2 .\" Copyright Neil Brown and others.
3 .\" This program is free software; you can redistribute it and/or modify
4 .\" it under the terms of the GNU General Public License as published by
5 .\" the Free Software Foundation; either version 2 of the License, or
6 .\" (at your option) any later version.
7 .\" See file COPYING in distribution for details.
10 mdadm \- manage MD devices
16 .BI mdadm " [mode] <raiddevice> [options] <component-devices>"
19 RAID devices are virtual devices created from two or more
20 real block devices. This allows multiple devices (typically disk
21 drives or partitions thereof) to be combined into a single device to
22 hold (for example) a single filesystem.
23 Some RAID levels include redundancy and so can survive some degree of
26 Linux Software RAID devices are implemented through the md (Multiple
27 Devices) device driver.
29 Currently, Linux supports
46 is not a Software RAID mechanism, but does involve
48 each device is a path to one common physical storage device.
49 New installations should not use md/multipath as it is not well
50 supported and has no ongoing development. Use the Device Mapper based
51 multipath-tools instead.
54 is also not true RAID, and it only involves one device. It
55 provides a layer over a true device that can be used to inject faults.
60 is a collection of devices that are
61 managed as a set. This is similar to the set of devices connected to
62 a hardware RAID controller. The set of devices may contain a number
63 of different RAID arrays each utilising some (or all) of the blocks from a
64 number of the devices in the set. For example, two devices in a 5-device set
65 might form a RAID1 using the whole devices. The remaining three might
66 have a RAID5 over the first half of each device, and a RAID0 over the
71 there is one set of metadata that describes all of
72 the arrays in the container. So when
76 device, the device just represents the metadata. Other normal arrays (RAID1
77 etc) can be created inside the container.
80 mdadm has several major modes of operation:
83 Assemble the components of a previously created
84 array into an active array. Components can be explicitly given
85 or can be searched for.
87 checks that the components
88 do form a bona fide array, and can, on request, fiddle superblock
89 information so as to assemble a faulty array.
93 Build an array that doesn't have per-device metadata (superblocks). For these
96 cannot differentiate between initial creation and subsequent assembly
97 of an array. It also cannot perform any checks that appropriate
98 components have been requested. Because of this, the
100 mode should only be used together with a complete understanding of
105 Create a new array with per-device metadata (superblocks).
106 Appropriate metadata is written to each device, and then the array
107 comprising those devices is activated. A 'resync' process is started
108 to make sure that the array is consistent (e.g. both sides of a mirror
109 contain the same data) but the content of the device is left otherwise
111 The array can be used as soon as it has been created. There is no
112 need to wait for the initial resync to finish.
115 .B "Follow or Monitor"
116 Monitor one or more md devices and act on any state changes. This is
117 only meaningful for RAID1, 4, 5, 6, 10 or multipath arrays, as
118 only these have interesting state. RAID0 or Linear never have
119 missing, spare, or failed drives, so there is nothing to monitor.
123 Grow (or shrink) an array, or otherwise reshape it in some way.
124 Currently supported growth options including changing the active size
125 of component devices and changing the number of active devices in RAID
126 levels 1/4/5/6, changing the RAID level between 1, 5, and 6, changing
127 the chunk size and layout for RAID5 and RAID5, as well as adding or
128 removing a write-intent bitmap.
131 .B "Incremental Assembly"
132 Add a single device to an appropriate array. If the addition of the
133 device makes the array runnable, the array will be started.
134 This provides a convenient interface to a
136 system. As each device is detected,
138 has a chance to include it in some array as appropriate.
141 flag is passed in we will remove the device from any active array
142 instead of adding it.
148 in this mode, then any arrays within that container will be assembled
153 This is for doing things to specific components of an array such as
154 adding new spares and removing faulty devices.
158 This is an 'everything else' mode that supports operations on active
159 arrays, operations on component devices such as erasing old superblocks, and
160 information gathering operations.
161 .\"This mode allows operations on independent devices such as examine MD
162 .\"superblocks, erasing old superblocks and stopping active arrays.
166 This mode does not act on a specific device or array, but rather it
167 requests the Linux Kernel to activate any auto-detected arrays.
170 .SH Options for selecting a mode are:
173 .BR \-A ", " \-\-assemble
174 Assemble a pre-existing array.
177 .BR \-B ", " \-\-build
178 Build a legacy array without superblocks.
181 .BR \-C ", " \-\-create
185 .BR \-F ", " \-\-follow ", " \-\-monitor
191 .BR \-G ", " \-\-grow
192 Change the size or shape of an active array.
195 .BR \-I ", " \-\-incremental
196 Add/remove a single device to/from an appropriate array, and possibly start the array.
200 Request that the kernel starts any auto-detected arrays. This can only
203 is compiled into the kernel \(em not if it is a module.
204 Arrays can be auto-detected by the kernel if all the components are in
205 primary MS-DOS partitions with partition type
207 and all use v0.90 metadata.
208 In-kernel autodetect is not recommended for new installations. Using
210 to detect and assemble arrays \(em possibly in an
212 \(em is substantially more flexible and should be preferred.
215 If a device is given before any options, or if the first option is
220 then the MANAGE mode is assumed.
221 Anything other than these will cause the
225 .SH Options that are not mode-specific are:
228 .BR \-h ", " \-\-help
229 Display general help message or, after one of the above options, a
230 mode-specific help message.
234 Display more detailed help about command line parsing and some commonly
238 .BR \-V ", " \-\-version
239 Print version information for mdadm.
242 .BR \-v ", " \-\-verbose
243 Be more verbose about what is happening. This can be used twice to be
245 The extra verbosity currently only affects
246 .B \-\-detail \-\-scan
248 .BR "\-\-examine \-\-scan" .
251 .BR \-q ", " \-\-quiet
252 Avoid printing purely informative messages. With this,
254 will be silent unless there is something really important to report.
257 .BR \-f ", " \-\-force
258 Be more forceful about certain operations. See the various modes for
259 the exact meaning of this option in different contexts.
262 .BR \-c ", " \-\-config=
263 Specify the config file. Default is to use
264 .BR /etc/mdadm.conf ,
265 or if that is missing then
266 .BR /etc/mdadm/mdadm.conf .
267 If the config file given is
269 then nothing will be read, but
271 will act as though the config file contained exactly
272 .B "DEVICE partitions containers"
275 to find a list of devices to scan, and
277 to find a list of containers to examine.
280 is given for the config file, then
282 will act as though the config file were empty.
285 .BR \-s ", " \-\-scan
288 for missing information.
289 In general, this option gives
291 permission to get any missing information (like component devices,
292 array devices, array identities, and alert destination) from the
293 configuration file (see previous option);
294 one exception is MISC mode when using
300 says to get a list of array devices from
304 .BR \-e ", " \-\-metadata=
305 Declare the style of RAID metadata (superblock) to be used. The
306 default is {DEFAULT_METADATA} for
308 and to guess for other operations.
309 The default can be overridden by setting the
318 .ie '{DEFAULT_METADATA}'0.90'
319 .IP "0, 0.90, default"
323 Use the original 0.90 format superblock. This format limits arrays to
324 28 component devices and limits component devices of levels 1 and
325 greater to 2 terabytes. It is also possible for there to be confusion
326 about whether the superblock applies to a whole device or just the
327 last partition, if that partition starts on a 64K boundary.
328 .ie '{DEFAULT_METADATA}'0.90'
329 .IP "1, 1.0, 1.1, 1.2"
331 .IP "1, 1.0, 1.1, 1.2 default"
333 Use the new version-1 format superblock. This has fewer restrictions.
334 It can easily be moved between hosts with different endian-ness, and a
335 recovery operation can be checkpointed and restarted. The different
336 sub-versions store the superblock at different locations on the
337 device, either at the end (for 1.0), at the start (for 1.1) or 4K from
338 the start (for 1.2). "1" is equivalent to "1.0".
339 'if '{DEFAULT_METADATA}'1.2' "default" is equivalent to "1.2".
341 Use the "Industry Standard" DDF (Disk Data Format) format defined by
343 When creating a DDF array a
345 will be created, and normal arrays can be created in that container.
347 Use the Intel(R) Matrix Storage Manager metadata format. This creates a
349 which is managed in a similar manner to DDF, and is supported by an
350 option-rom on some platforms:
352 .B http://www.intel.com/design/chipsets/matrixstorage_sb.htm
358 This will override any
360 setting in the config file and provides the identity of the host which
361 should be considered the home for any arrays.
