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 there-of) 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
45 is not a Software RAID mechanism, but does involve
48 each device is a path to one common physical storage device.
51 is also not true RAID, and it only involves one device. It
52 provides a layer over a true device that can be used to inject faults.
55 '''is a program that can be used to create, manage, and monitor
57 '''such it provides a similar set of functionality to the
60 '''The key differences between
67 '''is a single program and not a collection of programs.
70 '''can perform (almost) all of its functions without having a
71 '''configuration file and does not use one by default. Also
73 '''helps with management of the configuration
77 '''can provide information about your arrays (through Query, Detail, and Examine)
87 '''configuration file, at all. It has a different configuration file
88 '''with a different format and a different purpose.
91 mdadm has several major modes of operation:
94 Assemble the parts of a previously created
95 array into an active array. Components can be explicitly given
96 or can be searched for.
98 checks that the components
99 do form a bona fide array, and can, on request, fiddle superblock
100 information so as to assemble a faulty array.
104 Build an array that doesn't have per-device superblocks. For these
107 cannot differentiate between initial creation and subsequent assembly
108 of an array. It also cannot perform any checks that appropriate
109 devices have been requested. Because of this, the
111 mode should only be used together with a complete understanding of
116 Create a new array with per-device superblocks.
118 '''in several step create-add-add-run or it can all happen with one command.
121 .B "Follow or Monitor"
122 Monitor one or more md devices and act on any state changes. This is
123 only meaningful for raid1, 4, 5, 6, 10 or multipath arrays as
124 only these have interesting state. raid0 or linear never have
125 missing, spare, or failed drives, so there is nothing to monitor.
129 Grow (or shrink) an array, or otherwise reshape it in some way.
130 Currently supported growth options including changing the active size
131 of component devices in RAID level 1/4/5/6 and changing the number of
132 active devices in RAID1/5/6.
135 .B "Incremental Assembly"
136 Add a single device to an appropriate array. If the addition of the
137 device makes the array runnable, the array will be started.
138 This provides a convenient interface to a
140 system. As each device is detected,
142 has a chance to include it in some array as appropriate.
146 This is for doing things to specific components of an array such as
147 adding new spares and removing faulty devices.
151 This is an 'everything else' mode that supports operations on active
152 arrays, operations on component devices such as erasing old superblocks, and
153 information gathering operations.
154 '''This mode allows operations on independent devices such as examine MD
155 '''superblocks, erasing old superblocks and stopping active arrays.
159 .SH Options for selecting a mode are:
162 .BR -A ", " --assemble
163 Assemble a pre-existing array.
167 Build a legacy array without superblocks.
174 .BR -F ", " --follow ", " --monitor
181 Change the size or shape of an active array.
184 .BE -I ", " --incremental
185 Add a single device into an appropriate array, and possibly start the array.
188 If a device is given before any options, or if the first option is
193 then the MANAGE mode is assume.
194 Anything other than these will cause the
198 .SH Options that are not mode-specific are:
202 Display general help message or, after one of the above options, a
203 mode specific help message.
207 Display more detailed help about command line parsing and some commonly
211 .BR -V ", " --version
212 Print version information for mdadm.
215 .BR -v ", " --verbose
216 Be more verbose about what is happening. This can be used twice to be
218 The extra verbosity currently only affects
221 .BR "--examine --scan" .
225 Avoid printing purely informative messages. With this,
227 will be silent unless there is something really important to report.
231 Be less verbose. This is used with
239 gives an intermediate level of verbosity.
243 Be more forceful about certain operations. See the various modes of
244 the exact meaning of this option in different contexts.
247 .BR -c ", " --config=
248 Specify the config file. Default is to use
249 .BR /etc/mdadm.conf ,
250 or if that is missing, then
251 .BR /etc/mdadm/mdadm.conf .
252 If the config file given is
254 then nothing will be read, but
256 will act as though the config file contained exactly
257 .B "DEVICE partitions"
260 to find a list of devices to scan.
263 is given for the config file, then
265 will act as though the config file were empty.
271 for missing information.
272 In general, this option gives
274 permission to get any missing information, like component devices,
275 array devices, array identities, and alert destination from the
277 .BR /etc/mdadm.conf .
278 One exception is MISC mode when using
284 says to get a list of array devices from
288 .B -e ", " --metadata=
289 Declare the style of superblock (raid metadata) to be used. The
290 default is 0.90 for --create, and to guess for other operations.
