1MKFS.BTRFS(8) Btrfs Manual MKFS.BTRFS(8)
2
3
4
6 mkfs.btrfs - create a btrfs filesystem
7
9 mkfs.btrfs [options] <device> [<device>...]
10
12 mkfs.btrfs is used to create the btrfs filesystem on a single or
13 multiple devices. <device> is typically a block device but can be a
14 file-backed image as well. Multiple devices are grouped by UUID of the
15 filesystem.
16
17 Before mounting such filesystem, the kernel module must know all the
18 devices either via preceding execution of btrfs device scan or using
19 the device mount option. See section MULTIPLE DEVICES for more details.
20
22 -A|--alloc-start <offset>
23 (An option to help debugging chunk allocator.) Specify the
24 (physical) offset from the start of the device at which allocations
25 start. The default value is zero.
26
27 -b|--byte-count <size>
28 Specify the size of the filesystem. If this option is not used,
29 mkfs.btrfs uses the entire device space for the filesystem.
30
31 -d|--data <profile>
32 Specify the profile for the data block groups. Valid values are
33 raid0, raid1, raid5, raid6, raid10 or single or dup (case does not
34 matter).
35
36 See DUP PROFILES ON A SINGLE DEVICE for more.
37
38 -m|--metadata <profile>
39 Specify the profile for the metadata block groups. Valid values are
40 raid0, raid1, raid5, raid6, raid10, single or dup, (case does not
41 matter).
42
43 A single device filesystem will default to DUP, unless a SSD is
44 detected. Then it will default to single. The detection is based on
45 the value of /sys/block/DEV/queue/rotational, where DEV is the
46 short name of the device.
47
48 Note that the rotational status can be arbitrarily set by the
49 underlying block device driver and may not reflect the true status
50 (network block device, memory-backed SCSI devices etc). Use the
51 options --data/--metadata to avoid confusion.
52
53 See DUP PROFILES ON A SINGLE DEVICE for more details.
54
55 -M|--mixed
56 Normally the data and metadata block groups are isolated. The mixed
57 mode will remove the isolation and store both types in the same
58 block group type. This helps to utilize the free space regardless
59 of the purpose and is suitable for small devices. The separate
60 allocation of block groups leads to a situation where the space is
61 reserved for the other block group type, is not available for
62 allocation and can lead to ENOSPC state.
63
64 The recommended size for the mixed mode is for filesystems less
65 than 1GiB. The soft recommendation is to use it for filesystems
66 smaller than 5GiB. The mixed mode may lead to degraded performance
67 on larger filesystems, but is otherwise usable, even on multiple
68 devices.
69
70 The nodesize and sectorsize must be equal, and the block group
71 types must match.
72
73 Note
74 versions up to 4.2.x forced the mixed mode for devices smaller
75 than 1GiB. This has been removed in 4.3+ as it caused some
76 usability issues.
77
78 -l|--leafsize <size>
79 Alias for --nodesize. Deprecated.
80
81 -n|--nodesize <size>
82 Specify the nodesize, the tree block size in which btrfs stores
83 metadata. The default value is 16KiB (16384) or the page size,
84 whichever is bigger. Must be a multiple of the sectorsize, but not
85 larger than 64KiB (65536). Leafsize always equals nodesize and the
86 options are aliases.
87
88 Smaller node size increases fragmentation but lead to higher
89 b-trees which in turn leads to lower locking contention. Higher
90 node sizes give better packing and less fragmentation at the cost
91 of more expensive memory operations while updating the metadata
92 blocks.
93
94 Note
95 versions up to 3.11 set the nodesize to 4k.
96
97 -s|--sectorsize <size>
98 Specify the sectorsize, the minimum data block allocation unit.
99
100 The default value is the page size and is autodetected. If the
101 sectorsize differs from the page size, the created filesystem may
102 not be mountable by the kernel. Therefore it is not recommended to
103 use this option unless you are going to mount it on a system with
104 the appropriate page size.
105
106 -L|--label <string>
107 Specify a label for the filesystem. The string should be less than
108 256 bytes and must not contain newline characters.
109
110 -K|--nodiscard
111 Do not perform whole device TRIM operation on devices that are
112 capable of that. This does not affect discard/trim operation when
113 the filesystem is mounted. Please see the mount option discard for
114 that in btrfs(5).
115
116 -r|--rootdir <rootdir>
117 Populate the toplevel subvolume with files from rootdir. This does
118 not require root permissions and does not mount the filesystem.
119
120 -O|--features <feature1>[,<feature2>...]
121 A list of filesystem features turned on at mkfs time. Not all
122 features are supported by old kernels. To disable a feature, prefix
123 it with ^.
124
125 See section FILESYSTEM FEATURES for more details. To see all
126 available features that mkfs.btrfs supports run:
127
128 mkfs.btrfs -O list-all
129
130 -f|--force
131 Forcibly overwrite the block devices when an existing filesystem is
132 detected. By default, mkfs.btrfs will utilize libblkid to check for
133 any known filesystem on the devices. Alternatively you can use the
134 wipefs utility to clear the devices.
