6 mkfs.btrfs - create a btrfs filesystem
10 *mkfs.btrfs* [options] <device> [<device>...]
14 *mkfs.btrfs* is used to create the btrfs filesystem on a single or multiple
15 devices. <device> is typically a block device but can be a file-backed image
16 as well. Multiple devices are grouped by UUID of the filesystem.
18 Before mounting such filesystem, the kernel module must know all the devices
19 either via preceding execution of *btrfs device scan* or using the *device*
20 mount option. See section *MULTIPLE DEVICES* for more details.
24 *-b|--byte-count <size>*::
25 Specify the size of the filesystem. If this option is not used, then
26 mkfs.btrfs uses the entire device space for the filesystem.
28 *-d|--data <profile>*::
29 Specify the profile for the data block groups. Valid values are 'raid0',
30 'raid1', 'raid5', 'raid6', 'raid10' or 'single' or dup (case does not matter).
32 See 'DUP PROFILES ON A SINGLE DEVICE' for more.
34 *-m|--metadata <profile>*::
35 Specify the profile for the metadata block groups.
36 Valid values are 'raid0', 'raid1', 'raid5', 'raid6', 'raid10', 'single' or
37 'dup', (case does not matter).
39 A single device filesystem will default to 'DUP', unless a SSD is detected. Then
40 it will default to 'single'. The detection is based on the value of
41 `/sys/block/DEV/queue/rotational`, where 'DEV' is the short name of the device.
43 Note that the rotational status can be arbitrarily set by the underlying block
44 device driver and may not reflect the true status (network block device, memory-backed
45 SCSI devices etc). Use the options '--data/--metadata' to avoid confusion.
47 See 'DUP PROFILES ON A SINGLE DEVICE' for more details.
50 Normally the data and metadata block groups are isolated. The 'mixed' mode
51 will remove the isolation and store both types in the same block group type.
52 This helps to utilize the free space regardless of the purpose and is suitable
53 for small devices. The separate allocation of block groups leads to a situation
54 where the space is reserved for the other block group type, is not available for
55 allocation and can lead to ENOSPC state.
57 The recommended size for the mixed mode is for filesystems less than 1GiB. The
58 soft recommendation is to use it for filesystems smaller than 5GiB. The mixed
59 mode may lead to degraded performance on larger filesystems, but is otherwise
60 usable, even on multiple devices.
62 The 'nodesize' and 'sectorsize' must be equal, and the block group types must
65 NOTE: versions up to 4.2.x forced the mixed mode for devices smaller than 1GiB.
66 This has been removed in 4.3+ as it caused some usability issues.
68 *-l|--leafsize <size>*::
69 Alias for --nodesize. Deprecated.
71 *-n|--nodesize <size>*::
72 Specify the nodesize, the tree block size in which btrfs stores metadata. The
73 default value is 16KiB (16384) or the page size, whichever is bigger. Must be a
74 multiple of the sectorsize and a power of 2, but not larger than 64KiB (65536).
75 Leafsize always equals nodesize and the options are aliases.
77 Smaller node size increases fragmentation but leads to taller b-trees which in
78 turn leads to lower locking contention. Higher node sizes give better packing
79 and less fragmentation at the cost of more expensive memory operations while
80 updating the metadata blocks.
82 NOTE: versions up to 3.11 set the nodesize to 4k.
84 *-s|--sectorsize <size>*::
85 Specify the sectorsize, the minimum data block allocation unit.
87 The default value is the page size and is autodetected. If the sectorsize
88 differs from the page size, the created filesystem may not be mountable by the
89 kernel. Therefore it is not recommended to use this option unless you are going
90 to mount it on a system with the appropriate page size.
92 *-L|--label <string>*::
93 Specify a label for the filesystem. The 'string' should be less than 256
94 bytes and must not contain newline characters.
97 Do not perform whole device TRIM operation on devices that are capable of that.
98 This does not affect discard/trim operation when the filesystem is mounted.
99 Please see the mount option 'discard' for that in `btrfs`(5).
101 *-r|--rootdir <rootdir>*::
102 Populate the toplevel subvolume with files from 'rootdir'. This does not
103 require root permissions and does not mount the filesystem.
105 NOTE: This option may enlarge the image or file to ensure it's large enough to
106 contain the files from 'rootdir'.
