6 mkfs.btrfs - create a btrfs filesystem
11 $$[-A|--alloc-start <alloc-start>]$$
12 $$[-b|--byte-count <byte-count>]$$
13 $$[-d|--data <data-profile>]$$
14 $$[-m|--metadata <metadata profile>]$$
16 $$[-l|--leafsize <leafsize>]$$
17 $$[-n|--nodesize <nodesize>]$$
18 $$[-s|--sectorsize <sectorsize>]$$
19 $$[-L|--label <label>]$$
21 $$[-r|--rootdir <rootdir>]$$
22 $$[-O|--features <feature1>[,<feature2>...]]$$
23 $$[-U|--uuid <UUID>]$$
28 $$<device> [<device>...]$$
32 *mkfs.btrfs* is used to create the btrfs filesystem on a single or multiple
33 devices. <device> is typically a block device but can be a file-backed image
34 as well. Multiple devices are grouped by UUID of the filesystem.
36 Before mounting such filesystem, the kernel module must know all the devices
37 either via preceding execution of *btrfs device scan* or using the *device*
38 mount option. See section *MULTIPLE DEVICES* for more details.
42 *-A|--alloc-start <offset>*::
43 (An option to help debugging chunk allocator.)
44 Specify the (physical) offset from the start of the device at which allocations
45 start. The default value is zero.
47 *-b|--byte-count <size>*::
48 Specify the size of the filesystem. If this option is not used,
49 mkfs.btrfs uses the entire device space for the filesystem.
51 *-d|--data <profile>*::
52 Specify the profile for the data block groups. Valid values are 'raid0',
53 'raid1', 'raid5', 'raid6', 'raid10' or 'single' or dup (case does not matter).
55 See 'DUP PROFILES ON A SINGLE DEVICE' for more.
57 *-m|--metadata <profile>*::
58 Specify the profile for the metadata block groups.
59 Valid values are 'raid0', 'raid1', 'raid5', 'raid6', 'raid10', 'single' or
60 'dup', (case does not matter).
62 A single device filesystem will default to 'DUP', unless a SSD is detected. Then
63 it will default to 'single'. The detection is based on the value of
64 `/sys/block/DEV/queue/rotational`, where 'DEV' is the short name of the device.
66 Note that the rotational status can be arbitrarily set by the underlying block
67 device driver and may not reflect the true status (network block device, memory-backed
68 SCSI devices etc). Use the options '--data/--metadata' to avoid confusion.
70 See 'DUP PROFILES ON A SINGLE DEVICE' for more details.
73 Normally the data and metadata block groups are isolated. The 'mixed' mode
74 will remove the isolation and store both types in the same block group type.
75 This helps to utilize the free space regardless of the purpose and is suitable
76 for small devices. The separate allocation of block groups leads to a situation
77 where the space is reserved for the other block group type, is not available for
78 allocation and can lead to ENOSPC state.
80 The recommended size for the mixed mode is for filesystems less than 1GiB. The
81 soft recommendation is to use it for filesystems smaller than 5GiB. The mixed
82 mode may lead to degraded performance on larger filesystems, but is otherwise
83 usable, even on multiple devices.
85 The 'nodesize' and 'sectorsize' must be equal, and the block group types must
88 NOTE: versions up to 4.2.x forced the mixed mode for devices smaller than 1GiB.
89 This has been removed in 4.3+ as it caused some usability issues.
91 *-l|--leafsize <size>*::
92 Alias for --nodesize. Deprecated.
94 *-n|--nodesize <size>*::
95 Specify the nodesize, the tree block size in which btrfs stores metadata. The
96 default value is 16KiB (16384) or the page size, whichever is bigger. Must be a
97 multiple of the sectorsize, but not larger than 64KiB (65536). Leafsize always
98 equals nodesize and the options are aliases.
100 Smaller node size increases fragmentation but lead to higher b-trees which in
101 turn leads to lower locking contention. Higher node sizes give better packing
102 and less fragmentation at the cost of more expensive memory operations while
103 updating the metadata blocks.
105 NOTE: versions up to 3.11 set the nodesize to 4k.
107 *-s|--sectorsize <size>*::
108 Specify the sectorsize, the minimum data block allocation unit.
110 The default value is the page size and is autodetected. If the sectorsize
111 differs from the page size, the created filesystem may not be mountable by the
112 kernel. Therefore it is not recommended to use this option unless you are going
113 to mount it on a system with the appropriate page size.
115 *-L|--label <string>*::
116 Specify a label for the filesystem. The 'string' should be less than 256
117 bytes and must not contain newline characters.
120 Do not perform whole device TRIM operation on devices that are capable of that.
122 *-r|--rootdir <rootdir>*::
123 Populate the toplevel subvolume with files from 'rootdir'. This does not
124 require root permissions and does not mount the filesystem.
126 *-O|--features <feature1>[,<feature2>...]*::
127 A list of filesystem features turned on at mkfs time. Not all features are
128 supported by old kernels. To disable a feature, prefix it with '^'.
130 See section *FILESYSTEM FEATURES* for more details. To see all available
131 features that mkfs.btrfs supports run:
133 +mkfs.btrfs -O list-all+
136 Forcibly overwrite the block devices when an existing filesystem is detected.
137 By default, mkfs.btrfs will utilize 'libblkid' to check for any known
138 filesystem on the devices. Alternatively you can use the `wipefs` utility
139 to clear the devices.
142 Print only error or warning messages. Options --features or --help are unaffected.
145 Create the filesystem with the given 'UUID'. The UUID must not exist on any
146 filesystem currently present.
149 Print the *mkfs.btrfs* version and exit.
156 The default unit is 'byte'. All size parameters accept suffixes in the 1024
157 base. The recognized suffixes are: 'k', 'm', 'g', 't', 'p', 'e', both uppercase
163 Before mounting a multiple device filesystem, the kernel module must know the
164 association of the block devices that are attached to the filesystem UUID.