291 The default can be overridden by setting the
300 .IP "0, 0.90, default"
301 Use the original 0.90 format superblock. This format limits arrays to
302 28 componenet devices and limits component devices of levels 1 and
303 greater to 2 terabytes.
304 .IP "1, 1.0, 1.1, 1.2"
305 Use the new version-1 format superblock. This has few restrictions.
306 The different subversion store the superblock at different locations
307 on the device, either at the end (for 1.0), at the start (for 1.1) or
308 4K from the start (for 1.2).
313 This will over-ride any
315 setting in the config file and provides the identify of the host which
316 should be considered the home for any arrays.
318 When creating an array, the
320 will be recorded in the superblock. For version-1 superblocks, it will
321 be prefixed to the array name. For version-0.90 superblocks part of
322 the SHA1 hash of the hostname will be stored in the later half of the
325 When reporting information about an array, any array which is tagged
326 for the given homehost will be reported as such.
328 When using Auto-Assemble, only arrays tagged for the given homehost
331 .SH For create, build, or grow:
334 .BR -n ", " --raid-devices=
335 Specify the number of active devices in the array. This, plus the
336 number of spare devices (see below) must equal the number of
338 (including "\fBmissing\fP" devices)
339 that are listed on the command line for
341 Setting a value of 1 is probably
342 a mistake and so requires that
344 be specified first. A value of 1 will then be allowed for linear,
345 multipath, raid0 and raid1. It is never allowed for raid4 or raid5.
347 This number can only be changed using
349 for RAID1, RAID5 and RAID6 arrays, and only on kernels which provide
353 .BR -x ", " --spare-devices=
354 Specify the number of spare (eXtra) devices in the initial array.
355 Spares can also be added
356 and removed later. The number of component devices listed
357 on the command line must equal the number of raid devices plus the
358 number of spare devices.
363 Amount (in Kibibytes) of space to use from each drive in RAID1/4/5/6.
364 This must be a multiple of the chunk size, and must leave about 128Kb
365 of space at the end of the drive for the RAID superblock.
366 If this is not specified
367 (as it normally is not) the smallest drive (or partition) sets the
368 size, though if there is a variance among the drives of greater than 1%, a warning is
371 This value can be set with
373 for RAID level 1/4/5/6. If the array was created with a size smaller
374 than the currently active drives, the extra space can be accessed
377 The size can be given as
379 which means to choose the largest size that fits on all current drives.
383 Specify chunk size of kibibytes. The default is 64.
387 Specify rounding factor for linear array (==chunk size)
391 Set raid level. When used with
393 options are: linear, raid0, 0, stripe, raid1, 1, mirror, raid4, 4,
394 raid5, 5, raid6, 6, raid10, 10, multipath, mp, faulty. Obviously some of these are synonymous.
398 only linear, stripe, raid0, 0, raid1, multipath, mp, and faulty are valid.
400 Not yet supported with
404 .BR -p ", " --layout=
405 This option configures the fine details of data layout for raid5,
406 and raid10 arrays, and controls the failure modes for
409 The layout of the raid5 parity block can be one of
414 la, ra, ls, rs. The default is left-symmetric.
416 When setting the failure mode for
434 Each mode can be followed by a number which is used as a period
435 between fault generation. Without a number, the fault is generated
436 once on the first relevant request. With a number, the fault will be
437 generated after that many request, and will continue to be generated
438 every time the period elapses.
440 Multiple failure modes can be current simultaneously by using the
441 "--grow" option to set subsequent failure modes.
443 "clear" or "none" will remove any pending or periodic failure modes,
444 and "flush" will clear any persistent faults.
446 To set the parity with "--grow", the level of the array ("faulty")
447 must be specified before the fault mode is specified.
449 Finally, the layout options for RAID10 are one of 'n', 'o' or 'p' followed
450 by a small number. The default is 'n2'.
453 signals 'near' copies. Multiple copies of one data block are at
454 similar offsets in different devices.
457 signals 'offset' copies. Rather than the chunks being duplicated
458 within a stripe, whole stripes are duplicated but are rotated by one
459 device so duplicate blocks are on different devices. Thus subsequent
460 copies of a block are in the next drive, and are one chunk further
465 (multiple copies have very different offsets). See md(4) for more
466 detail about 'near' and 'far'.
468 The number is the number of copies of each datablock. 2 is normal, 3
469 can be useful. This number can be at most equal to the number of
470 devices in the array. It does not need to divide evenly into that
471 number (e.g. it is perfectly legal to have an 'n2' layout for an array
472 with an odd number of devices).