109 Shrink the filesystem to its minimal size, only works with *-r|--rootdir*
112 NOTE: If the destination is regular file, this option will also reduce the
113 file size. Or it will only reduce the filesystem available space.
114 Extra space will not be usable unless resized using 'btrfs filesystem resize'.
116 *-O|--features <feature1>[,<feature2>...]*::
117 A list of filesystem features turned on at mkfs time. Not all features are
118 supported by old kernels. To disable a feature, prefix it with '^'.
120 See section *FILESYSTEM FEATURES* for more details. To see all available
121 features that mkfs.btrfs supports run:
123 +mkfs.btrfs -O list-all+
126 Forcibly overwrite the block devices when an existing filesystem is detected.
127 By default, mkfs.btrfs will utilize 'libblkid' to check for any known
128 filesystem on the devices. Alternatively you can use the `wipefs` utility
129 to clear the devices.
132 Print only error or warning messages. Options --features or --help are unaffected.
135 Create the filesystem with the given 'UUID'. The UUID must not exist on any
136 filesystem currently present.
139 Print the *mkfs.btrfs* version and exit.
144 *-A|--alloc-start <offset>*::
145 *deprecated, will be removed*
146 (An option to help debugging chunk allocator.)
147 Specify the (physical) offset from the start of the device at which allocations
148 start. The default value is zero.
152 The default unit is 'byte'. All size parameters accept suffixes in the 1024
153 base. The recognized suffixes are: 'k', 'm', 'g', 't', 'p', 'e', both uppercase
159 Before mounting a multiple device filesystem, the kernel module must know the
160 association of the block devices that are attached to the filesystem UUID.
162 There is typically no action needed from the user. On a system that utilizes a
163 udev-like daemon, any new block device is automatically registered. The rules
164 call *btrfs device scan*.
166 The same command can be used to trigger the device scanning if the btrfs kernel
167 module is reloaded (naturally all previous information about the device
168 registration is lost).
170 Another possibility is to use the mount options *device* to specify the list of
171 devices to scan at the time of mount.
173 # mount -o device=/dev/sdb,device=/dev/sdc /dev/sda /mnt
175 NOTE: that this means only scanning, if the devices do not exist in the system,
176 mount will fail anyway. This can happen on systems without initramfs/initrd and
177 root partition created with RAID1/10/5/6 profiles. The mount action can happen
178 before all block devices are discovered. The waiting is usually done on the
179 initramfs/initrd systems.
181 As of kernel 4.14, RAID5/6 is still considered experimental and shouldn't be
182 employed for production use.
187 Features that can be enabled during creation time. See also `btrfs`(5) section
188 'FILESYSTEM FEATURES'.
191 (kernel support since 2.6.37)
193 mixed data and metadata block groups, also set by option '--mixed'
196 (default since btrfs-progs 3.12, kernel support since 3.7)
198 increased hardlink limit per file in a directory to 65536, older kernels
199 supported a varying number of hardlinks depending on the sum of all file name
200 sizes that can be stored into one metadata block
203 (kernel support since 3.9)
205 extended format for RAID5/6, also enabled if raid5 or raid6 block groups
209 (default since btrfs-progs 3.18, kernel support since 3.10)
211 reduced-size metadata for extent references, saves a few percent of metadata
214 (kernel support since 3.14)
216 improved representation of file extents where holes are not explicitly
217 stored as an extent, saves a few percent of metadata if sparse files are used
219 BLOCK GROUPS, CHUNKS, RAID
220 --------------------------
222 The highlevel organizational units of a filesystem are block groups of three types:
223 data, metadata and system.
226 store data blocks and nothing else
229 store internal metadata in b-trees, can store file data if they fit into the
233 store structures that describe the mapping between the physical devices and the
234 linear logical space representing the filesystem
236 Other terms commonly used:
240 a logical range of space of a given profile, stores data, metadata or both;
241 sometimes the terms are used interchangeably
243 A typical size of metadata block group is 256MiB (filesystem smaller than
244 50GiB) and 1GiB (larger than 50GiB), for data it's 1GiB. The system block group
245 size is a few megabytes.
