\fBmd_chunk_alignment\fP \(em If set to 1, and a Physical Volume is placed
directly upon an md device, LVM2 will align its data blocks with the
md device's stripe-width.
.IP
\fBdata_alignment_detection\fP \(em If set to 1, and your kernel provides
topology information in sysfs for the Physical Volume, the start of data
area will be aligned on a multiple of the ’minimum_io_size’ or
’optimal_io_size’ exposed in sysfs.  minimum_io_size is the smallest
request the device can perform without incurring a read-modify-write
penalty (e.g. MD's chunk size).  optimal_io_size is the device's
preferred unit of receiving I/O (e.g. MD's stripe width).  minimum_io_size
is used if optimal_io_size is undefined (0).  If both \fBmd_chunk_alignment\fP
and \fBdata_alignment_detection\fP are enabled the result of
\fBdata_alignment_detection\fP is used.
.IP
\fBdata_alignment\fP \(em Default alignment (in KB) of start of data area
when creating a new Physical Volume using the \fBlvm2\fP format.
If a Physical Volume is placed directly upon an md device and
\fBmd_chunk_alignment\fP or \fBdata_alignment_detection\fP is enabled
this parameter is ignored.  Set to 0 to use the default alignment of
64KB or the page size, if larger.
.IP
\fBdata_alignment_offset_detection\fP \(em If set to 1, and your kernel
provides topology information in sysfs for the Physical Volume, the
start of the aligned data area of the Physical Volume will be shifted
by the alignment_offset exposed in sysfs.
.sp
To see the location of the first Physical Extent of an existing Physical Volume
use \fBpvs -o +pe_start\fP .  It will be a multiple of the requested
\fBdata_alignment\fP plus the alignment_offset from
\fBdata_alignment_offset_detection\fP (if enabled) or the pvcreate
commandline.
.IP
\fBdisable_after_error_count\fP \(em During each LVM operation errors received
from each device are counted.  If the counter of a particular device exceeds
the limit set here, no further I/O is sent to that device for the remainder of
the respective operation. Setting the parameter to 0 disables the counters
altogether.
.IP
\fBpv_min_size\fP \(em
Minimal size (in KB) of the block device which can be used as a PV.
In clustered environment all nodes have to use the same value.
Any value smaller than 512KB is ignored.  Up to and include version 2.02.84
the default was 512KB.  From 2.02.85 onwards it was changed to 2MB to
avoid floppy drives by default.
.IP
\fBissue_discards\fP \(em
Issue discards to a logical volumes's underlying physical volume(s) when the
logical volume is no longer using the physical volumes' space (e.g. lvremove,
lvreduce, etc).  Discards inform the storage that a region is no longer in use.
Storage that supports discards advertise the protocol specific way discards
should be issued by the kernel (TRIM, UNMAP, or WRITE SAME with UNMAP bit set).
Not all storage will support or benefit from discards but SSDs and thinly
provisioned LUNs generally do.  If set to 1, discards will only be issued if
both the storage and kernel provide support.
.IP
.TP
\fBallocation\fP \(em Space allocation policies
.IP
\fBcling_tag_list\fP \(em List of PV tags matched by the \fBcling\fP allocation policy.
.IP
When searching for free space to extend an LV, the \fBcling\fP
allocation policy will choose space on the same PVs as the last
segment of the existing LV.  If there is insufficient space and a
list of tags is defined here, it will check whether any of them are
attached to the PVs concerned and then seek to match those PV tags
between existing extents and new extents.
.IP
The @ prefix for tags is required.
Use the special tag "@*" as a wildcard to match any PV tag and so use
all PV tags for this purpose.
.IP
For example, LVs are mirrored between two sites within a single VG.
PVs are tagged with either @site1 or @site2 to indicate where
they are situated and these two PV tags are selected for use with this
allocation policy:
.IP
cling_tag_list = [ "@site1", "@site2" ]
.TP
\fBlog\fP \(em Default log settings
.IP
\fBfile\fP \(em Location of log file.  If this entry is not present, no
log file is written.
.IP
\fBoverwrite\fP \(em Set to 1 to overwrite the log file each time a tool
is invoked.  By default tools append messages to the log file.
.IP
\fBlevel\fP \(em Log level (0-9) of messages to write to the file.
9 is the most verbose; 0 should produce no output.
.IP
\fBverbose\fP \(em Default level (0-3) of messages sent to stdout or stderr.
3 is the most verbose; 0 should produce the least output.
.IP
\fBsyslog\fP \(em Set to 1 (the default) to send log messages through syslog.
Turn off by setting to 0.  If you set to an integer greater than one,
this is used - unvalidated - as the facility.  The default is LOG_USER.