476 same as --layout (thus explaining the p of
480 .BR -b ", " --bitmap=
481 Specify a file to store a write-intent bitmap in. The file should not
482 exist unless --force is also given. The same file should be provided
483 when assembling the array. If the word
485 is given, then the bitmap is stored with the metadata on the array,
486 and so is replicated on all devices. If the word
490 mode, then any bitmap that is present is removed.
492 To help catch typing errors, the filename must contain at least one
493 slash ('/') if it is a real file (not 'internal' or 'none').
495 Note: external bitmaps are only known to work on ext2 and ext3.
496 Storing bitmap files on other filesystems may result in serious problems.
500 Set the chunksize of the bitmap. Each bit corresponds to that many
501 Kilobytes of storage.
502 When using a file based bitmap, the default is to use the smallest
503 size that is atleast 4 and requires no more than 2^21 chunks.
506 bitmap, the chunksize is automatically determined to make best use of
511 .BR -W ", " --write-mostly
512 subsequent devices lists in a
517 command will be flagged as 'write-mostly'. This is valid for RAID1
518 only and means that the 'md' driver will avoid reading from these
519 devices if at all possible. This can be useful if mirroring over a
524 Specify that write-behind mode should be enabled (valid for RAID1
525 only). If an argument is specified, it will set the maximum number
526 of outstanding writes allowed. The default value is 256.
527 A write-intent bitmap is required in order to use write-behind
528 mode, and write-behind is only attempted on drives marked as
535 that the array pre-existed and is known to be clean. It can be useful
536 when trying to recover from a major failure as you can be sure that no
537 data will be affected unless you actually write to the array. It can
538 also be used when creating a RAID1 or RAID10 if you want to avoid the
539 initial resync, however this practice - while normally safe - is not
540 recommended. Use this ony if you really know what you are doing.
544 This is needed when --grow is used to increase the number of
545 raid-devices in a RAID5 if there are no spare devices available.
546 See the section below on RAID_DEVICE CHANGES. The file should be
547 stored on a separate device, not on the raid array being reshaped.
553 for the array. This is currently only effective when creating an
554 array with a version-1 superblock. The name is a simple textual
555 string that can be used to identify array components when assembling.
561 run the array, even if some of the components
562 appear to be active in another array or filesystem. Normally
564 will ask for confirmation before including such components in an
565 array. This option causes that question to be suppressed.
571 accept the geometry and layout specified without question. Normally
573 will not allow creation of an array with only one device, and will try
574 to create a raid5 array with one missing drive (as this makes the
575 initial resync work faster). With
578 will not try to be so clever.
581 .BR -a ", " "--auto{=no,yes,md,mdp,part,p}{NN}"
582 Instruct mdadm to create the device file if needed, possibly allocating
583 an unused minor number. "md" causes a non-partitionable array
584 to be used. "mdp", "part" or "p" causes a partitionable array (2.6 and
585 later) to be used. "yes" requires the named md device to have
586 a 'standard' format, and the type and minor number will be determined
587 from this. See DEVICE NAMES below.
589 The argument can also come immediately after
592 If --auto is not given on the command line or in the config file, then
598 is also given, then any
600 entries in the config file will over-ride the
602 instruction given on the command line.
604 For partitionable arrays,
606 will create the device file for the whole array and for the first 4
607 partitions. A different number of partitions can be specified at the
608 end of this option (e.g.
610 If the device name ends with a digit, the partition names add a 'p',
611 and a number, e.g. "/dev/home1p3". If there is no
612 trailing digit, then the partition names just have a number added,
613 e.g. "/dev/scratch3".
615 If the md device name is in a 'standard' format as described in DEVICE
616 NAMES, then it will be created, if necessary, with the appropriate
617 number based on that name. If the device name is not in one of these
618 formats, then a unused minor number will be allocated. The minor
619 number will be considered unused if there is no active array for that
620 number, and there is no entry in /dev for that number and with a
631 it will also create symlinks from
633 with names starting with
641 to enforce this even if it is suppressing
649 uuid of array to assemble. Devices which don't have this uuid are
653 .BR -m ", " --super-minor=
654 Minor number of device that array was created for. Devices which
655 don't have this minor number are excluded. If you create an array as
656 /dev/md1, then all superblocks will contain the minor number 1, even if
657 the array is later assembled as /dev/md2.
659 Giving the literal word "dev" for
663 to use the minor number of the md device that is being assembled.
