Manpage/btrfs-device

From btrfs Wiki
(Difference between revisions)
Jump to: navigation, search
(Update from git, v5.14.2)
(remove content)
 
Line 1: Line 1:
=btrfs-device(8) manual page=
 
 
{{GeneratedManpage
 
{{GeneratedManpage
 
|name=btrfs-device}}
 
|name=btrfs-device}}
 
==NAME==
 
btrfs-device - manage devices of btrfs filesystems
 
 
==SYNOPSIS==
 
 
<p><b>btrfs device</b> <em>&lt;subcommand&gt;</em> <em>&lt;args&gt;</em></p>
 
==DESCRIPTION==
 
 
<p>The <b>btrfs device</b> command group is used to manage devices of the btrfs filesystems.</p>
 
==DEVICE MANAGEMENT==
 
 
<p>Btrfs filesystem can be created on top of single or multiple block devices.
 
Data and metadata are organized in allocation profiles with various redundancy
 
policies. There&#8217;s some similarity with traditional RAID levels, but this could
 
be confusing to users familiar with the traditional meaning. Due to the
 
similarity, the RAID terminology is widely used in the documentation.  See
 
[[Manpage/mkfs.btrfs|mkfs.btrfs(8)]] for more details and the exact profile capabilities and
 
constraints.</p>
 
<p>The device management works on a mounted filesystem. Devices can be added,
 
removed or replaced, by commands provided by <b>btrfs device</b> and <b>btrfs replace</b>.</p>
 
<p>The profiles can be also changed, provided there&#8217;s enough workspace to do the
 
conversion, using the <b>btrfs balance</b> command and namely the filter <em>convert</em>.</p>
 
<dl>
 
<dt>
 
Type
 
<dd>
 
<p>
 
The block group profile type is the main distinction of the information stored
 
on the block device. User data are called <em>Data</em>, the internal data structures
 
managed by filesystem are <em>Metadata</em> and <em>System</em>.
 
</p>
 
 
<dt>
 
Profile
 
<dd>
 
<p>
 
A profile describes an allocation policy based on the redundancy/replication
 
constraints in connection with the number of devices. The profile applies to
 
data and metadata block groups separately. Eg. <em>single</em>, <em>RAID1</em>.
 
</p>
 
 
<dt>
 
RAID level
 
<dd>
 
<p>
 
Where applicable, the level refers to a profile that matches constraints of the
 
standard RAID levels. At the moment the supported ones are: RAID0, RAID1,
 
RAID10, RAID5 and RAID6.
 
</p>
 
 
</dl>
 
<p>See the section <b>TYPICAL USECASES</b> for some examples.</p>
 
==SUBCOMMAND==
 
 
<dl>
 
<dt>
 
<b>add</b> [-Kf] <em>&lt;device&gt;</em> [<em>&lt;device&gt;</em>&#8230;] <em>&lt;path&gt;</em>
 
<dd>
 
<p>
 
Add device(s) to the filesystem identified by <em>&lt;path&gt;</em>.
 
</p>
 
<p>If applicable, a whole device discard (TRIM) operation is performed prior to
 
adding the device. A device with existing filesystem detected by [http://man7.org/linux/man-pages/man8/blkid.8.html blkid(8)]
 
will prevent device addition and has to be forced. Alternatively the filesystem
 
can be wiped from the device using eg. the [http://man7.org/linux/man-pages/man8/wipefs.8.html wipefs(8)] tool.</p>
 
<p>The operation is instant and does not affect existing data. The operation merely
 
adds the device to the filesystem structures and creates some block groups
 
headers.</p>
 
<p><tt>Options</tt></p>
 
<dl>
 
<dt>
 
-K|--nodiscard
 
<dd>
 
<p>
 
do not perform discard (TRIM) by default
 
</p>
 
 
<dt>
 
-f|--force
 
<dd>
 
<p>
 
force overwrite of existing filesystem on the given disk(s)
 
</p>
 
 
<dt>
 
--enqueue
 
<dd>
 
<p>
 
wait if there&#8217;s another exclusive operation running, otherwise continue
 
</p>
 
 
</dl>
 
 
<dt>
 
<b>remove</b> [options] <em>&lt;device&gt;</em>|<em>&lt;devid&gt;</em> [<em>&lt;device&gt;</em>|<em>&lt;devid&gt;</em>&#8230;] <em>&lt;path&gt;</em>
 