See /usr/include/sys/syslog.h for safe facility values to use.
For example, LOG_LOCAL0 might be 128.
.IP
\fBindent\fP \(em When set to 1 (the default) messages are indented
according to their severity, two spaces per level.
Set to 0 to turn off indentation.
.IP
\fBcommand_names\fP \(em When set to 1, the command name is used as a
prefix for each message.
Default is 0 (off).
.IP
\fBprefix\fP \(em Prefix used for all messages (after the command name).
Default is two spaces.
.IP
\fBactivation\fP \(em Set to 1 to log messages while
devices are suspended during activation.
Only set this temporarily while debugging a problem because
in low memory situations this setting can cause your machine to lock up.
.TP
\fBbackup\fP \(em Configuration for metadata backups.
.IP
\fBarchive_dir\fP \(em Directory used for automatic metadata archives.
Backup copies of former metadata for each volume group are archived here.
Defaults to "#DEFAULT_ARCHIVE_DIR#".
.IP
\fBbackup_dir\fP \(em Directory used for automatic metadata backups.
A single backup copy of the current metadata for each volume group
is stored here.
Defaults to "#DEFAULT_BACKUP_DIR#".
.IP
\fBarchive\fP \(em Whether or not tools automatically archive existing
metadata into \fBarchive_dir\fP before making changes to it.
Default is 1 (automatic archives enabled).
Set to 0 to disable.
Disabling this might make metadata recovery difficult or impossible
if something goes wrong.
.IP
\fBbackup\fP \(em Whether or not tools make an automatic backup
into \fBbackup_dir\fP after changing metadata.
Default is 1 (automatic backups enabled).  Set to 0 to disable.
Disabling this might make metadata recovery difficult or impossible
if something goes wrong.
.IP
\fBretain_min\fP \(em Minimum number of archives to keep.
Defaults to 10.
.IP
\fBretain_days\fP \(em Minimum number of days to keep archive files.
Defaults to 30.
.TP
\fBshell\fP \(em LVM2 built-in readline shell settings
.IP
\fBhistory_size\fP \(em Maximum number of lines of shell history to retain (default 100) in $HOME/.lvm_history
.TP
\fBglobal\fP \(em Global settings
.IP
\fBtest\fP \(em If set to 1, run tools in test mode i.e. no changes to
the on-disk metadata will get made.  It's equivalent to having the
-t option on every command.
.IP
\fBactivation\fP \(em Set to 0 to turn off all communication with
the device-mapper driver.  Useful if you want to manipulate logical
volumes while device-mapper is not present in your kernel.
.IP
\fBproc\fP \(em Mount point of proc filesystem.
Defaults to /proc.
.IP
\fBumask\fP \(em File creation mask for any files and directories created.
Interpreted as octal if the first digit is zero.
Defaults to 077.
Use 022 to allow other users to read the files by default.
.IP
\fBformat\fP \(em The default value of \fB--metadatatype\fP used
to determine which format of metadata to use when creating new
physical volumes and volume groups. \fBlvm1\fP or \fBlvm2\fP.
.IP
\fBfallback_to_lvm1\fP \(em Set this to 1 if you need to
be able to switch between 2.4 kernels using LVM1 and kernels
including device-mapper.
The LVM2 tools should be installed as normal and
the LVM1 tools should be installed with a .lvm1 suffix e.g.
vgscan.lvm1.
If an LVM2 tool is then run but unable to communicate
with device-mapper, it will automatically invoke the equivalent LVM1
version of the tool.  Note that for LVM1 tools to
manipulate physical volumes and volume groups created by LVM2 you
must use \fB--metadataformat lvm1\fP when creating them.
.IP
\fBlibrary_dir\fP \(em A directory searched for LVM2's shared libraries
ahead of the places \fBdlopen\fP (3) searches.
.IP
\fBformat_libraries\fP \(em A list of shared libraries to load that contain
code to process different formats of metadata. For example, liblvm2formatpool.so
is needed to read GFS pool metadata if LVM2 was configured \fB--with-pool=shared\fP.
.IP
\fBlocking_type\fP \(em What type of locking to use.
1 is the default, which use flocks on files in \fBlocking_dir\fP
(see below) to
avoid conflicting LVM2 commands running concurrently on a single
machine. 0 disables locking and risks corrupting your metadata.
If set to 2, the tools will load the external \fBlocking_library\fP
(see below).
If the tools were configured \fB--with-cluster=internal\fP
(the default) then 3 means to use built-in cluster-wide locking.
Type 4 enforces read-only metadata and forbids any operations that
might want to modify Volume Group metadata.