667 will look for super blocks with a minor number of 0.
671 Specify the name of the array to assemble. This must be the name
672 that was specified when creating the array. It must either match
673 then name stored in the superblock exactly, or it must match
676 is added to the start of the given name.
680 Assemble the array even if some superblocks appear out-of-date
684 Attempt to start the array even if fewer drives were given than were
685 present last time the array was active. Normally if not all the
686 expected drives are found and
688 is not used, then the array will be assembled but not started.
691 an attempt will be made to start it anyway.
695 This is the reverse of
697 in that it inhibits the started if array unless all expected drives
698 are present. This is only needed with
700 and can be used if you physical connections to devices are
701 not as reliable as you would like.
704 .BR -a ", " "--auto{=no,yes,md,mdp,part}"
705 See this option under Create and Build options.
708 .BR -b ", " --bitmap=
709 Specify the bitmap file that was given when the array was created. If
712 bitmap, there is no need to specify this when assembling the array.
718 was used to grow the number of raid-devices in a RAID5, and the system
719 crashed during the critical section, then the same
721 must be presented to --assemble to allow possibly corrupted data to be
725 .BR -U ", " --update=
726 Update the superblock on each device while assembling the array. The
727 argument given to this flag can be one of
741 option will adjust the superblock of an array what was created on a Sparc
742 machine running a patched 2.2 Linux kernel. This kernel got the
743 alignment of part of the superblock wrong. You can use the
744 .B "--examine --sparc2.2"
747 to see what effect this would have.
751 option will update the
753 field on each superblock to match the minor number of the array being
755 This can be useful if
757 reports a different "Preferred Minor" to
759 In some cases this update will be performed automatically
760 by the kernel driver. In particular the update happens automatically
761 at the first write to an array with redundancy (RAID level 1 or
762 greater) on a 2.6 (or later) kernel.
766 option will change the uuid of the array. If a UUID is given with the
767 "--uuid" option that UUID will be used as a new UUID and will
769 be used to help identify the devices in the array.
770 If no "--uuid" is given, a random uuid is chosen.
774 option will change the
776 of the array as stored in the superblock. This is only supported for
777 version-1 superblocks.
781 option will change the
783 as recorded in the superblock. For version-0 superblocks, this is the
784 same as updating the UUID.
785 For version-1 superblocks, this involves updating the name.
789 option will cause the array to be marked
791 meaning that any redundancy in the array (e.g. parity for raid5,
792 copies for raid1) may be incorrect. This will cause the raid system
793 to perform a "resync" pass to make sure that all redundant information
798 option allows arrays to be moved between machines with different
800 When assembling such an array for the first time after a move, giving
801 .B "--update=byteorder"
804 to expect superblocks to have their byteorder reversed, and will
805 correct that order before assembling the array. This is only valid
806 with original (Version 0.90) superblocks.
810 option will correct the summaries in the superblock. That is the
811 counts of total, working, active, failed, and spare devices.
815 will rarely be of use. It applies to version 1.1 and 1.2 metadata
816 only (where the metadata is at the start of the device) and is only
817 useful when the component device has changed size (typically become
818 larger). The version 1 metadata records the amount of the device that
819 can be used to store data, so if a device in a version 1.1 or 1.2
820 array becomes larger, the metadata will still be visible, but the
821 extra space will not. In this case it might be useful to assemble the
823 .BR --update=devicesize .
826 to determine the maximum usable amount of space on each device and
827 update the relevant field in the metadata.
830 .B --auto-update-homehost
831 This flag is only meaning with auto-assembly (see discussion below).
832 In that situation, if no suitable arrays are found for this homehost,
834 will recan for any arrays at all and will assemble them and update the
835 homehost to match the current host.
841 hot-add listed devices.
845 re-add a device that was recently removed from an array.
849 remove listed devices. They must not be active. i.e. they should
850 be failed or spare devices.
854 mark listed devices as faulty.
861 Each of these options require that the first device list is the array
862 to be acted upon and the remainder are component devices to be added,
863 removed, or marked as fault. Several different operations can be
864 specified for different devices, e.g.