363 When creating an array, the
365 will be recorded in the metadata. For version-1 superblocks, it will
366 be prefixed to the array name. For version-0.90 superblocks, part of
367 the SHA1 hash of the hostname will be stored in the later half of the
370 When reporting information about an array, any array which is tagged
371 for the given homehost will be reported as such.
373 When using Auto-Assemble, only arrays tagged for the given homehost
374 will be allowed to use 'local' names (i.e. not ending in '_' followed
375 by a digit string). See below under
376 .BR "Auto Assembly" .
378 .SH For create, build, or grow:
381 .BR \-n ", " \-\-raid\-devices=
382 Specify the number of active devices in the array. This, plus the
383 number of spare devices (see below) must equal the number of
385 (including "\fBmissing\fP" devices)
386 that are listed on the command line for
388 Setting a value of 1 is probably
389 a mistake and so requires that
391 be specified first. A value of 1 will then be allowed for linear,
392 multipath, RAID0 and RAID1. It is never allowed for RAID4, RAID5 or RAID6.
394 This number can only be changed using
396 for RAID1, RAID4, RAID5 and RAID6 arrays, and only on kernels which provide
397 the necessary support.
400 .BR \-x ", " \-\-spare\-devices=
401 Specify the number of spare (eXtra) devices in the initial array.
402 Spares can also be added
403 and removed later. The number of component devices listed
404 on the command line must equal the number of RAID devices plus the
405 number of spare devices.
408 .BR \-z ", " \-\-size=
409 Amount (in Kibibytes) of space to use from each drive in RAID levels 1/4/5/6.
410 This must be a multiple of the chunk size, and must leave about 128Kb
411 of space at the end of the drive for the RAID superblock.
412 If this is not specified
413 (as it normally is not) the smallest drive (or partition) sets the
414 size, though if there is a variance among the drives of greater than 1%, a warning is
417 A suffix of 'M' or 'G' can be given to indicate Megabytes or
418 Gigabytes respectively.
420 This value can be set with
422 for RAID level 1/4/5/6. If the array was created with a size smaller
423 than the currently active drives, the extra space can be accessed
426 The size can be given as
428 which means to choose the largest size that fits on all current drives.
430 This value can not be used with
432 metadata such as DDF and IMSM.
435 .BR \-Z ", " \-\-array-size=
436 This is only meaningful with
438 and its effect is not persistent: when the array is stopped and
439 restarted the default array size will be restored.
441 Setting the array-size causes the array to appear smaller to programs
442 that access the data. This is particularly needed before reshaping an
443 array so that it will be smaller. As the reshape is not reversible,
444 but setting the size with
446 is, it is required that the array size is reduced as appropriate
447 before the number of devices in the array is reduced.
449 A suffix of 'M' or 'G' can be given to indicate Megabytes or
450 Gigabytes respectively.
453 restores the apparent size of the array to be whatever the real
454 amount of available space is.
457 .BR \-c ", " \-\-chunk=
458 Specify chunk size of kibibytes. The default when creating an
459 array is 512KB. To ensure compatibility with earlier versions, the
460 default when Building and array with no persistent metadata is 64KB.
461 This is only meaningful for RAID0, RAID4, RAID5, RAID6, and RAID10.
463 A suffix of 'M' or 'G' can be given to indicate Megabytes or
464 Gigabytes respectively.
468 Specify rounding factor for a Linear array. The size of each
469 component will be rounded down to a multiple of this size.
470 This is a synonym for
472 but highlights the different meaning for Linear as compared to other
473 RAID levels. The default is 64K if a kernel earlier than 2.6.16 is in
474 use, and is 0K (i.e. no rounding) in later kernels.
477 .BR \-l ", " \-\-level=
478 Set RAID level. When used with
480 options are: linear, raid0, 0, stripe, raid1, 1, mirror, raid4, 4,
481 raid5, 5, raid6, 6, raid10, 10, multipath, mp, faulty, container.
482 Obviously some of these are synonymous.
486 metadata type is requested, only the
488 level is permitted, and it does not need to be explicitly given.
492 only linear, stripe, raid0, 0, raid1, multipath, mp, and faulty are valid.
496 to change the RAID level in some cases. See LEVEL CHANGES below.
499 .BR \-p ", " \-\-layout=
500 This option configures the fine details of data layout for RAID5, RAID6,
501 and RAID10 arrays, and controls the failure modes for
504 The layout of the RAID5 parity block can be one of
505 .BR left\-asymmetric ,
506 .BR left\-symmetric ,
507 .BR right\-asymmetric ,
508 .BR right\-symmetric ,
509 .BR la ", " ra ", " ls ", " rs .
511 .BR left\-symmetric .
513 It is also possible to cause RAID5 to use a RAID4-like layout by
519 Finally for RAID5 there are DDF\-compatible layouts,
520 .BR ddf\-zero\-restart ,
521 .BR ddf\-N\-restart ,
523 .BR ddf\-N\-continue .
525 These same layouts are available for RAID6. There are also 4 layouts
526 that will provide an intermediate stage for converting between RAID5
527 and RAID6. These provide a layout which is identical to the
528 corresponding RAID5 layout on the first N\-1 devices, and has the 'Q'
529 syndrome (the second 'parity' block used by RAID6) on the last device.
531 .BR left\-symmetric\-6 ,
532 .BR right\-symmetric\-6 ,
533 .BR left\-asymmetric\-6 ,
534 .BR right\-asymmetric\-6 ,
536 .BR parity\-first\-6 .
538 When setting the failure mode for level
541 .BR write\-transient ", " wt ,
542 .BR read\-transient ", " rt ,
543 .BR write\-persistent ", " wp ,
544 .BR read\-persistent ", " rp ,
546 .BR read\-fixable ", " rf ,
547 .BR clear ", " flush ", " none .
549 Each failure mode can be followed by a number, which is used as a period
550 between fault generation. Without a number, the fault is generated
551 once on the first relevant request. With a number, the fault will be
552 generated after that many requests, and will continue to be generated
553 every time the period elapses.
555 Multiple failure modes can be current simultaneously by using the
557 option to set subsequent failure modes.
559 "clear" or "none" will remove any pending or periodic failure modes,
560 and "flush" will clear any persistent faults.
562 Finally, the layout options for RAID10 are one of 'n', 'o' or 'f' followed
563 by a small number. The default is 'n2'. The supported options are:
566 signals 'near' copies. Multiple copies of one data block are at
567 similar offsets in different devices.
570 signals 'offset' copies. Rather than the chunks being duplicated
571 within a stripe, whole stripes are duplicated but are rotated by one
572 device so duplicate blocks are on different devices. Thus subsequent
573 copies of a block are in the next drive, and are one chunk further
578 (multiple copies have very different offsets).
579 See md(4) for more detail about 'near', 'offset', and 'far'.
581 The number is the number of copies of each datablock. 2 is normal, 3
582 can be useful. This number can be at most equal to the number of
583 devices in the array. It does not need to divide evenly into that
584 number (e.g. it is perfectly legal to have an 'n2' layout for an array
585 with an odd number of devices).
587 When an array is converted between RAID5 and RAID6 an intermediate
588 RAID6 layout is used in which the second parity block (Q) is always on
589 the last device. To convert a RAID5 to RAID6 and leave it in this new
590 layout (which does not require re-striping) use
591 .BR \-\-layout=preserve .
592 This will try to avoid any restriping.
594 The converse of this is
595 .B \-\-layout=normalise
596 which will change a non-standard RAID6 layout into a more standard
603 (thus explaining the p of
607 .BR \-b ", " \-\-bitmap=
608 Specify a file to store a write-intent bitmap in. The file should not
611 is also given. The same file should be provided
612 when assembling the array. If the word
614 is given, then the bitmap is stored with the metadata on the array,
615 and so is replicated on all devices. If the word
619 mode, then any bitmap that is present is removed.
621 To help catch typing errors, the filename must contain at least one
622 slash ('/') if it is a real file (not 'internal' or 'none').
624 Note: external bitmaps are only known to work on ext2 and ext3.
625 Storing bitmap files on other filesystems may result in serious problems.
628 .BR \-\-bitmap\-chunk=
629 Set the chunksize of the bitmap. Each bit corresponds to that many
630 Kilobytes of storage.
631 When using a file based bitmap, the default is to use the smallest
632 size that is at-least 4 and requires no more than 2^21 chunks.