<dd>
 
<p>
 
Remove device(s) from a filesystem identified by <em>&lt;path&gt;</em>
 
</p>
 
<p>Device removal must satisfy the profile constraints, otherwise the command
 
fails. The filesystem must be converted to profile(s) that would allow the
 
removal. This can typically happen when going down from 2 devices to 1 and
 
using the RAID1 profile. See the <b>TYPICAL USECASES</b> section below.</p>
 
<p>The operation can take long as it needs to move all data from the device.</p>
 
<p>It is possible to delete the device that was used to mount the filesystem. The
 
device entry in the mount table will be replaced by another device name with
 
the lowest device id.</p>
 
<p>If the filesystem is mounted in degraded mode (-o degraded), special term
 
<em>missing</em> can be used for <em>device</em>. In that case, the first device that is
 
described by the filesystem metadata, but not present at the mount time will be
 
removed.</p>
 
<blockquote><b>Note:</b>
 
In most cases, there is only one missing device in degraded mode,
 
otherwise mount fails. If there are two or more devices missing (e.g. possible
 
in RAID6), you need specify <em>missing</em> as many times as the number of missing
 
devices to remove all of them.</blockquote>
 
<p><tt>Options</tt></p>
 
<dl>
 
<dt>
 
--enqueue
 
<dd>
 
<p>
 
wait if there&#8217;s another exclusive operation running, otherwise continue
 
</p>
 
 
</dl>
 
 
<dt>
 
<b>delete</b> <em>&lt;device&gt;</em>|<em>&lt;devid&gt;</em> [<em>&lt;device&gt;</em>|<em>&lt;devid&gt;</em>&#8230;] <em>&lt;path&gt;</em>
 
<dd>
 
<p>
 
Alias of remove kept for backward compatibility
 
</p>
 
 
<dt>
 
<b>ready</b> <em>&lt;device&gt;</em>
 
<dd>
 
<p>
 
Wait until all devices of a multiple-device filesystem are scanned and
 
registered within the kernel module. This is to provide a way for automatic
 
filesystem mounting tools to wait before the mount can start. The device scan
 
is only one of the preconditions and the mount can fail for other reasons.
 
Normal users usually do not need this command and may safely ignore it.
 
</p>
 
 
<dt>
 
<b>scan</b> [options] [<em>&lt;device&gt;</em> [<em>&lt;device&gt;</em>&#8230;]]
 
<dd>
 
<p>
 
Scan devices for a btrfs filesystem and register them with the kernel module.
 
This allows mounting multiple-device filesystem by specifying just one from the
 
whole group.
 
</p>
 
<p>If no devices are passed, all block devices that blkid reports to contain btrfs
 
are scanned.</p>
 
<p>The options <em>--all-devices</em> or <em>-d</em> can be used as a fallback in case blkid is
 
not available.  If used, behavior is the same as if no devices are passed.</p>
 
<p>The command can be run repeatedly. Devices that have been already registered
 
remain as such. Reloading the kernel module will drop this information. There&#8217;s
 
an alternative way of mounting multiple-device filesystem without the need for
 
prior scanning. See the mount option <em>device</em>.</p>
 
<p><tt>Options</tt></p>
 
<dl>
 
<dt>
 
-d|--all-devices
 
<dd>
 
<p>
 
Enumerate and register all devices, use as a fallback in case blkid is not
 
available.
 
</p>
 
 
<dt>
 
-u|--forget
 
<dd>
 
<p>
 
Unregister a given device or all stale devices if no path is given, the device
 
must be unmounted otherwise it&#8217;s an error.
 
</p>
 
 
</dl>
 
 
<dt>
 
<b>stats</b> [options] <em>&lt;path&gt;</em>|<em>&lt;device&gt;</em>
 
<dd>
 
<p>
 
Read and print the device IO error statistics for all devices of the given
 
filesystem identified by <em>&lt;path&gt;</em> or for a single <em>&lt;device&gt;</em>. The filesystem must
 
be mounted.  See section <b>DEVICE STATS</b> for more information about the reported
 
statistics and the meaning.
 
</p>
 
<p><tt>Options</tt></p>
 
<dl>
 
<dt>
 
-z|--reset
 
<dd>
 
<p>
 
Print the stats and reset the values to zero afterwards.
 
</p>
 
 
<dt>
 
-c|--check
 
<dd>
 
<p>
 
Check if the stats are all zeros and return 0 if it is so. Set bit 6 of the
 
return code if any of the statistics is no-zero. The error values is 65 if
 
reading stats from at least one device failed, otherwise it&#8217;s 64.
 
</p>
 
 
</dl>
 
 
<dt>
 
<b>usage</b> [options] <em>&lt;path&gt;</em> [<em>&lt;path&gt;</em>&#8230;]
 
<dd>
 
<p>
 
Show detailed information about internal allocations on devices.
 