866 mdadm /dev/md0 --add /dev/sda1 --fail /dev/sdb1 --remove /dev/sdb1
868 Each operation applies to all devices listed until the next
871 If an array is using a write-intent bitmap, then devices which have
872 been removed can be re-added in a way that avoids a full
873 reconstruction but instead just updated the blocks that have changed
874 since the device was removed. For arrays with persistent metadata
875 (superblocks) this is done automatically. For arrays created with
877 mdadm needs to be told that this device we removed recently with
880 Devices can only be removed from an array if they are not in active
881 use. i.e. that must be spares or failed devices. To remove an active
882 device, it must be marked as
890 Examine a device to see
891 (1) if it is an md device and (2) if it is a component of an md
893 Information about what is discovered is presented.
897 Print detail of one or more md devices.
900 .BR -E ", " --examine
901 Print content of md superblock on device(s).
904 If an array was created on a 2.2 Linux kernel patched with RAID
905 support, the superblock will have been created incorrectly, or at
906 least incompatibly with 2.4 and later kernels. Using the
910 will fix the superblock before displaying it. If this appears to do
911 the right thing, then the array can be successfully assembled using
912 .BR "--assemble --update=sparc2.2" .
915 .BR -X ", " --examine-bitmap
916 Report information about a bitmap file.
920 start a partially built array.
924 deactivate array, releasing all resources.
927 .BR -o ", " --readonly
928 mark array as readonly.
931 .BR -w ", " --readwrite
932 mark array as readwrite.
936 If the device contains a valid md superblock, the block is
937 over-written with zeros. With
939 the block where the superblock would be is over-written even if it
940 doesn't appear to be valid.
948 is set to reflect the status of the device.
952 For each md device given, wait for any resync, recovery, or reshape
953 activity to finish before returning.
955 will return with success if it actually waited for every device
956 listed, otherwise it will return failure.
958 .SH For Incremental Assembly mode:
960 .BR --rebuild-map ", " -r
962 .RB ( /var/run/mdadm/map )
965 uses to help track which arrays are currently being assembled.
969 Run any array assembled as soon as a minimal number of devices are
970 available, rather than waiting until all expected devices are present.
978 file for arrays that are being incrementally assembled and will try to
979 start any that are not already started. If any such array is listed
982 as requiring an external bitmap, that bitmap will be attached first.
984 .SH For Monitor mode:
987 Give a mail address to send alerts to.
990 .BR -p ", " --program ", " --alert
991 Give a program to be run whenever an event is detected.
995 Cause all events to be reported through 'syslog'. The messages have
996 facility of 'daemon' and varying priorities.
1000 Give a delay in seconds.
1002 polls the md arrays and then waits this many seconds before polling
1003 again. The default is 60 seconds.
1006 .BR -f ", " --daemonise
1009 to run as a background daemon if it decides to monitor anything. This
1010 causes it to fork and run in the child, and to disconnect form the
1011 terminal. The process id of the child is written to stdout.
1014 which will only continue monitoring if a mail address or alert program
1015 is found in the config file.
1018 .BR -i ", " --pid-file
1021 is running in daemon mode, write the pid of the daemon process to
1022 the specified file, instead of printing it on standard output.
1025 .BR -1 ", " --oneshot
1026 Check arrays only once. This will generate
1028 events and more significantly
1034 .B " mdadm --monitor --scan -1"
1036 from a cron script will ensure regular notification of any degraded arrays.
1042 alert for every array found at startup. This alert gets mailed and
1043 passed to the alert program. This can be used for testing that alert
1044 message do get through successfully.
1051 .I md-device options-and-component-devices...
1054 .B mdadm --assemble --scan
1055 .I md-devices-and-options...
1058 .B mdadm --assemble --scan
1062 This usage assembles one or more raid arrays from pre-existing components.
1063 For each array, mdadm needs to know the md device, the identity of the
1064 array, and a number of component-devices. These can be found in a number of ways.
1066 In the first usage example (without the
1068 the first device given is the md device.
1069 In the second usage example, all devices listed are treated as md
1070 devices and assembly is attempted.
1071 In the third (where no devices are listed) all md devices that are
1072 listed in the configuration file are assembled.
1074 If precisely one device is listed, but
1080 was given and identify information is extracted from the configuration file.
1082 The identity can be given with the
1086 option, can be found in the config file, or will be taken from the
1087 super block on the first component-device listed on the command line.
1089 Devices can be given on the
1091 command line or in the config file. Only devices which have an md
1092 superblock which contains the right identity will be considered for
1095 The config file is only used if explicitly named with
1097 or requested with (a possibly implicit)
1105 is not given, then the config file will only be used to find the
1106 identity of md arrays.