635 bitmap, the chunksize defaults to 64Meg, or larger if necessary to
636 fit the bitmap into the available space.
638 A suffix of 'M' or 'G' can be given to indicate Megabytes or
639 Gigabytes respectively.
642 .BR \-W ", " \-\-write\-mostly
643 subsequent devices listed in a
648 command will be flagged as 'write-mostly'. This is valid for RAID1
649 only and means that the 'md' driver will avoid reading from these
650 devices if at all possible. This can be useful if mirroring over a
654 .BR \-\-write\-behind=
655 Specify that write-behind mode should be enabled (valid for RAID1
656 only). If an argument is specified, it will set the maximum number
657 of outstanding writes allowed. The default value is 256.
658 A write-intent bitmap is required in order to use write-behind
659 mode, and write-behind is only attempted on drives marked as
663 .BR \-\-assume\-clean
666 that the array pre-existed and is known to be clean. It can be useful
667 when trying to recover from a major failure as you can be sure that no
668 data will be affected unless you actually write to the array. It can
669 also be used when creating a RAID1 or RAID10 if you want to avoid the
670 initial resync, however this practice \(em while normally safe \(em is not
671 recommended. Use this only if you really know what you are doing.
673 When the devices that will be part of a new array were filled
674 with zeros before creation the operator knows the array is
675 actually clean. If that is the case, such as after running
676 badblocks, this argument can be used to tell mdadm the
677 facts the operator knows.
680 .BR \-\-backup\-file=
683 is used to increase the number of raid-devices in a RAID5 or RAID6 if
684 there are no spare devices available, or to shrink, change RAID level
685 or layout. See the GROW MODE section below on RAID\-DEVICES CHANGES.
686 The file must be stored on a separate device, not on the RAID array
690 .BR \-N ", " \-\-name=
693 for the array. This is currently only effective when creating an
694 array with a version-1 superblock, or an array in a DDF container.
695 The name is a simple textual string that can be used to identify array
696 components when assembling. If name is needed but not specified, it
697 is taken from the basename of the device that is being created.
709 run the array, even if some of the components
710 appear to be active in another array or filesystem. Normally
712 will ask for confirmation before including such components in an
713 array. This option causes that question to be suppressed.
716 .BR \-f ", " \-\-force
719 accept the geometry and layout specified without question. Normally
721 will not allow creation of an array with only one device, and will try
722 to create a RAID5 array with one missing drive (as this makes the
723 initial resync work faster). With
726 will not try to be so clever.
729 .BR \-a ", " "\-\-auto{=yes,md,mdp,part,p}{NN}"
730 Instruct mdadm how to create the device file if needed, possibly allocating
731 an unused minor number. "md" causes a non-partitionable array
732 to be used (though since Linux 2.6.28, these array devices are in fact
733 partitionable). "mdp", "part" or "p" causes a partitionable array (2.6 and
734 later) to be used. "yes" requires the named md device to have
735 a 'standard' format, and the type and minor number will be determined
736 from this. With mdadm 3.0, device creation is normally left up to
738 so this option is unlikely to be needed.
739 See DEVICE NAMES below.
741 The argument can also come immediately after
746 is not given on the command line or in the config file, then
752 is also given, then any
754 entries in the config file will override the
756 instruction given on the command line.
758 For partitionable arrays,
760 will create the device file for the whole array and for the first 4
761 partitions. A different number of partitions can be specified at the
762 end of this option (e.g.
764 If the device name ends with a digit, the partition names add a 'p',
766 .IR /dev/md/home1p3 .
767 If there is no trailing digit, then the partition names just have a
769 .IR /dev/md/scratch3 .
771 If the md device name is in a 'standard' format as described in DEVICE
772 NAMES, then it will be created, if necessary, with the appropriate
773 device number based on that name. If the device name is not in one of these
774 formats, then a unused device number will be allocated. The device
775 number will be considered unused if there is no active array for that
776 number, and there is no entry in /dev for that number and with a
777 non-standard name. Names that are not in 'standard' format are only
778 allowed in "/dev/md/".
782 .\".BR \-\-symlink = no
787 .\"to create devices in
789 .\"it will also create symlinks from
791 .\"with names starting with
797 .\"to suppress this, or
798 .\".B \-\-symlink=yes
799 .\"to enforce this even if it is suppressing
807 .BR \-u ", " \-\-uuid=
808 uuid of array to assemble. Devices which don't have this uuid are
812 .BR \-m ", " \-\-super\-minor=
813 Minor number of device that array was created for. Devices which
814 don't have this minor number are excluded. If you create an array as
815 /dev/md1, then all superblocks will contain the minor number 1, even if
816 the array is later assembled as /dev/md2.
818 Giving the literal word "dev" for
822 to use the minor number of the md device that is being assembled.
825 .B \-\-super\-minor=dev
826 will look for super blocks with a minor number of 0.
829 is only relevant for v0.90 metadata, and should not normally be used.
835 .BR \-N ", " \-\-name=
836 Specify the name of the array to assemble. This must be the name
837 that was specified when creating the array. It must either match
838 the name stored in the superblock exactly, or it must match
841 prefixed to the start of the given name.
844 .BR \-f ", " \-\-force
845 Assemble the array even if the metadata on some devices appears to be
848 cannot find enough working devices to start the array, but can find
849 some devices that are recorded as having failed, then it will mark
850 those devices as working so that the array can be started.
851 An array which requires
853 to be started may contain data corruption. Use it carefully.
857 Attempt to start the array even if fewer drives were given than were
858 present last time the array was active. Normally if not all the
859 expected drives are found and
861 is not used, then the array will be assembled but not started.
864 an attempt will be made to start it anyway.
868 This is the reverse of
870 in that it inhibits the startup of array unless all expected drives
871 are present. This is only needed with
873 and can be used if the physical connections to devices are
874 not as reliable as you would like.
877 .BR \-a ", " "\-\-auto{=no,yes,md,mdp,part}"
878 See this option under Create and Build options.
881 .BR \-b ", " \-\-bitmap=
882 Specify the bitmap file that was given when the array was created. If
885 bitmap, there is no need to specify this when assembling the array.
888 .BR \-\-backup\-file=
891 was used while reshaping an array (e.g. changing number of devices or
892 chunk size) and the system crashed during the critical section, then the same
896 to allow possibly corrupted data to be restored, and the reshape
900 .BR \-\-invalid\-backup
901 If the file needed for the above option is not available for any
902 reason an empty file can be given together with this option to
903 indicate that the backup file is invalid. In this case the data that
904 was being rearranged at the time of the crash could be irrecoverably
905 lost, but the rest of the array may still be recoverable. This option
906 should only be used as a last resort if there is no way to recover the
911 .BR \-U ", " \-\-update=
912 Update the superblock on each device while assembling the array. The
913 argument given to this flag can be one of
928 option will adjust the superblock of an array what was created on a Sparc
929 machine running a patched 2.2 Linux kernel. This kernel got the
930 alignment of part of the superblock wrong. You can use the
931 .B "\-\-examine \-\-sparc2.2"
934 to see what effect this would have.
938 option will update the
940 field on each superblock to match the minor number of the array being
942 This can be useful if
944 reports a different "Preferred Minor" to
946 In some cases this update will be performed automatically
947 by the kernel driver. In particular the update happens automatically
948 at the first write to an array with redundancy (RAID level 1 or
949 greater) on a 2.6 (or later) kernel.
953 option will change the uuid of the array. If a UUID is given with the
955 option that UUID will be used as a new UUID and will
957 be used to help identify the devices in the array.
960 is given, a random UUID is chosen.
964 option will change the
966 of the array as stored in the superblock. This is only supported for
967 version-1 superblocks.
971 option will change the
973 as recorded in the superblock. For version-0 superblocks, this is the
974 same as updating the UUID.
975 For version-1 superblocks, this involves updating the name.
979 option will cause the array to be marked
981 meaning that any redundancy in the array (e.g. parity for RAID5,
982 copies for RAID1) may be incorrect. This will cause the RAID system
983 to perform a "resync" pass to make sure that all redundant information
988 option allows arrays to be moved between machines with different
990 When assembling such an array for the first time after a move, giving
991 .B "\-\-update=byteorder"
994 to expect superblocks to have their byteorder reversed, and will
995 correct that order before assembling the array. This is only valid
996 with original (Version 0.90) superblocks.
1000 option will correct the summaries in the superblock. That is the
1001 counts of total, working, active, failed, and spare devices.