All changes to logical volumes and their states are communicated
using locks.
.IP
\fBwait_for_locks\fP \(em When set to 1, the default, the tools
wait if a lock request cannot be satisfied immediately.
When set to 0, the operation is aborted instead.
.IP
\fBlocking_dir\fP \(em The directory LVM2 places its file locks
if \fBlocking_type\fP is set to 1.  The default is \fB/var/lock/lvm\fP.
.IP
\fBlocking_library\fP \(em The name of the external locking
library to load if \fBlocking_type\fP is set to 2.
The default is \fBliblvm2clusterlock.so\fP.  If you need to write
such a library, look at the lib/locking source code directory.
.TP
\fBtags\fP \(em Host tag settings
.IP
\fBhosttags\fP \(em If set to 1, create a host tag with the machine name.
Setting this to 0 does nothing, neither creating nor destroying any tag.
The machine name used is the nodename as returned by \fBuname\fP (2).
.IP
Additional host tags to be set can be listed here as subsections.
The @ prefix for tags is optional.
Each of these host tag subsections can contain a \fBhost_list\fP
array of host names. If any one of these entries matches the machine
name exactly then the host tag gets defined on this particular host,
otherwise it doesn't.
.IP
After lvm.conf has been processed, LVM2 works through each host
tag that has been defined in turn, and if there is a configuration
file called lvm_\fB<host_tag>\fP.conf it attempts to load it.
Any settings read in override settings found in earlier files.
Any additional host tags defined get appended to the search list,
so in turn they can lead to further configuration files being processed.
Use \fBlvm dumpconfig\fP to check the result of config
file processing.
.IP
The following example always sets host tags \fBtag1\fP and
sets \fBtag2\fP on machines fs1 and fs2:
.IP
tags { tag1 { } tag2 { host_list = [ "fs1", "fs2" ] } }
.IP
These options are useful if you are replicating configuration files
around a cluster.  Use of \fBhosttags = 1\fP means every machine
can have static and identical local configuration files yet use
different settings and activate different logical volumes by
default.  See also \fBvolume_list\fP below and \fB--addtag\fP
in \fBlvm\fP (8).
.TP
\fBactivation\fP \(em Settings affecting device-mapper activation
.IP
\fBmissing_stripe_filler\fP \(em When activating an incomplete logical
volume in partial mode, this option dictates how the missing data is
replaced.  A value of "error" will cause activation to create error
mappings for the missing data, meaning that read access to missing
portions of the volume will result in I/O errors. You can instead also
use a device path, and in that case this device will be used in place of
missing stripes. However, note that using anything other than
"error" with mirrored or snapshotted volumes is likely to result in data
corruption.  For instructions on how to create a device that always
returns zeros, see \fBlvcreate\fP (8).
.IP
\fBmirror_region_size\fP \(em Unit size in KB for copy operations
when mirroring.
.IP
\fBreadahead\fP \(em Used when there is no readahead value stored
in the volume group metadata.  Set to \fBnone\fP to disable
readahead in these circumstances or \fBauto\fP to use the default
value chosen by the kernel.
.IP
\fBreserved_memory\fP, \fBreserved_stack\fP \(em How many KB to reserve
for LVM2 to use while logical volumes are suspended.  If insufficient
memory is reserved before suspension, there is a risk of machine deadlock.
.IP
\fBprocess_priority\fP \(em The nice value to use while devices are
suspended.  This is set to a high priority so that logical volumes
are suspended (with I/O generated by other processes to those
logical volumes getting queued) for the shortest possible time.
.IP
\fBvolume_list\fP \(em This acts as a filter through which
all requests to activate a logical volume on this machine
are passed.  A logical volume is only activated if it matches
an item in the list.  Tags must be preceded by @ and are checked
against all tags defined in the logical volume and volume group
metadata for a match.
@* is short-hand to check every tag set on the host machine (see
\fBtags\fP above).
Logical volume and volume groups can also be included in the list
by name e.g. vg00, vg00/lvol1.
.IP
\fBread_only_volume_list\fP \(em This acts as a filter through
which all requests to activate a logical volume on this machine
are passed.  A logical volume is activated in read-only mode (instead
of read-write) if it matches an item in the list.  Volumes must first
pass the \fBvolume_list\fP filter, if present.  Tags must be preceded
by @ and are checked against all tags defined in the logical volume
and volume group metadata for a match.
@* is short-hand to check every tag set on the host machine (see
\fBtags\fP above).
Logical volume and volume groups can also be included in the list
by name e.g. vg00, vg00/lvol1.