248 a block group profile type that utilizes RAID-like features on multiple
249 devices: striping, mirroring, parity
252 when used in connection with block groups refers to the allocation strategy
253 and constraints, see the section 'PROFILES' for more details
258 There are the following block group types available:
260 [ cols="^,^,^,^,^",width="60%" ]
261 |=============================================================
262 .2+^.<h| Profile 3+^.^h| Redundancy .2+^.<h| Min/max devices
263 ^.^h| Copies ^.^h| Parity ^.<h| Striping
264 | single | 1 | | | 1/any
265 | DUP | 2 / 1 device | | | 1/any ^(see note 1)^
266 | RAID0 | | | 1 to N | 2/any
267 | RAID1 | 2 | | | 2/any
268 | RAID10 | 2 | | 1 to N | 4/any
269 | RAID5 | 1 | 1 | 2 to N - 1 | 2/any ^(see note 2)^
270 | RAID6 | 1 | 2 | 3 to N - 2 | 3/any ^(see note 3)^
271 |=============================================================
273 WARNING: It's not recommended to build btrfs with RAID0/1/10/5/6 profiles on
274 partitions from the same device. Neither redundancy nor performance will be
277 'Note 1:' DUP may exist on more than 1 device if it starts on a single device and
278 another one is added. Since version 4.5.1, *mkfs.btrfs* will let you create DUP
281 'Note 2:' It's not recommended to use 2 devices with RAID5. In that case,
282 parity stripe will contain the same data as the data stripe, making RAID5
283 degraded to RAID1 with more overhead.
285 'Note 3:' It's also not recommended to use 3 devices with RAID6, unless you
286 want to get effectively 3 copies in a RAID1-like manner (but not exactly that).
287 N-copies RAID1 is not implemented.
289 DUP PROFILES ON A SINGLE DEVICE
290 -------------------------------
292 The mkfs utility will let the user create a filesystem with profiles that write
293 the logical blocks to 2 physical locations. Whether there are really 2
294 physical copies highly depends on the underlying device type.
296 For example, a SSD drive can remap the blocks internally to a single copy--thus
297 deduplicating them. This negates the purpose of increased redundancy and just
298 wastes filesystem space without providing the expected level of redundancy.
300 The duplicated data/metadata may still be useful to statistically improve the
301 chances on a device that might perform some internal optimizations. The actual
302 details are not usually disclosed by vendors. For example we could expect that
303 not all blocks get deduplicated. This will provide a non-zero probability of
304 recovery compared to a zero chance if the single profile is used. The user
305 should make the tradeoff decision. The deduplication in SSDs is thought to be
306 widely available so the reason behind the mkfs default is to not give a false
309 As another example, the widely used USB flash or SD cards use a translation
310 layer between the logical and physical view of the device. The data lifetime
311 may be affected by frequent plugging. The memory cells could get damaged,
312 hopefully not destroying both copies of particular data in case of DUP.
314 The wear levelling techniques can also lead to reduced redundancy, even if the
315 device does not do any deduplication. The controllers may put data written in
316 a short timespan into the same physical storage unit (cell, block etc). In case
317 this unit dies, both copies are lost. BTRFS does not add any artificial delay
318 between metadata writes.
320 The traditional rotational hard drives usually fail at the sector level.
322 In any case, a device that starts to misbehave and repairs from the DUP copy
323 should be replaced! *DUP is not backup*.
328 **SMALL FILESYSTEMS AND LARGE NODESIZE**
330 The combination of small filesystem size and large nodesize is not recommended
331 in general and can lead to various ENOSPC-related issues during mount time or runtime.
333 Since mixed block group creation is optional, we allow small
334 filesystem instances with differing values for 'sectorsize' and 'nodesize'
335 to be created and could end up in the following situation:
337 # mkfs.btrfs -f -n 65536 /dev/loop0
338 btrfs-progs v3.19-rc2-405-g976307c
339 See http://btrfs.wiki.kernel.org for more information.
341 Performing full device TRIM (512.00MiB) ...
343 UUID: 49fab72e-0c8b-466b-a3ca-d1bfe56475f0
346 Filesystem size: 512.00MiB
347 Block group profiles:
349 Metadata: DUP 40.00MiB
352 Incompat features: extref, skinny-metadata
356 1 512.00MiB /dev/loop0
358 # mount /dev/loop0 /mnt/
359 mount: mount /dev/loop0 on /mnt failed: No space left on device
361 The ENOSPC occurs during the creation of the UUID tree. This is caused
362 by large metadata blocks and space reservation strategy that allocates more
363 than can fit into the filesystem.
368 *mkfs.btrfs* is part of btrfs-progs.
369 Please refer to the btrfs wiki http://btrfs.wiki.kernel.org for