1005 option will rarely be of use. It applies to version 1.1 and 1.2 metadata
1006 only (where the metadata is at the start of the device) and is only
1007 useful when the component device has changed size (typically become
1008 larger). The version 1 metadata records the amount of the device that
1009 can be used to store data, so if a device in a version 1.1 or 1.2
1010 array becomes larger, the metadata will still be visible, but the
1011 extra space will not. In this case it might be useful to assemble the
1013 .BR \-\-update=devicesize .
1016 to determine the maximum usable amount of space on each device and
1017 update the relevant field in the metadata.
1021 option can be used when an array has an internal bitmap which is
1022 corrupt in some way so that assembling the array normally fails. It
1023 will cause any internal bitmap to be ignored.
1027 .B \-\-auto\-update\-homehost
1028 This flag is only meaningful with auto-assembly (see discussion below).
1029 In that situation, if no suitable arrays are found for this homehost,
1031 will rescan for any arrays at all and will assemble them and update the
1032 homehost to match the current host.
1035 .SH For Manage mode:
1038 .BR \-t ", " \-\-test
1039 Unless a more serious error occurred,
1041 will exit with a status of 2 if no changes were made to the array and
1042 0 if at least one change was made.
1043 This can be useful when an indirect specifier such as
1048 is used in requesting an operation on the array.
1050 will report failure if these specifiers didn't find any match.
1053 .BR \-a ", " \-\-add
1054 hot-add listed devices.
1055 If a device appears to have recently been part of the array
1056 (possibly it failed or was removed) the device is re\-added as describe
1058 If that fails or the device was never part of the array, the device is
1059 added as a hot-spare.
1060 If the array is degraded, it will immediately start to rebuild data
1063 Note that this and the following options are only meaningful on array
1064 with redundancy. They don't apply to RAID0 or Linear.
1068 re\-add a device that was previous removed from an array.
1069 If the metadata on the device reports that it is a member of the
1070 array, and the slot that it used is still vacant, then the device will
1071 be added back to the array in the same position. This will normally
1072 cause the data for that device to be recovered. However based on the
1073 event count on the device, the recovery may only require sections that
1074 are flagged a write-intent bitmap to be recovered or may not require
1075 any recovery at all.
1077 When used on an array that has no metadata (i.e. it was built with
1079 it will be assumed that bitmap-based recovery is enough to make the
1080 device fully consistent with the array.
1084 can be accompanied by
1085 .BR \-\-update=devicesize .
1086 See the description of this option when used in Assemble mode for an
1087 explanation of its use.
1089 If the device name given is
1091 then mdadm will try to find any device that looks like it should be
1092 part of the array but isn't and will try to re\-add all such devices.
1095 .BR \-r ", " \-\-remove
1096 remove listed devices. They must not be active. i.e. they should
1097 be failed or spare devices. As well as the name of a device file
1106 The first causes all failed device to be removed. The second causes
1107 any device which is no longer connected to the system (i.e an 'open'
1110 to be removed. This will only succeed for devices that are spares or
1111 have already been marked as failed.
1114 .BR \-f ", " \-\-fail
1115 mark listed devices as faulty.
1116 As well as the name of a device file, the word
1118 can be given. This will cause any device that has been detached from
1119 the system to be marked as failed. It can then be removed.
1127 .BR \-\-write\-mostly
1128 Subsequent devices that are added or re\-added will have the 'write-mostly'
1129 flag set. This is only valid for RAID1 and means that the 'md' driver
1130 will avoid reading from these devices if possible.
1133 Subsequent devices that are added or re\-added will have the 'write-mostly'
1137 Each of these options requires that the first device listed is the array
1138 to be acted upon, and the remainder are component devices to be added,
1139 removed, marked as faulty, etc. Several different operations can be
1140 specified for different devices, e.g.
1142 mdadm /dev/md0 \-\-add /dev/sda1 \-\-fail /dev/sdb1 \-\-remove /dev/sdb1
1144 Each operation applies to all devices listed until the next
1147 If an array is using a write-intent bitmap, then devices which have
1148 been removed can be re\-added in a way that avoids a full
1149 reconstruction but instead just updates the blocks that have changed
1150 since the device was removed. For arrays with persistent metadata
1151 (superblocks) this is done automatically. For arrays created with
1153 mdadm needs to be told that this device we removed recently with
1156 Devices can only be removed from an array if they are not in active
1157 use, i.e. that must be spares or failed devices. To remove an active
1158 device, it must first be marked as
1164 .BR \-Q ", " \-\-query
1165 Examine a device to see
1166 (1) if it is an md device and (2) if it is a component of an md
1168 Information about what is discovered is presented.
1171 .BR \-D ", " \-\-detail
1172 Print details of one or more md devices.
1175 .BR \-\-detail\-platform
1176 Print details of the platform's RAID capabilities (firmware / hardware
1177 topology) for a given metadata format.
1180 .BR \-Y ", " \-\-export
1185 output will be formatted as
1187 pairs for easy import into the environment.
1190 .BR \-E ", " \-\-examine
1191 Print contents of the metadata stored on the named device(s).
1192 Note the contrast between
1197 applies to devices which are components of an array, while
1199 applies to a whole array which is currently active.
1202 If an array was created on a SPARC machine with a 2.2 Linux kernel
1203 patched with RAID support, the superblock will have been created
1204 incorrectly, or at least incompatibly with 2.4 and later kernels.
1209 will fix the superblock before displaying it. If this appears to do
1210 the right thing, then the array can be successfully assembled using
1211 .BR "\-\-assemble \-\-update=sparc2.2" .
1214 .BR \-X ", " \-\-examine\-bitmap
1215 Report information about a bitmap file.
1216 The argument is either an external bitmap file or an array component
1217 in case of an internal bitmap. Note that running this on an array
1220 does not report the bitmap for that array.
1223 .BR \-R ", " \-\-run
1224 start a partially assembled array. If
1226 did not find enough devices to fully start the array, it might leaving
1227 it partially assembled. If you wish, you can then use
1229 to start the array in degraded mode.
1232 .BR \-S ", " \-\-stop
1233 deactivate array, releasing all resources.
1236 .BR \-o ", " \-\-readonly
1237 mark array as readonly.
1240 .BR \-w ", " \-\-readwrite
1241 mark array as readwrite.
1244 .B \-\-zero\-superblock
1245 If the device contains a valid md superblock, the block is
1246 overwritten with zeros. With
1248 the block where the superblock would be is overwritten even if it
1249 doesn't appear to be valid.
1252 .B \-\-kill\-subarray=
1253 If the device is a container and the argument to \-\-kill\-subarray
1254 specifies an inactive subarray in the container, then the subarray is
1255 deleted. Deleting all subarrays will leave an 'empty-container' or
1256 spare superblock on the drives. See \-\-zero\-superblock for completely
1257 removing a superblock. Note that some formats depend on the subarray
1258 index for generating a UUID, this command will fail if it would change
1259 the UUID of an active subarray.
1262 .B \-\-update\-subarray=
1263 If the device is a container and the argument to \-\-update\-subarray
1264 specifies a subarray in the container, then attempt to update the given
1265 superblock field in the subarray. See below in
1270 .BR \-t ", " \-\-test
1275 is set to reflect the status of the device. See below in
1280 .BR \-W ", " \-\-wait
1281 For each md device given, wait for any resync, recovery, or reshape
1282 activity to finish before returning.
1284 will return with success if it actually waited for every device
1285 listed, otherwise it will return failure.
1289 For each md device given, or each device in /proc/mdstat if
1291 is given, arrange for the array to be marked clean as soon as possible.
1293 will return with success if the array uses external metadata and we
1294 successfully waited. For native arrays this returns immediately as the
1295 kernel handles dirty-clean transitions at shutdown. No action is taken
1296 if safe-mode handling is disabled.
1298 .SH For Incremental Assembly mode:
1300 .BR \-\-rebuild\-map ", " \-r
1301 Rebuild the map file
1302 .RB ( /var/run/mdadm/map )
1305 uses to help track which arrays are currently being assembled.
1308 .BR \-\-run ", " \-R
1309 Run any array assembled as soon as a minimal number of devices are
1310 available, rather than waiting until all expected devices are present.
1313 .BR \-\-scan ", " \-s
1314 Only meaningful with
1318 file for arrays that are being incrementally assembled and will try to
1319 start any that are not already started. If any such array is listed
1322 as requiring an external bitmap, that bitmap will be attached first.