135
136 -q|--quiet
137 Print only error or warning messages. Options --features or --help
138 are unaffected.
139
140 -U|--uuid <UUID>
141 Create the filesystem with the given UUID. The UUID must not exist
142 on any filesystem currently present.
143
144 -V|--version
145 Print the mkfs.btrfs version and exit.
146
147 --help
148 Print help.
149
151 The default unit is byte. All size parameters accept suffixes in the
152 1024 base. The recognized suffixes are: k, m, g, t, p, e, both
153 uppercase and lowercase.
154
156 Before mounting a multiple device filesystem, the kernel module must
157 know the association of the block devices that are attached to the
158 filesystem UUID.
159
160 There is typically no action needed from the user. On a system that
161 utilizes a udev-like daemon, any new block device is automatically
162 registered. The rules call btrfs device scan.
163
164 The same command can be used to trigger the device scanning if the
165 btrfs kernel module is reloaded (naturally all previous information
166 about the device registration is lost).
167
168 Another possibility is to use the mount options device to specify the
169 list of devices to scan at the time of mount.
170
171 # mount -o device=/dev/sdb,device=/dev/sdc /dev/sda /mnt
172
173
174 Note
175 that this means only scanning, if the devices do not exist in the
176 system, mount will fail anyway. This can happen on systems without
177 initramfs/initrd and root partition created with RAID1/10/5/6
178 profiles. The mount action can happen before all block devices are
179 discovered. The waiting is usually done on the initramfs/initrd
180 systems.
181
182 As of kernel 4.9, RAID5/6 is still considered experimental and
183 shouldn’t be employed for production use.
184
186 Features that can be enabled during creation time. See also btrfs(5)
187 section FILESYSTEM FEATURES.
188
189 mixed-bg
190 (kernel support since 2.6.37)
191
192 mixed data and metadata block groups, also set by option --mixed
193
194 extref
195 (default since btrfs-progs 3.12, kernel support since 3.7)
196
197 increased hardlink limit per file in a directory to 65536, older
198 kernels supported a varying number of hardlinks depending on the
199 sum of all file name sizes that can be stored into one metadata
200 block
201
202 raid56
203 (kernel support since 3.9)
204
205 extended format for RAID5/6, also enabled if raid5 or raid6 block
206 groups are selected
207
208 skinny-metadata
209 (default since btrfs-progs 3.18, kernel support since 3.10)
210
211 reduced-size metadata for extent references, saves a few percent of
212 metadata
213
214 no-holes
215 (kernel support since 3.14)
216
217 improved representation of file extents where holes are not
218 explicitly stored as an extent, saves a few percent of metadata if
219 sparse files are used
220
222 The highlevel organizational units of a filesystem are block groups of
223 three types: data, metadata and system.
224
225 DATA
226 store data blocks and nothing else
227
228 METADATA
229 store internal metadata in b-trees, can store file data if they fit
230 into the inline limit
231
232 SYSTEM
233 store structures that describe the mapping between the physical
234 devices and the linear logical space representing the filesystem
235
236 Other terms commonly used:
237
238 block group, chunk
239 a logical range of space of a given profile, stores data, metadata
240 or both; sometimes the terms are used interchangeably
241
242 A typical size of metadata block group is 256MiB (filesystem
243 smaller than 50GiB) and 1GiB (larger than 50GiB), for data it’s
244 1GiB. The system block group size is a few megabytes.
245
246 RAID
247 a block group profile type that utilizes RAID-like features on
248 multiple devices: striping, mirroring, parity
249
250 profile
251 when used in connection with block groups refers to the allocation
252 strategy and constraints, see the section PROFILES for more details
253
255 There are the following block group types available:
256
257 ┌────────┬────────────────────────────────────┬────────────┐
258 │ │ │ │
259 │Profile │ Redundancy │ Min/max │
260 │ ├──────────────┬────────┬────────────┤ devices │
261 │ │ │ │ │ │
262 │ │ Copies │ Parity │ Striping │ │
263 ├────────┼──────────────┼────────┼────────────┼────────────┤
264 │ │ │ │ │ │
265 │single │ 1 │ │ │ 1/any │
266 ├────────┼──────────────┼────────┼────────────┼────────────┤
267 │ │ │ │ │ │
268 │ DUP │ 2 / 1 device │ │ │ 1/any (see │
269 │ │ │ │ │ note 1) │
270 ├────────┼──────────────┼────────┼────────────┼────────────┤
271 │ │ │ │ │ │
272 │ RAID0 │ │ │ 1 to N │ 2/any │
273 ├────────┼──────────────┼────────┼────────────┼────────────┤
274 │ │ │ │ │ │
275 │ RAID1 │ 2 │ │ │ 2/any │
276 ├────────┼──────────────┼────────┼────────────┼────────────┤
277 │ │ │ │ │ │
278 │RAID10 │ 2 │ │ 1 to N │ 4/any │
279 ├────────┼──────────────┼────────┼────────────┼────────────┤
280 │ │ │ │ │ │
281 │ RAID5 │ 1 │ 1 │ 2 to N - 1 │ 2/any (see │
282 │ │ │ │ │ note 2) │
283 ├────────┼──────────────┼────────┼────────────┼────────────┤
284 │ │ │ │ │ │
285 │ RAID6 │ 1 │ 2 │ 3 to N - 2 │ 3/any (see │
286 │ │ │ │ │ note 3) │
287 └────────┴──────────────┴────────┴────────────┴────────────┘
288
289 Warning
290 It’s not recommended to build btrfs with RAID0/1/10/5/6 profiles on
291 partitions from the same device. Neither redundancy nor performance
292 will be improved.