1108 Normally the array will be started after it is assembled. However if
1110 is not given and insufficient drives were listed to start a complete
1111 (non-degraded) array, then the array is not started (to guard against
1112 usage errors). To insist that the array be started in this case (as
1113 may work for RAID1, 4, 5, 6, or 10), give the
1117 If the md device does not exist, then it will be created providing the
1118 intent is clear. i.e. the name must be in a standard form, or the
1120 option must be given to clarify how and whether the device should be
1123 This can be useful for handling partitioned devices (which don't have
1124 a stable device number - it can change after a reboot) and when using
1125 "udev" to manage your
1127 tree (udev cannot handle md devices because of the unusual device
1128 initialisation conventions).
1130 If the option to "auto" is "mdp" or "part" or (on the command line
1131 only) "p", then mdadm will create a partitionable array, using the
1132 first free one that is not in use, and does not already have an entry
1133 in /dev (apart from numeric /dev/md* entries).
1135 If the option to "auto" is "yes" or "md" or (on the command line)
1136 nothing, then mdadm will create a traditional, non-partitionable md
1139 It is expected that the "auto" functionality will be used to create
1140 device entries with meaningful names such as "/dev/md/home" or
1141 "/dev/md/root", rather than names based on the numerical array number.
1143 When using this option to create a partitionable array, the device
1144 files for the first 4 partitions are also created. If a different
1145 number is required it can be simply appended to the auto option.
1146 e.g. "auto=part8". Partition names are created by appending a digit
1147 string to the device name, with an intervening "p" if the device name
1152 option is also available in Build and Create modes. As those modes do
1153 not use a config file, the "auto=" config option does not apply to
1161 and no devices are listed,
1163 will first attempt to assemble all the arrays listed in the config
1168 has been specified (either in the config file or on the command line),
1170 will look further for possible arrays and will try to assemble
1171 anything that it finds which is tagged as belonging to the given
1172 homehost. This is the only situation where
1174 will assemble arrays without being given specific device name or
1175 identify information for the array.
1179 finds a consistent set of devices that look like they should comprise
1180 an array, and if the superblock is tagged as belonging to the given
1181 home host, it will automatically choose a device name and try to
1182 assemble the array. If the array uses version-0.90 metadata, then the
1184 number as recorded in the superblock is used to create a name in
1188 If the array uses version-1 metadata, then the
1190 from the superblock is used to similarly create a name in
1192 The name will have any 'host' prefix stripped first.
1196 cannot find any array for the given host at all, and if
1197 .B --auto-update-homehost
1200 will search again for any array (not just an array created for this
1201 host) and will assemble each assuming
1202 .IR --update=homehost .
1203 This will change the host tag in the superblock so that on the next run,
1204 these arrays will be found without the second pass. The intention of
1205 this feature is to support transitioning a set of md arrays to using
1208 The reason for requiring arrays to be tagged with the homehost for
1209 auto assembly is to guard against problems that can arise when moving
1210 devices from one host to another.
1220 .BI --raid-devices= Z
1224 This usage is similar to
1226 The difference is that it creates an array without a superblock. With
1227 these arrays there is no difference between initially creating the array and
1228 subsequently assembling the array, except that hopefully there is useful
1229 data there in the second case.
1231 The level may raid0, linear, multipath, or faulty, or one of their
1232 synonyms. All devices must be listed and the array will be started
1244 .BI --raid-devices= Z
1248 This usage will initialise a new md array, associate some devices with
1249 it, and activate the array.
1253 option is given (as described in more detail in the section on
1254 Assemble mode), then the md device will be created with a suitable
1255 device number if necessary.
1257 As devices are added, they are checked to see if they contain raid
1258 superblocks or filesystems. They are also checked to see if the variance in
1259 device size exceeds 1%.
1261 If any discrepancy is found, the array will not automatically be run, though
1264 can override this caution.
1266 To create a "degraded" array in which some devices are missing, simply
1267 give the word "\fBmissing\fP"
1268 in place of a device name. This will cause
1270 to leave the corresponding slot in the array empty.
1271 For a RAID4 or RAID5 array at most one slot can be
1272 "\fBmissing\fP"; for a RAID6 array at most two slots.
1273 For a RAID1 array, only one real device needs to be given. All of the
1277 When creating a RAID5 array,
1279 will automatically create a degraded array with an extra spare drive.