1325 .BR \-\-fail ", " \-f
1326 This allows the hot-plug system to remove devices that have fully disappeared
1327 from the kernel. It will first fail and then remove the device from any
1328 array it belongs to.
1329 The device name given should be a kernel device name such as "sda",
1335 Only used with \-\-fail. The 'path' given will be recorded so that if
1336 a new device appears at the same location it can be automatically
1337 added to the same array. This allows the failed device to be
1338 automatically replaced by a new device without metadata if it appears
1339 at specified path. This option is normally only set by a
1343 .SH For Monitor mode:
1345 .BR \-m ", " \-\-mail
1346 Give a mail address to send alerts to.
1349 .BR \-p ", " \-\-program ", " \-\-alert
1350 Give a program to be run whenever an event is detected.
1353 .BR \-y ", " \-\-syslog
1354 Cause all events to be reported through 'syslog'. The messages have
1355 facility of 'daemon' and varying priorities.
1358 .BR \-d ", " \-\-delay
1359 Give a delay in seconds.
1361 polls the md arrays and then waits this many seconds before polling
1362 again. The default is 60 seconds. Since 2.6.16, there is no need to
1363 reduce this as the kernel alerts
1365 immediately when there is any change.
1368 .BR \-r ", " \-\-increment
1369 Give a percentage increment.
1371 will generate RebuildNN events with the given percentage increment.
1374 .BR \-f ", " \-\-daemonise
1377 to run as a background daemon if it decides to monitor anything. This
1378 causes it to fork and run in the child, and to disconnect from the
1379 terminal. The process id of the child is written to stdout.
1382 which will only continue monitoring if a mail address or alert program
1383 is found in the config file.
1386 .BR \-i ", " \-\-pid\-file
1389 is running in daemon mode, write the pid of the daemon process to
1390 the specified file, instead of printing it on standard output.
1393 .BR \-1 ", " \-\-oneshot
1394 Check arrays only once. This will generate
1396 events and more significantly
1402 .B " mdadm \-\-monitor \-\-scan \-1"
1404 from a cron script will ensure regular notification of any degraded arrays.
1407 .BR \-t ", " \-\-test
1410 alert for every array found at startup. This alert gets mailed and
1411 passed to the alert program. This can be used for testing that alert
1412 message do get through successfully.
1416 This inhibits the functionality for moving spares between arrays.
1417 Only one monitoring process started with
1419 but without this flag is allowed, otherwise the two could interfere
1426 .B mdadm \-\-assemble
1427 .I md-device options-and-component-devices...
1430 .B mdadm \-\-assemble \-\-scan
1431 .I md-devices-and-options...
1434 .B mdadm \-\-assemble \-\-scan
1438 This usage assembles one or more RAID arrays from pre-existing components.
1439 For each array, mdadm needs to know the md device, the identity of the
1440 array, and a number of component-devices. These can be found in a number of ways.
1442 In the first usage example (without the
1444 the first device given is the md device.
1445 In the second usage example, all devices listed are treated as md
1446 devices and assembly is attempted.
1447 In the third (where no devices are listed) all md devices that are
1448 listed in the configuration file are assembled. If not arrays are
1449 described by the configuration file, then any arrays that
1450 can be found on unused devices will be assembled.
1452 If precisely one device is listed, but
1458 was given and identity information is extracted from the configuration file.
1460 The identity can be given with the
1466 option, will be taken from the md-device record in the config file, or
1467 will be taken from the super block of the first component-device
1468 listed on the command line.
1470 Devices can be given on the
1472 command line or in the config file. Only devices which have an md
1473 superblock which contains the right identity will be considered for
1476 The config file is only used if explicitly named with
1478 or requested with (a possibly implicit)
1483 .B /etc/mdadm/mdadm.conf
1488 is not given, then the config file will only be used to find the
1489 identity of md arrays.
1491 Normally the array will be started after it is assembled. However if
1493 is not given and not all expected drives were listed, then the array
1494 is not started (to guard against usage errors). To insist that the
1495 array be started in this case (as may work for RAID1, 4, 5, 6, or 10),
1504 does not create any entries in
1508 It does record information in
1509 .B /var/run/mdadm/map
1512 to choose the correct name.
1516 detects that udev is not configured, it will create the devices in
1520 In Linux kernels prior to version 2.6.28 there were two distinctly
1521 different types of md devices that could be created: one that could be
1522 partitioned using standard partitioning tools and one that could not.
1523 Since 2.6.28 that distinction is no longer relevant as both type of
1524 devices can be partitioned.
1526 will normally create the type that originally could not be partitioned
1527 as it has a well defined major number (9).
1529 Prior to 2.6.28, it is important that mdadm chooses the correct type
1530 of array device to use. This can be controlled with the
1532 option. In particular, a value of "mdp" or "part" or "p" tells mdadm
1533 to use a partitionable device rather than the default.
1535 In the no-udev case, the value given to
1537 can be suffixed by a number. This tells
1539 to create that number of partition devices rather than the default of 4.
1543 can also be given in the configuration file as a word starting
1545 on the ARRAY line for the relevant array.
1552 and no devices are listed,
1554 will first attempt to assemble all the arrays listed in the config
1557 In no array at listed in the config (other than those marked
1559 it will look through the available devices for possible arrays and
1560 will try to assemble anything that it finds. Arrays which are tagged
1561 as belonging to the given homehost will be assembled and started
1562 normally. Arrays which do not obviously belong to this host are given
1563 names that are expected not to conflict with anything local, and are
1564 started "read-auto" so that nothing is written to any device until the
1565 array is written to. i.e. automatic resync etc is delayed.
1569 finds a consistent set of devices that look like they should comprise
1570 an array, and if the superblock is tagged as belonging to the given
1571 home host, it will automatically choose a device name and try to
1572 assemble the array. If the array uses version-0.90 metadata, then the
1574 number as recorded in the superblock is used to create a name in
1578 If the array uses version-1 metadata, then the
1580 from the superblock is used to similarly create a name in
1582 (the name will have any 'host' prefix stripped first).
1584 This behaviour can be modified by the
1588 configuration file. This line can indicate that specific metadata
1589 type should, or should not, be automatically assembled. If an array
1590 is found which is not listed in
1592 and has a metadata format that is denied by the
1594 line, then it will not be assembled.
1597 line can also request that all arrays identified as being for this
1598 homehost should be assembled regardless of their metadata type.
1601 for further details.
1606 cannot find any array for the given host at all, and if
1607 .B \-\-auto\-update\-homehost
1610 will search again for any array (not just an array created for this
1611 host) and will assemble each assuming
1612 .BR \-\-update=homehost .
1613 This will change the host tag in the superblock so that on the next run,
1614 these arrays will be found without the second pass. The intention of
1615 this feature is to support transitioning a set of md arrays to using
1618 The reason for requiring arrays to be tagged with the homehost for
1619 auto assembly is to guard against problems that can arise when moving
1620 devices from one host to another.
1631 .BI \-\-raid\-devices= Z
1635 This usage is similar to
1637 The difference is that it creates an array without a superblock. With
1638 these arrays there is no difference between initially creating the array and
1639 subsequently assembling the array, except that hopefully there is useful
1640 data there in the second case.
1642 The level may raid0, linear, raid1, raid10, multipath, or faulty, or
1643 one of their synonyms. All devices must be listed and the array will
1644 be started once complete. It will often be appropriate to use
1645 .B \-\-assume\-clean
1646 with levels raid1 or raid10.
1657 .BI \-\-raid\-devices= Z
1661 This usage will initialise a new md array, associate some devices with
1662 it, and activate the array.
1664 The named device will normally not exist when
1665 .I "mdadm \-\-create"
1666 is run, but will be created by
1668 once the array becomes active.
1670 As devices are added, they are checked to see if they contain RAID
1671 superblocks or filesystems. They are also checked to see if the variance in
1672 device size exceeds 1%.
1674 If any discrepancy is found, the array will not automatically be run, though
1677 can override this caution.
1679 To create a "degraded" array in which some devices are missing, simply
1680 give the word "\fBmissing\fP"
1681 in place of a device name. This will cause
1683 to leave the corresponding slot in the array empty.
1684 For a RAID4 or RAID5 array at most one slot can be
1685 "\fBmissing\fP"; for a RAID6 array at most two slots.