293
294 Note 1: DUP may exist on more than 1 device if it starts on a single
295 device and another one is added. Since version 4.5.1, mkfs.btrfs will
296 let you create DUP on multiple devices.
297
298 Note 2: It’s not recommended to use 2 devices with RAID5. In that case,
299 parity stripe will contain the same data as the data stripe, making
300 RAID5 degraded to RAID1 with more overhead.
301
302 Note 3: It’s also not recommended to use 3 devices with RAID6, unless
303 you want to get effectively 3 copies in a RAID1-like manner (but not
304 exactly that). N-copies RAID1 is not implemented.
305
307 The mkfs utility will let the user create a filesystem with profiles
308 that write the logical blocks to 2 physical locations. Whether there
309 are really 2 physical copies highly depends on the underlying device
310 type.
311
312 For example, a SSD drive can remap the blocks internally to a single
313 copy thus deduplicating them. This negates the purpose of increased
314 redundancy and just wastes filesystem space without the expected level
315 of redundancy.
316
317 The duplicated data/metadata may still be useful to statistically
318 improve the chances on a device that might perform some internal
319 optimizations. The actual details are not usually disclosed by vendors.
320 For example we could expect that not all blocks get deduplicated. This
321 will provide a non-zero probability of recovery compared to a zero
322 chance if the single profile is used. The user should make the tradeoff
323 decision. The deduplication in SSDs is thought to be widely available
324 so the reason behind the mkfs default is to not give a false sense of
325 redundancy.
326
327 As another example, the widely used USB flash or SD cards use a
328 translation layer between the logical and physical view of the device.
329 The data lifetime may be affected by frequent plugging. The memory
330 cells could get damaged, hopefully not destroying both copies of
331 particular data in case of DUP.
332
333 The wear levelling techniques can also lead to reduced redundancy, even
334 if the device does not do any deduplication. The controllers may put
335 data written in a short timespan into the same physical storage unit
336 (cell, block etc). In case this unit dies, both copies are lost. BTRFS
337 does not add any artificial delay between metadata writes.
338
339 The traditional rotational hard drives usually fail at the sector
340 level.
341
342 In any case, a device that starts to misbehave and repairs from the DUP
343 copy should be replaced! DUP is not backup.
344
346 SMALL FILESYSTEMS AND LARGE NODESIZE
347
348 The combination of small filesystem size and large nodesize is not
349 recommended in general and can lead to various ENOSPC-related issues
350 during mount time or runtime.
351
352 Since mixed block group creation is optional, we allow small filesystem
353 instances with differing values for sectorsize and nodesize to be
354 created and could end up in the following situation:
355
356 # mkfs.btrfs -f -n 65536 /dev/loop0
357 btrfs-progs v3.19-rc2-405-g976307c
358 See http://btrfs.wiki.kernel.org for more information.
359
360 Performing full device TRIM (512.00MiB) ...
361 Label: (null)
362 UUID: 49fab72e-0c8b-466b-a3ca-d1bfe56475f0
363 Node size: 65536
364 Sector size: 4096
365 Filesystem size: 512.00MiB
366 Block group profiles:
367 Data: single 8.00MiB
368 Metadata: DUP 40.00MiB
369 System: DUP 12.00MiB
370 SSD detected: no
371 Incompat features: extref, skinny-metadata
372 Number of devices: 1
373 Devices:
374 ID SIZE PATH
375 1 512.00MiB /dev/loop0
376
377 # mount /dev/loop0 /mnt/
378 mount: mount /dev/loop0 on /mnt failed: No space left on device
379
380 The ENOSPC occurs during the creation of the UUID tree. This is caused
381 by large metadata blocks and space reservation strategy that allocates
382 more than can fit into the filesystem.
383
385 mkfs.btrfs is part of btrfs-progs. Please refer to the btrfs wiki
386 http://btrfs.wiki.kernel.org for further details.
387
389 btrfs(5), btrfs(8), wipefs(8)
390
391
392
393Btrfs v4.9.1 08/06/2017 MKFS.BTRFS(8)