1280 This is because building the spare into a degraded array is in general faster than resyncing
1281 the parity on a non-degraded, but not clean, array. This feature can
1282 be over-ridden with the
1286 When creating an array with version-1 metadata a name for the host is
1288 If this is not given with the
1292 will chose a name based on the last component of the name of the
1293 device being created. So if
1295 is being created, then the name
1300 is being created, then the name
1304 A new array will normally get a randomly assigned 128bit UUID which is
1305 very likely to be unique. If you have a specific need, you can choose
1306 a UUID for the array by giving the
1308 option. Be warned that creating two arrays with the same UUID is a
1309 recipe for disaster. Also, using
1311 when creating a v0.90 array will silently override any
1316 '''option is given, it is not necessary to list any component-devices in this command.
1317 '''They can be added later, before a
1321 '''is given, the apparent size of the smallest drive given is used.
1323 The General Management options that are valid with --create are:
1326 insist on running the array even if some devices look like they might
1331 start the array readonly - not supported yet.
1339 .I options... devices...
1342 This usage will allow individual devices in an array to be failed,
1343 removed or added. It is possible to perform multiple operations with
1344 on command. For example:
1346 .B " mdadm /dev/md0 -f /dev/hda1 -r /dev/hda1 -a /dev/hda1"
1352 and will then remove it from the array and finally add it back
1353 in as a spare. However only one md array can be affected by a single
1364 MISC mode includes a number of distinct operations that
1365 operate on distinct devices. The operations are:
1368 The device is examined to see if it is
1369 (1) an active md array, or
1370 (2) a component of an md array.
1371 The information discovered is reported.
1375 The device should be an active md device.
1377 will display a detailed description of the array.
1381 will cause the output to be less detailed and the format to be
1382 suitable for inclusion in
1383 .BR /etc/mdadm.conf .
1386 will normally be 0 unless
1388 failed to get useful information about the device(s). However if the
1390 option is given, then the exit status will be:
1394 The array is functioning normally.
1397 The array has at least one failed device.
1400 The array has multiple failed devices and hence is unusable (raid4 or
1404 There was an error while trying to get information about the device.
1409 The device should be a component of an md array.
1411 will read the md superblock of the device and display the contents.
1416 then multiple devices that are components of the one array
1417 are grouped together and reported in a single entry suitable
1419 .BR /etc/mdadm.conf .
1423 without listing any devices will cause all devices listed in the
1424 config file to be examined.
1428 The devices should be active md arrays which will be deactivated, as
1429 long as they are not currently in use.
1433 This will fully activate a partially assembled md array.
1437 This will mark an active array as read-only, providing that it is
1438 not currently being used.
1444 array back to being read/write.
1448 For all operations except
1451 will cause the operation to be applied to all arrays listed in
1456 causes all devices listed in the config file to be examined.
1464 .I options... devices...
1469 to periodically poll a number of md arrays and to report on any events
1472 will never exit once it decides that there are arrays to be checked,
1473 so it should normally be run in the background.
1475 As well as reporting events,
1477 may move a spare drive from one array to another if they are in the
1480 and if the destination array has a failed drive but no spares.
1482 If any devices are listed on the command line,
1484 will only monitor those devices. Otherwise all arrays listed in the
1485 configuration file will be monitored. Further, if
1487 is given, then any other md devices that appear in
1489 will also be monitored.
1491 The result of monitoring the arrays is the generation of events.
1492 These events are passed to a separate program (if specified) and may
1493 be mailed to a given E-mail address.
1495 When passing event to program, the program is run once for each event
1496 and is given 2 or 3 command-line arguments. The first is the
1497 name of the event (see below). The second is the name of the
1498 md device which is affected, and the third is the name of a related
1499 device if relevant, such as a component device that has failed.
1503 is given, then a program or an E-mail address must be specified on the
1504 command line or in the config file. If neither are available, then
1506 will not monitor anything.
1510 will continue monitoring as long as something was found to monitor. If
1511 no program or email is given, then each event is reported to
1514 The different events are:
1518 .B DeviceDisappeared
1519 An md array which previously was configured appears to no longer be
1520 configured. (syslog priority: Critical)
1524 was told to monitor an array which is RAID0 or Linear, then it will
1526 .B DeviceDisappeared
1527 with the extra information
1529 This is because RAID0 and Linear do not support the device-failed,
1530 hot-spare and resync operations which are monitored.