1686 For a RAID1 array, only one real device needs to be given. All of the
1690 When creating a RAID5 array,
1692 will automatically create a degraded array with an extra spare drive.
1693 This is because building the spare into a degraded array is in general
1694 faster than resyncing the parity on a non-degraded, but not clean,
1695 array. This feature can be overridden with the
1699 When creating an array with version-1 metadata a name for the array is
1701 If this is not given with the
1705 will choose a name based on the last component of the name of the
1706 device being created. So if
1708 is being created, then the name
1713 is being created, then the name
1717 When creating a partition based array, using
1719 with version-1.x metadata, the partition type should be set to
1721 (non fs-data). This type selection allows for greater precision since
1722 using any other [RAID auto-detect (0xFD) or a GNU/Linux partition (0x83)],
1723 might create problems in the event of array recovery through a live cdrom.
1725 A new array will normally get a randomly assigned 128bit UUID which is
1726 very likely to be unique. If you have a specific need, you can choose
1727 a UUID for the array by giving the
1729 option. Be warned that creating two arrays with the same UUID is a
1730 recipe for disaster. Also, using
1732 when creating a v0.90 array will silently override any
1737 .\"option is given, it is not necessary to list any component-devices in this command.
1738 .\"They can be added later, before a
1742 .\"is given, the apparent size of the smallest drive given is used.
1744 When creating an array within a
1747 can be given either the list of devices to use, or simply the name of
1748 the container. The former case gives control over which devices in
1749 the container will be used for the array. The latter case allows
1751 to automatically choose which devices to use based on how much spare
1754 The General Management options that are valid with
1759 insist on running the array even if some devices look like they might
1764 start the array readonly \(em not supported yet.
1771 .I options... devices...
1774 This usage will allow individual devices in an array to be failed,
1775 removed or added. It is possible to perform multiple operations with
1776 on command. For example:
1778 .B " mdadm /dev/md0 \-f /dev/hda1 \-r /dev/hda1 \-a /dev/hda1"
1784 and will then remove it from the array and finally add it back
1785 in as a spare. However only one md array can be affected by a single
1788 When a device is added to an active array, mdadm checks to see if it
1789 has metadata on it which suggests that it was recently a member of the
1790 array. If it does, it tries to "re\-add" the device. If there have
1791 been no changes since the device was removed, or if the array has a
1792 write-intent bitmap which has recorded whatever changes there were,
1793 then the device will immediately become a full member of the array and
1794 those differences recorded in the bitmap will be resolved.
1804 MISC mode includes a number of distinct operations that
1805 operate on distinct devices. The operations are:
1808 The device is examined to see if it is
1809 (1) an active md array, or
1810 (2) a component of an md array.
1811 The information discovered is reported.
1815 The device should be an active md device.
1817 will display a detailed description of the array.
1821 will cause the output to be less detailed and the format to be
1822 suitable for inclusion in
1826 will normally be 0 unless
1828 failed to get useful information about the device(s); however, if the
1830 option is given, then the exit status will be:
1834 The array is functioning normally.
1837 The array has at least one failed device.
1840 The array has multiple failed devices such that it is unusable.
1843 There was an error while trying to get information about the device.
1847 .B \-\-detail\-platform
1848 Print detail of the platform's RAID capabilities (firmware / hardware
1849 topology). If the metadata is specified with
1853 then the return status will be:
1857 metadata successfully enumerated its platform components on this system
1860 metadata is platform independent
1863 metadata failed to find its platform components on this system
1867 .B \-\-update\-subarray=
1868 If the device is a container and the argument to \-\-update\-subarray
1869 specifies a subarray in the container, then attempt to update the given
1870 superblock field in the subarray. Similar to updating an array in
1871 "assemble" mode, the field to update is selected by
1875 option. Currently only
1881 option updates the subarray name in the metadata, it may not affect the
1882 device node name or the device node symlink until the subarray is
1883 re\-assembled. If updating
1885 would change the UUID of an active subarray this operation is blocked,
1886 and the command will end in an error.
1890 The device should be a component of an md array.
1892 will read the md superblock of the device and display the contents.
1897 is given, then multiple devices that are components of the one array
1898 are grouped together and reported in a single entry suitable
1904 without listing any devices will cause all devices listed in the
1905 config file to be examined.
1909 The devices should be active md arrays which will be deactivated, as
1910 long as they are not currently in use.
1914 This will fully activate a partially assembled md array.
1918 This will mark an active array as read-only, providing that it is
1919 not currently being used.
1925 array back to being read/write.
1929 For all operations except
1932 will cause the operation to be applied to all arrays listed in
1937 causes all devices listed in the config file to be examined.
1940 .BR \-b ", " \-\-brief
1941 Be less verbose. This is used with
1949 gives an intermediate level of verbosity.
1955 .B mdadm \-\-monitor
1956 .I options... devices...
1961 to periodically poll a number of md arrays and to report on any events
1964 will never exit once it decides that there are arrays to be checked,
1965 so it should normally be run in the background.
1967 As well as reporting events,
1969 may move a spare drive from one array to another if they are in the
1974 and if the destination array has a failed drive but no spares.
1976 If any devices are listed on the command line,
1978 will only monitor those devices. Otherwise all arrays listed in the
1979 configuration file will be monitored. Further, if
1981 is given, then any other md devices that appear in
1983 will also be monitored.
1985 The result of monitoring the arrays is the generation of events.
1986 These events are passed to a separate program (if specified) and may
1987 be mailed to a given E-mail address.
1989 When passing events to a program, the program is run once for each event,
1990 and is given 2 or 3 command-line arguments: the first is the
1991 name of the event (see below), the second is the name of the
1992 md device which is affected, and the third is the name of a related
1993 device if relevant (such as a component device that has failed).
1997 is given, then a program or an E-mail address must be specified on the
1998 command line or in the config file. If neither are available, then
2000 will not monitor anything.
2004 will continue monitoring as long as something was found to monitor. If
2005 no program or email is given, then each event is reported to
2008 The different events are:
2012 .B DeviceDisappeared
2013 An md array which previously was configured appears to no longer be
2014 configured. (syslog priority: Critical)
2018 was told to monitor an array which is RAID0 or Linear, then it will
2020 .B DeviceDisappeared
2021 with the extra information
2023 This is because RAID0 and Linear do not support the device-failed,
2024 hot-spare and resync operations which are monitored.
2028 An md array started reconstruction. (syslog priority: Warning)
2034 is a two-digit number (ie. 05, 48). This indicates that rebuild
2035 has passed that many percent of the total. The events are generated
2036 with fixed increment since 0. Increment size may be specified with
2037 a commandline option (default is 20). (syslog priority: Warning)
2041 An md array that was rebuilding, isn't any more, either because it
2042 finished normally or was aborted. (syslog priority: Warning)
2046 An active component device of an array has been marked as
2047 faulty. (syslog priority: Critical)
2051 A spare component device which was being rebuilt to replace a faulty
2052 device has failed. (syslog priority: Critical)
2056 A spare component device which was being rebuilt to replace a faulty
2057 device has been successfully rebuilt and has been made active.
2058 (syslog priority: Info)
2062 A new md array has been detected in the
2064 file. (syslog priority: Info)
2068 A newly noticed array appears to be degraded. This message is not
2071 notices a drive failure which causes degradation, but only when
2073 notices that an array is degraded when it first sees the array.
2074 (syslog priority: Critical)
2078 A spare drive has been moved from one array in a
2082 to another to allow a failed drive to be replaced.
2083 (syslog priority: Info)
2089 has been told, via the config file, that an array should have a certain
2090 number of spare devices, and
2092 detects that it has fewer than this number when it first sees the
2093 array, it will report a
2096 (syslog priority: Warning)
2100 An array was found at startup, and the
2103 (syslog priority: Info)
2113 cause Email to be sent. All events cause the program to be run.
2114 The program is run with two or three arguments: the event
2115 name, the array device and possibly a second device.
2117 Each event has an associated array device (e.g.
2119 and possibly a second device. For
2124 the second device is the relevant component device.
2127 the second device is the array that the spare was moved from.
2131 to move spares from one array to another, the different arrays need to
2132 be labeled with the same
2134 or the spares must be allowed to migrate through matching POLICY domains
2135 in the configuration file. The
2137 name can be any string; it is only necessary that different spare
2138 groups use different names.