166 There is typically no action needed from the user. On a system that utilizes a
167 udev-like daemon, any new block device is automatically registered. The rules
168 call *btrfs device scan*.
170 The same command can be used to trigger the device scanning if the btrfs kernel
171 module is reloaded (naturally all previous information about the device
172 registration is lost).
174 Another possibility is to use the mount options *device* to specify the list of
175 devices to scan at the time of mount.
177 # mount -o device=/dev/sdb,device=/dev/sdc /dev/sda /mnt
179 NOTE: that this means only scanning, if the devices do not exist in the system,
180 mount will fail anyway. This can happen on systems without initramfs/initrd and
181 root partition created with RAID1/10/5/6 profiles. The mount action can happen
182 before all block devices are discovered. The waiting is usually done on the
183 initramfs/initrd systems.
189 mixed data and metadata block groups, also set by option '--mixed'
192 (default since btrfs-progs 3.12, kernel support since 3.7)
194 increased hardlink limit per file in a directory to 65536, older kernels
195 supported a varying number of hardlinks depending on the sum of all file name
196 sizes that can be stored into one metadata block
199 extended format for RAID5/6, also enabled if raid5 or raid6 block groups
203 (default since btrfs-progs 3.18, kernel support since 3.10)
205 reduced-size metadata for extent references, saves a few percent of metadata
208 improved representation of file extents where holes are not explicitly
209 stored as an extent, saves a few percent of metadata if sparse files are used
211 BLOCK GROUPS, CHUNKS, RAID
212 --------------------------
214 The highlevel organizational units of a filesystem are block groups of three types:
215 data, metadata and system.
218 store data blocks and nothing else
221 store internal metadata in b-trees, can store file data if they fit into the
225 store structures that describe the mapping between the physical devices and the
226 linear logical space representing the filesystem
228 Other terms commonly used:
232 a logical range of space of a given profile, stores data, metadata or both;
233 sometimes the terms are used interchangably
235 A typical size of metadata block group is 256MiB (filesystem smaller than
236 50GiB) and 1GiB (larger than 50GiB), for data it's 1GiB. The system block group
237 size is a few megabytes.
240 a block group profile type that utilizes RAID-like features on multiple
241 devices: striping, mirroring, parity
244 when used in connection with block groups refers to the allocation strategy
245 and constraints, see the section 'PROFILES' for more details
250 There are the following block group types available:
252 [ cols="^,^,^,^,^",width="60%" ]
253 |=============================================================
254 .2+^.<h| Profile 3+^.^h| Redundancy .2+^.<h| Min/max devices
255 ^.^h| Copies ^.^h| Parity ^.<h| Striping
256 | single | 1 | | | 1/any
257 | DUP | 2 / 1 device | | | 1/1 ^(see note)^
258 | RAID0 | | | 1 to N | 2/any
259 | RAID1 | 2 | | | 2/any
260 | RAID10 | 2 | | 1 to N | 4/any
261 | RAID5 | 1 | 1 | 2 to N - 1 | 2/any
262 | RAID6 | 1 | 2 | 3 to N - 2 | 3/any
263 |=============================================================
265 'Note:' DUP may exist on more than 1 device if it starts on a single device and
266 another one is added, but *mkfs.btrfs* will not let you create DUP on multiple
269 DUP PROFILES ON A SINGLE DEVICE
270 -------------------------------
272 The mkfs utility will let the user create a filesystem with profiles that write
273 the logical blocks to 2 physical locations. Whether there are really 2
274 physical copies highly depends on the underlying device type.
276 For example, a SSD drive can remap the blocks internally to a single copy thus
277 deduplicating them. This negates the purpose of increased redunancy and just
280 The duplicated data/metadata may still be useful to statistically improve the
281 chances on a device that might perform some internal optimizations. The actual
282 details are not usually disclosed by vendors. As another example, the widely
283 used USB flash or SD cards use a translation layer. The data lifetime may
284 be affected by frequent plugging. The memory cells could get damaged, hopefully
285 not destroying both copies of particular data.
287 The traditional rotational hard drives usually fail at the sector level.
289 In any case, a device that starts to misbehave and repairs from the DUP copy
290 should be replaced! *DUP is not backup*.
295 **SMALL FILESYSTEMS AND LARGE NODESIZE**
297 The combination of small filesystem size and large nodesize is not recommended
298 in general and can lead to various ENOSPC-related issues during mount time or runtime.
300 Since mixed block group creation is optional, we allow small
301 filesystem instances with differing values for 'sectorsize' and 'nodesize'
302 to be created and could end up in the following situation:
304 # mkfs.btrfs -f -n 65536 /dev/loop0
305 btrfs-progs v3.19-rc2-405-g976307c
306 See http://btrfs.wiki.kernel.org for more information.
308 Performing full device TRIM (512.00MiB) ...
310 UUID: 49fab72e-0c8b-466b-a3ca-d1bfe56475f0
313 Filesystem size: 512.00MiB
314 Block group profiles:
316 Metadata: DUP 40.00MiB
319 Incompat features: extref, skinny-metadata
323 1 512.00MiB /dev/loop0
325 # mount /dev/loop0 /mnt/
326 mount: mount /dev/loop0 on /mnt failed: No space left on device
328 The ENOSPC occurs during the creation of the UUID tree. This is caused
329 by large metadata blocks and space reservation strategy that allocates more
330 than can fit into the filesystem.
335 *mkfs.btrfs* is part of btrfs-progs.
336 Please refer to the btrfs wiki http://btrfs.wiki.kernel.org for
341 `btrfs`(8), `wipefs`(8)