.TP
\fBmetadata\fP \(em Advanced metadata settings
.IP
\fBpvmetadatacopies\fP \(em When creating a physical volume using the
LVM2 metadata format, this is the default number of copies of metadata
to store on each physical volume.
Currently it can be set to 0, 1 or 2.  The default is 1.
If set to 2, one copy is placed at the beginning of the disk
and the other is placed at the end.
It can be overridden on the command line with \fB--pvmetadatacopies\fP
(see \fBpvcreate\fP).
If creating a volume group with just one physical volume, it's a
good idea to have 2 copies.  If creating a large volume group with
many physical volumes, you may decide that 3 copies of the metadata
is sufficient, i.e. setting it to 1 on three of the physical volumes,
and 0 on the rest.  Every volume group must contain at least one
physical volume with at least 1 copy of the metadata (unless using
the text files described below).  The disadvantage of having lots
of copies is that every time the tools access the volume group, every
copy of the metadata has to be accessed, and this slows down the
tools.
.IP
\fBpvmetadatasize\fP \(em Approximate number of sectors to set aside
for each copy of the metadata. Volume groups with large numbers of
physical or logical volumes, or volumes groups containing complex
logical volume structures will need additional space for their metadata.
The metadata areas are treated as circular buffers, so
unused space becomes filled with an archive of the most recent
previous versions of the metadata.
.IP
\fBpvmetadataignore\fP When creating a physical volume using the LVM2
metadata format, this states whether metadata areas should be ignored.
The default is "n".  If metadata areas on a physical volume are ignored,
LVM will not not store metadata in the metadata areas present on newly
created Physical Volumes.  The option can be overridden on the command
line with \fB--metadataignore\fP (See \fBpvcreate\fP and \fBpvchange\fP).
Metadata areas cannot be created or extended after Logical Volumes have
been allocated on the device.
If you do not want to store metadata on this device, it is still wise
always to allocate a metadata area (use a non-zero value for
\fB--pvmetadatacopies\fP) in case you need it in the future and to use
this option to instruct LVM2 to ignore it.
.IP
\fBvgmetadatacopies\fP \(em When creating a volume group using the
LVM2 metadata format, this is the default number of copies of metadata
desired across all the physical volumes in the volume group.  If set to
a non-zero value, LVM will automatically set or clear the metadataignore
flag on the physical volumes (see \fBpvcreate\fP and \fBpvchange\fP
\fB--metadataignore\fP) in order to achieve the desired number of metadata
copies.  An LVM command that adds or removes physical volumes (for example,
\fBvgextend\fP, \fBvgreduce\fP, \fBvgsplit\fP, or \fBvgmerge\fP), may cause
LVM to automatically set or clear the metadataignore flags.  Also, if
physical volumes go missing or reappear, or a new number of copies is
explicitly set (see \fBvgchange --vgmetadatacopies\fP), LVM may adjust
the metadataignore flags.
Set \fBvgmetadatacopies\fP to 0 instructs LVM not to set or clear the
metadataignore flags automatically.  You may set a value larger than the
sum of all metadata areas on all physical volumes.  The value can
be overridden on the command line with \fB--vgmetadatacopies\fP for various
commands (for example, \fBvgcreate\fP and \fBvgchange\fP), and can be
queryied with the \fBvg_mda_copies\fP field of \fBvgs\fP.  This option
is useful for volume groups containing large numbers of physical volumes
with metadata as it may be used to minimize metadata read and write overhead.
.IP
\fBdirs\fP \(em List of directories holding live copies of LVM2
metadata as text files.  These directories must not be on logical
volumes.  It is possible to use LVM2 with a couple of directories
here, preferably on different (non-logical-volume) filesystems
and with no other on-disk metadata, \fBpvmetadatacopies = 0\fP.
Alternatively these directories can be in addition to the
on-disk metadata areas.  This feature was created during the
development of the LVM2 metadata before the new on-disk metadata
areas were designed and no longer gets tested.
It is not supported under low-memory conditions, and it is
important never to edit these metadata files unless you fully
understand how things work: to make changes you should always use
the tools as normal, or else vgcfgbackup, edit backup, vgcfgrestore.
.SH FILES
.I #DEFAULT_SYS_DIR#/lvm.conf
.br
.I #DEFAULT_ARCHIVE_DIR#
.br
.I #DEFAULT_BACKUP_DIR#
.br
.I #DEFAULT_CACHE_DIR#/.cache
.br
.I #DEFAULT_LOCK_DIR#
.SH SEE ALSO
.BR lvm (8),
.BR umask (2),
.BR uname (2),
.BR dlopen (3),
.BR syslog (3),
.BR syslog.conf (5)