2142 detects that an array in a spare group has fewer active
2143 devices than necessary for the complete array, and has no spare
2144 devices, it will look for another array in the same spare group that
2145 has a full complement of working drive and a spare. It will then
2146 attempt to remove the spare from the second drive and add it to the
2148 If the removal succeeds but the adding fails, then it is added back to
2151 If the spare group for a degraded array is not defined,
2153 will look at the rules of spare migration specified by POLICY lines in
2155 and then follow similar steps as above if a matching spare is found.
2158 The GROW mode is used for changing the size or shape of an active
2160 For this to work, the kernel must support the necessary change.
2161 Various types of growth are being added during 2.6 development,
2162 including restructuring a RAID5 array to have more active devices.
2164 Currently the only support available is to
2166 change the "size" attribute
2167 for RAID1, RAID5 and RAID6.
2169 increase or decrease the "raid\-devices" attribute of RAID1, RAID5,
2172 change the chunk-size and layout of RAID5 and RAID6.
2174 convert between RAID1 and RAID5, and between RAID5 and RAID6.
2176 add a write-intent bitmap to any array which supports these bitmaps, or
2177 remove a write-intent bitmap from such an array.
2180 GROW mode is not currently supported for
2182 or arrays inside containers.
2185 Normally when an array is built the "size" it taken from the smallest
2186 of the drives. If all the small drives in an arrays are, one at a
2187 time, removed and replaced with larger drives, then you could have an
2188 array of large drives with only a small amount used. In this
2189 situation, changing the "size" with "GROW" mode will allow the extra
2190 space to start being used. If the size is increased in this way, a
2191 "resync" process will start to make sure the new parts of the array
2194 Note that when an array changes size, any filesystem that may be
2195 stored in the array will not automatically grow to use the space. The
2196 filesystem will need to be explicitly told to use the extra space.
2198 Also the size of an array cannot be changed while it has an active
2199 bitmap. If an array has a bitmap, it must be removed before the size
2200 can be changed. Once the change it complete a new bitmap can be created.
2202 .SS RAID\-DEVICES CHANGES
2204 A RAID1 array can work with any number of devices from 1 upwards
2205 (though 1 is not very useful). There may be times which you want to
2206 increase or decrease the number of active devices. Note that this is
2207 different to hot-add or hot-remove which changes the number of
2210 When reducing the number of devices in a RAID1 array, the slots which
2211 are to be removed from the array must already be vacant. That is, the
2212 devices which were in those slots must be failed and removed.
2214 When the number of devices is increased, any hot spares that are
2215 present will be activated immediately.
2217 Changing the number of active devices in a RAID5 or RAID6 is much more
2218 effort. Every block in the array will need to be read and written
2219 back to a new location. From 2.6.17, the Linux Kernel is able to
2220 increase the number of devices in a RAID5 safely, including restarting
2221 an interrupted "reshape". From 2.6.31, the Linux Kernel is able to
2222 increase or decrease the number of devices in a RAID5 or RAID6.
2224 When decreasing the number of devices, the size of the array will also
2225 decrease. If there was data in the array, it could get destroyed and
2226 this is not reversible. To help prevent accidents,
2228 requires that the size of the array be decreased first with
2229 .BR "mdadm --grow --array-size" .
2230 This is a reversible change which simply makes the end of the array
2231 inaccessible. The integrity of any data can then be checked before
2232 the non-reversible reduction in the number of devices is request.
2234 When relocating the first few stripes on a RAID5 or RAID6, it is not
2235 possible to keep the data on disk completely consistent and
2236 crash-proof. To provide the required safety, mdadm disables writes to
2237 the array while this "critical section" is reshaped, and takes a
2238 backup of the data that is in that section. For grows, this backup may be
2239 stored in any spare devices that the array has, however it can also be
2240 stored in a separate file specified with the
2242 option, and is required to be specified for shrinks, RAID level
2243 changes and layout changes. If this option is used, and the system
2244 does crash during the critical period, the same file must be passed to
2246 to restore the backup and reassemble the array. When shrinking rather
2247 than growing the array, the reshape is done from the end towards the
2248 beginning, so the "critical section" is at the end of the reshape.
2252 Changing the RAID level of any array happens instantaneously. However
2253 in the RAID5 to RAID6 case this requires a non-standard layout of the
2254 RAID6 data, and in the RAID6 to RAID5 case that non-standard layout is
2255 required before the change can be accomplished. So while the level
2256 change is instant, the accompanying layout change can take quite a
2259 is required. If the array is not simultaneously being grown or
2260 shrunk, so that the array size will remain the same - for example,
2261 reshaping a 3-drive RAID5 into a 4-drive RAID6 - the backup file will
2262 be used not just for a "cricital section" but throughout the reshape
2263 operation, as described below under LAYOUT CHANGES.
2265 .SS CHUNK-SIZE AND LAYOUT CHANGES
2267 Changing the chunk-size of layout without also changing the number of
2268 devices as the same time will involve re-writing all blocks in-place.
2269 To ensure against data loss in the case of a crash, a
2271 must be provided for these changes. Small sections of the array will
2272 be copied to the backup file while they are being rearranged. This
2273 means that all the data is copied twice, once to the backup and once
2274 to the new layout on the array, so this type of reshape will go very
2277 If the reshape is interrupted for any reason, this backup file must be
2279 .B "mdadm --assemble"
2280 so the array can be reassembled. Consequently the file cannot be
2281 stored on the device being reshaped.
2286 A write-intent bitmap can be added to, or removed from, an active
2287 array. Either internal bitmaps, or bitmaps stored in a separate file,
2288 can be added. Note that if you add a bitmap stored in a file which is
2289 in a filesystem that is on the RAID array being affected, the system
2290 will deadlock. The bitmap must be on a separate filesystem.
2292 .SH INCREMENTAL MODE
2296 .B mdadm \-\-incremental
2302 .B mdadm \-\-incremental \-\-fail
2306 .B mdadm \-\-incremental \-\-rebuild\-map
2309 .B mdadm \-\-incremental \-\-run \-\-scan
2312 This mode is designed to be used in conjunction with a device
2313 discovery system. As devices are found in a system, they can be
2315 .B "mdadm \-\-incremental"
2316 to be conditionally added to an appropriate array.
2318 Conversely, it can also be used with the
2320 flag to do just the opposite and find whatever array a particular device
2321 is part of and remove the device from that array.
2323 If the device passed is a
2325 device created by a previous call to
2327 then rather than trying to add that device to an array, all the arrays
2328 described by the metadata of the container will be started.
2331 performs a number of tests to determine if the device is part of an
2332 array, and which array it should be part of. If an appropriate array
2333 is found, or can be created,
2335 adds the device to the array and conditionally starts the array.
2339 will normally only add devices to an array which were previously working
2340 (active or spare) parts of that array. The support for automatic
2341 inclusion of a new drive as a spare in some array requires
2342 a configuration through POLICY in config file.
2346 makes are as follow:
2348 Is the device permitted by
2350 That is, is it listed in a
2352 line in that file. If
2354 is absent then the default it to allow any device. Similar if
2356 contains the special word
2358 then any device is allowed. Otherwise the device name given to
2360 must match one of the names or patterns in a
2365 Does the device have a valid md superblock. If a specific metadata
2366 version is request with
2370 then only that style of metadata is accepted, otherwise
2372 finds any known version of metadata. If no
2374 metadata is found, the device may be still added to an array
2375 as a spare if POLICY allows.
2379 Does the metadata match an expected array?
2380 The metadata can match in two ways. Either there is an array listed
2383 which identifies the array (either by UUID, by name, by device list,
2384 or by minor-number), or the array was created with a
2390 or on the command line.
2393 is not able to positively identify the array as belonging to the
2394 current host, the device will be rejected.
2398 keeps a list of arrays that it has partially assembled in
2399 .B /var/run/mdadm/map
2401 .B /var/run/mdadm.map
2402 if the directory doesn't exist. Or maybe even
2403 .BR /dev/.mdadm.map ).
2404 If no array exists which matches
2405 the metadata on the new device,
2407 must choose a device name and unit number. It does this based on any
2410 or any name information stored in the metadata. If this name
2411 suggests a unit number, that number will be used, otherwise a free
2412 unit number will be chosen. Normally
2414 will prefer to create a partitionable array, however if the
2418 suggests that a non-partitionable array is preferred, that will be
2421 If the array is not found in the config file and its metadata does not
2422 identify it as belonging to the "homehost", then
2424 will choose a name for the array which is certain not to conflict with
2425 any array which does belong to this host. It does this be adding an
2426 underscore and a small number to the name preferred by the metadata.