</p>
 
<p>The level of detail can differ if the command is run under a regular or the
 
root user (due to use of restricted ioctls). The first example below is for
 
normal user (warning included) and the next one with root on the same
 
filesystem:</p>
 
<pre>WARNING: cannot read detailed chunk info, per-device usage will not be shown, run as root
 
/dev/sdc1, ID: 1
 
  Device size:          931.51GiB
 
  Device slack:              0.00B
 
  Unallocated:          931.51GiB</pre>
 
<pre>/dev/sdc1, ID: 1
 
  Device size:          931.51GiB
 
  Device slack:              0.00B
 
  Data,single:          641.00GiB
 
  Data,RAID0/3:            1.00GiB
 
  Metadata,single:        19.00GiB
 
  System,single:          32.00MiB
 
  Unallocated:          271.48GiB</pre>
 
<div>
 
<ul>
 
<li>
 
<p>
 
<em>Device size</em>&#8201;&#8212;&#8201;size of the device as seen by the filesystem (may be
 
different than actual device size)
 
</p>
 
</li>
 
<li>
 
<p>
 
<em>Device slack</em>&#8201;&#8212;&#8201;portion of device not used by the filesystem but still
 
available in the physical space provided by the device, eg. after a device shrink
 
</p>
 
</li>
 
<li>
 
<p>
 
<em>Data,single</em>, <em>Metadata,single</em>, <em>System,single</em>&#8201;&#8212;&#8201;in general, list of block
 
group type (Data, Metadata, System) and profile (single, RAID1, &#8230;) allocated
 
on the device
 
</p>
 
</li>
 
<li>
 
<p>
 
<em>Data,RAID0/3</em>&#8201;&#8212;&#8201;in particular, striped profiles RAID0/RAID10/RAID5/RAID6 with
 
the number of devices on which the stripes are allocated, multiple occurences
 
of the same profile can appear in case a new device has been added and all new
 
available stripes have been used for writes
 
</p>
 
</li>
 
<li>
 
<p>
 
<em>Unallocated</em>&#8201;&#8212;&#8201;remaining space that the filesystem can still use for new
 
block groups
 
</p>
 
</li>
 
</ul>
 
</div>
 
<p><tt>Options</tt></p>
 
<dl>
 
<dt>
 
-b|--raw
 
<dd>
 
<p>
 
raw numbers in bytes, without the <em>B</em> suffix
 
</p>
 
 
<dt>
 
-h|--human-readable
 
<dd>
 
<p>
 
print human friendly numbers, base 1024, this is the default
 
</p>
 
 
<dt>
 
-H
 
<dd>
 
<p>
 
print human friendly numbers, base 1000
 
</p>
 
 
<dt>
 
--iec
 
<dd>
 
<p>
 
select the 1024 base for the following options, according to the IEC standard
 
</p>
 
 
<dt>
 
--si
 
<dd>
 
<p>
 
select the 1000 base for the following options, according to the SI standard
 
</p>
 
 
<dt>
 
-k|--kbytes
 
<dd>
 
<p>
 
show sizes in KiB, or kB with --si
 
</p>
 
 
<dt>
 
-m|--mbytes
 
<dd>
 
<p>
 
show sizes in MiB, or MB with --si
 
</p>
 
 
<dt>
 
-g|--gbytes
 
<dd>
 
<p>
 
show sizes in GiB, or GB with --si
 
</p>
 
 
<dt>
 
-t|--tbytes
 
<dd>
 
<p>
 
show sizes in TiB, or TB with --si
 
</p>
 
 
</dl>
 
 
</dl>
 
<p>If conflicting options are passed, the last one takes precedence.</p>
 
==TYPICAL USECASES==
 
 
===STARTING WITH A SINGLE-DEVICE FILESYSTEM===
 
 
<p>Assume we&#8217;ve created a filesystem on a block device <em>/dev/sda</em> with profile
 
<em>single/single</em> (data/metadata), the device size is 50GiB and we&#8217;ve used the
 
whole device for the filesystem. The mount point is <em>/mnt</em>.</p>
 
<p>The amount of data stored is 16GiB, metadata have allocated 2GiB.</p>
 
====ADD NEW DEVICE====
 
 
<p>We want to increase the total size of the filesystem and keep the profiles. The
 
size of the new device <em>/dev/sdb</em> is 100GiB.</p>
 
<pre>$ btrfs device add /dev/sdb /mnt</pre>
 
<p>The amount of free data space increases by less than 100GiB, some space is
 
allocated for metadata.</p>
 
====CONVERT TO RAID1====
 
 
<p>Now we want to increase the redundancy level of both data and metadata, but
 
we&#8217;ll do that in steps. Note, that the device sizes are not equal and we&#8217;ll use
 
that to show the capabilities of split data/metadata and independent profiles.</p>
 
<p>The constraint for RAID1 gives us at most 50GiB of usable space and exactly 2
 
copies will be stored on the devices.</p>
 
<p>First we&#8217;ll convert the metadata. As the metadata occupy less than 50GiB and
 
there&#8217;s enough workspace for the conversion process, we can do:</p>
 
<pre>$ btrfs balance start -mconvert=raid1 /mnt</pre>
 
<p>This operation can take a while, because all metadata have to be moved and all
 
block pointers updated. Depending on the physical locations of the old and new
 
blocks, the disk seeking is the key factor affecting performance.</p>
 
<p>You&#8217;ll note that the system block group has been also converted to RAID1, this
 
normally happens as the system block group also holds metadata (the physical to
 
logical mappings).</p>
 
<p>What changed:</p>
 
<ul>
 
<li>
 
<p>
 
available data space decreased by 3GiB, usable roughly (50 - 3) + (100 - 3) = 144 GiB
 
</p>
 
</li>
 
<li>
 
<p>
 
metadata redundancy increased
 
</p>
 
</li>
 
</ul>
 
<p>IOW, the unequal device sizes allow for combined space for data yet improved
 
redundancy for metadata. If we decide to increase redundancy of data as well,
 
we&#8217;re going to lose 50GiB of the second device for obvious reasons.</p>
 
<pre>$ btrfs balance start -dconvert=raid1 /mnt</pre>
 
<p>The balance process needs some workspace (ie. a free device space without any
 
data or metadata block groups) so the command could fail if there&#8217;s too much
 
data or the block groups occupy the whole first device.</p>
 
<p>The device size of <em>/dev/sdb</em> as seen by the filesystem remains unchanged, but
 
the logical space from 50-100GiB will be unused.</p>
 
====REMOVE DEVICE====
 
 
<p>Device removal must satisfy the profile constraints, otherwise the command
 
fails. For example:</p>
 
<pre>$ btrfs device remove /dev/sda /mnt
 
ERROR: error removing device '/dev/sda': unable to go below two devices on raid1</pre>
 
<p>In order to remove a device, you need to convert the profile in this case:</p>
 
<pre>$ btrfs balance start -mconvert=dup -dconvert=single /mnt
 
$ btrfs device remove /dev/sda /mnt</pre>
 
==DEVICE STATS==
 
 
<p>The device stats keep persistent record of several error classes related to
 
doing IO. The current values are printed at mount time and updated during
 
filesystem lifetime or from a scrub run.</p>
 
<pre>$ btrfs device stats /dev/sda3
 
[/dev/sda3].write_io_errs  0
 
[/dev/sda3].read_io_errs    0
 
[/dev/sda3].flush_io_errs  0
 
[/dev/sda3].corruption_errs 0
 
[/dev/sda3].generation_errs 0</pre>
 
<dl>
 
<dt>
 
write_io_errs
 
<dd>
 
<p>
 
Failed writes to the block devices, means that the layers beneath the
 
filesystem were not able to satisfy the write request.
 
</p>
 
 
<dt>
 
read_io_errors
 
<dd>
 
<p>
 
Read request analogy to write_io_errs.
 
</p>
 
 
<dt>
 
flush_io_errs
 
<dd>
 
<p>
 
Number of failed writes with the <em>FLUSH</em> flag set. The flushing is a method of
 
forcing a particular order between write requests and is crucial for
 
implementing crash consistency. In case of btrfs, all the metadata blocks must
 
be permanently stored on the block device before the superblock is written.
 
</p>
 
 
<dt>
 
corruption_errs
 
<dd>
 
<p>
 
A block checksum mismatched or a corrupted metadata header was found.
 
</p>
 
 
<dt>
 
generation_errs
 
<dd>
 
<p>
 
The block generation does not match the expected value (eg. stored in the
 
parent node).
 
</p>
 
 
</dl>
 
<p>Since kernel 5.14 the device stats are also available in textual form in
 
<em>/sys/fs/btrfs/FSID/devinfo/DEVID/error_stats</em>.</p>
 
==EXIT STATUS==
 
 
<p><b>btrfs device</b> returns a zero exit status if it succeeds. Non zero is
 
returned in case of failure.</p>
 
<p>If the <em>-s</em> option is used, <b>btrfs device stats</b> will add 64 to the
 
exit status if any of the error counters is non-zero.</p>
 
==AVAILABILITY==
 
 
<p><b>btrfs</b> is part of btrfs-progs.
 
Please refer to the btrfs wiki http://btrfs.wiki.kernel.org for
 
further details.</p>
 
==SEE ALSO==
 
 
<p>[[Manpage/mkfs.btrfs|mkfs.btrfs(8)]],
 
[[Manpage/btrfs-replace|btrfs-replace(8)]],
 
[[Manpage/btrfs-balance|btrfs-balance(8)]]</p>
 
[[Category:Manpage]]
 

Latest revision as of 12:29, 12 January 2022

Note: manual pages are located at read-the-docs site, please update your links.


Personal tools