2428 Once an appropriate array is found or created and the device is added,
2430 must decide if the array is ready to be started. It will
2431 normally compare the number of available (non-spare) devices to the
2432 number of devices that the metadata suggests need to be active. If
2433 there are at least that many, the array will be started. This means
2434 that if any devices are missing the array will not be restarted.
2440 in which case the array will be run as soon as there are enough
2441 devices present for the data to be accessible. For a RAID1, that
2442 means one device will start the array. For a clean RAID5, the array
2443 will be started as soon as all but one drive is present.
2445 Note that neither of these approaches is really ideal. If it can
2446 be known that all device discovery has completed, then
2450 can be run which will try to start all arrays that are being
2451 incrementally assembled. They are started in "read-auto" mode in
2452 which they are read-only until the first write request. This means
2453 that no metadata updates are made and no attempt at resync or recovery
2454 happens. Further devices that are found before the first write can
2455 still be added safely.
2458 This section describes environment variables that affect how mdadm
2463 Setting this value to 1 will prevent mdadm from automatically launching
2464 mdmon. This variable is intended primarily for debugging mdadm/mdmon.
2470 does not create any device nodes in /dev, but leaves that task to
2474 appears not to be configured, or if this environment variable is set
2477 will create and devices that are needed.
2481 .B " mdadm \-\-query /dev/name-of-device"
2483 This will find out if a given device is a RAID array, or is part of
2484 one, and will provide brief information about the device.
2486 .B " mdadm \-\-assemble \-\-scan"
2488 This will assemble and start all arrays listed in the standard config
2489 file. This command will typically go in a system startup file.
2491 .B " mdadm \-\-stop \-\-scan"
2493 This will shut down all arrays that can be shut down (i.e. are not
2494 currently in use). This will typically go in a system shutdown script.
2496 .B " mdadm \-\-follow \-\-scan \-\-delay=120"
2498 If (and only if) there is an Email address or program given in the
2499 standard config file, then
2500 monitor the status of all arrays listed in that file by
2501 polling them ever 2 minutes.
2503 .B " mdadm \-\-create /dev/md0 \-\-level=1 \-\-raid\-devices=2 /dev/hd[ac]1"
2505 Create /dev/md0 as a RAID1 array consisting of /dev/hda1 and /dev/hdc1.
2508 .B " echo 'DEVICE /dev/hd*[0\-9] /dev/sd*[0\-9]' > mdadm.conf"
2510 .B " mdadm \-\-detail \-\-scan >> mdadm.conf"
2512 This will create a prototype config file that describes currently
2513 active arrays that are known to be made from partitions of IDE or SCSI drives.
2514 This file should be reviewed before being used as it may
2515 contain unwanted detail.
2517 .B " echo 'DEVICE /dev/hd[a\-z] /dev/sd*[a\-z]' > mdadm.conf"
2519 .B " mdadm \-\-examine \-\-scan \-\-config=mdadm.conf >> mdadm.conf"
2521 This will find arrays which could be assembled from existing IDE and
2522 SCSI whole drives (not partitions), and store the information in the
2523 format of a config file.
2524 This file is very likely to contain unwanted detail, particularly
2527 entries. It should be reviewed and edited before being used as an
2530 .B " mdadm \-\-examine \-\-brief \-\-scan \-\-config=partitions"
2532 .B " mdadm \-Ebsc partitions"
2534 Create a list of devices by reading
2535 .BR /proc/partitions ,
2536 scan these for RAID superblocks, and printout a brief listing of all
2539 .B " mdadm \-Ac partitions \-m 0 /dev/md0"
2541 Scan all partitions and devices listed in
2542 .BR /proc/partitions
2545 out of all such devices with a RAID superblock with a minor number of 0.
2547 .B " mdadm \-\-monitor \-\-scan \-\-daemonise > /var/run/mdadm"
2549 If config file contains a mail address or alert program, run mdadm in
2550 the background in monitor mode monitoring all md devices. Also write
2551 pid of mdadm daemon to
2552 .BR /var/run/mdadm .
2554 .B " mdadm \-Iq /dev/somedevice"
2556 Try to incorporate newly discovered device into some array as
2559 .B " mdadm \-\-incremental \-\-rebuild\-map \-\-run \-\-scan"
2561 Rebuild the array map from any current arrays, and then start any that
2564 .B " mdadm /dev/md4 --fail detached --remove detached"
2566 Any devices which are components of /dev/md4 will be marked as faulty
2567 and then remove from the array.
2569 .B " mdadm --grow /dev/md4 --level=6 --backup-file=/root/backup-md4
2573 which is currently a RAID5 array will be converted to RAID6. There
2574 should normally already be a spare drive attached to the array as a
2575 RAID6 needs one more drive than a matching RAID5.
2577 .B " mdadm --create /dev/md/ddf --metadata=ddf --raid-disks 6 /dev/sd[a-f]"
2579 Create a DDF array over 6 devices.
2581 .B " mdadm --create /dev/md/home -n3 -l5 -z 30000000 /dev/md/ddf"
2583 Create a RAID5 array over any 3 devices in the given DDF set. Use
2584 only 30 gigabytes of each device.
2586 .B " mdadm -A /dev/md/ddf1 /dev/sd[a-f]"
2588 Assemble a pre-exist ddf array.
2590 .B " mdadm -I /dev/md/ddf1"
2592 Assemble all arrays contained in the ddf array, assigning names as
2595 .B " mdadm \-\-create \-\-help"
2597 Provide help about the Create mode.
2599 .B " mdadm \-\-config \-\-help"
2601 Provide help about the format of the config file.
2603 .B " mdadm \-\-help"
2605 Provide general help.
2615 lists all active md devices with information about them.
2617 uses this to find arrays when
2619 is given in Misc mode, and to monitor array reconstruction
2624 The config file lists which devices may be scanned to see if
2625 they contain MD super block, and gives identifying information
2626 (e.g. UUID) about known MD arrays. See
2630 .SS /var/run/mdadm/map
2633 mode is used, this file gets a list of arrays currently being created.
2636 does not exist as a directory, then
2637 .B /var/run/mdadm.map
2640 is not available (as may be the case during early boot),
2642 is used on the basis that
2644 is usually available very early in boot.
2649 understand two sorts of names for array devices.
2651 The first is the so-called 'standard' format name, which matches the
2652 names used by the kernel and which appear in
2655 The second sort can be freely chosen, but must reside in
2657 When giving a device name to
2659 to create or assemble an array, either full path name such as
2663 can be given, or just the suffix of the second sort of name, such as
2669 chooses device names during auto-assembly or incremental assembly, it
2670 will sometimes add a small sequence number to the end of the name to
2671 avoid conflicted between multiple arrays that have the same name. If
2673 can reasonably determine that the array really is meant for this host,
2674 either by a hostname in the metadata, or by the presence of the array
2677 then it will leave off the suffix if possible.
2678 Also if the homehost is specified as
2681 will only use a suffix if a different array of the same name already
2682 exists or is listed in the config file.
2684 The standard names for non-partitioned arrays (the only sort of md
2685 array available in 2.4 and earlier) are of the form
2689 where NN is a number.
2690 The standard names for partitionable arrays (as available from 2.6
2691 onwards) are of the form
2695 Partition numbers should be indicated by added "pMM" to these, thus "/dev/md/d1p2".
2697 From kernel version, 2.6.28 the "non-partitioned array" can actually
2698 be partitioned. So the "md_dNN" names are no longer needed, and
2699 partitions such as "/dev/mdNNpXX" are possible.
2703 was previously known as
2707 is completely separate from the
2709 package, and does not use the
2711 configuration file at all.
2714 For further information on mdadm usage, MD and the various levels of
2717 .B http://linux\-raid.osdl.org/
2719 (based upon Jakob \(/Ostergaard's Software\-RAID.HOWTO)
2721 .\"for new releases of the RAID driver check out:
2724 .\".UR ftp://ftp.kernel.org/pub/linux/kernel/people/mingo/raid-patches
2725 .\"ftp://ftp.kernel.org/pub/linux/kernel/people/mingo/raid-patches
2730 .\".UR http://www.cse.unsw.edu.au/~neilb/patches/linux-stable/
2731 .\"http://www.cse.unsw.edu.au/~neilb/patches/linux-stable/
2734 The latest version of
2736 should always be available from
2738 .B http://www.kernel.org/pub/linux/utils/raid/mdadm/