1534 An md array started reconstruction. (syslog priority: Warning)
1540 is 20, 40, 60, or 80, this indicates that rebuild has passed that many
1541 percentage of the total. (syslog priority: Warning)
1545 An md array that was rebuilding, isn't any more, either because it
1546 finished normally or was aborted. (syslog priority: Warning)
1550 An active component device of an array has been marked as
1551 faulty. (syslog priority: Critical)
1555 A spare component device which was being rebuilt to replace a faulty
1556 device has failed. (syslog priority: Critial)
1560 A spare component device which was being rebuilt to replace a faulty
1561 device has been successfully rebuilt and has been made active.
1562 (syslog priority: Info)
1566 A new md array has been detected in the
1568 file. (syslog priority: Info)
1572 A newly noticed array appears to be degraded. This message is not
1575 notices a drive failure which causes degradation, but only when
1577 notices that an array is degraded when it first sees the array.
1578 (syslog priority: Critial)
1582 A spare drive has been moved from one array in a
1584 to another to allow a failed drive to be replaced.
1585 (syslog priority: Info)
1591 has been told, via the config file, that an array should have a certain
1592 number of spare devices, and
1594 detects that it has fewer that this number when it first sees the
1595 array, it will report a
1598 (syslog priority: Warning)
1602 An array was found at startup, and the
1605 (syslog priority: Info)
1615 cause Email to be sent. All events cause the program to be run.
1616 The program is run with two or three arguments, they being the event
1617 name, the array device and possibly a second device.
1619 Each event has an associated array device (e.g.
1621 and possibly a second device. For
1626 the second device is the relevant component device.
1629 the second device is the array that the spare was moved from.
1633 to move spares from one array to another, the different arrays need to
1634 be labelled with the same
1636 in the configuration file. The
1638 name can be any string. It is only necessary that different spare
1639 groups use different names.
1643 detects that an array which is in a spare group has fewer active
1644 devices than necessary for the complete array, and has no spare
1645 devices, it will look for another array in the same spare group that
1646 has a full complement of working drive and a spare. It will then
1647 attempt to remove the spare from the second drive and add it to the
1649 If the removal succeeds but the adding fails, then it is added back to
1653 The GROW mode is used for changing the size or shape of an active
1655 For this to work, the kernel must support the necessary change.
1656 Various types of growth are being added during 2.6 development,
1657 including restructuring a raid5 array to have more active devices.
1659 Currently the only support available is to
1661 change the "size" attribute
1662 for RAID1, RAID5 and RAID6.
1664 increase the "raid-disks" attribute of RAID1, RAID5, and RAID6.
1666 add a write-intent bitmap to any array which support these bitmaps, or
1667 remove a write-intent bitmap from such an array.
1671 Normally when an array is built the "size" it taken from the smallest
1672 of the drives. If all the small drives in an arrays are, one at a
1673 time, removed and replaced with larger drives, then you could have an
1674 array of large drives with only a small amount used. In this
1675 situation, changing the "size" with "GROW" mode will allow the extra
1676 space to start being used. If the size is increased in this way, a
1677 "resync" process will start to make sure the new parts of the array
1680 Note that when an array changes size, any filesystem that may be
1681 stored in the array will not automatically grow to use the space. The
1682 filesystem will need to be explicitly told to use the extra space.
1684 .SS RAID-DEVICES CHANGES
1686 A RAID1 array can work with any number of devices from 1 upwards
1687 (though 1 is not very useful). There may be times which you want to
1688 increase or decrease the number of active devices. Note that this is
1689 different to hot-add or hot-remove which changes the number of
1692 When reducing the number of devices in a RAID1 array, the slots which
1693 are to be removed from the array must already be vacant. That is, the
1694 devices that which were in those slots must be failed and removed.
1696 When the number of devices is increased, any hot spares that are
1697 present will be activated immediately.
1699 Increasing the number of active devices in a RAID5 is much more
1700 effort. Every block in the array will need to be read and written
1701 back to a new location. From 2.6.17, the Linux Kernel is able to do
1702 this safely, including restart and interrupted "reshape".
1704 When relocating the first few stripes on a raid5, it is not possible
1705 to keep the data on disk completely consistent and crash-proof. To
1706 provide the required safety, mdadm disables writes to the array while
1707 this "critical section" is reshaped, and takes a backup of the data
1708 that is in that section. This backup is normally stored in any spare
1709 devices that the array has, however it can also be stored in a
1710 separate file specified with the
1712 option. If this option is used, and the system does crash during the
1713 critical period, the same file must be passed to
1715 to restore the backup and reassemble the array.
1719 A write-intent bitmap can be added to, or removed from, an active
1720 array. Either internal bitmaps, or bitmaps stored in a separate file
1721 can be added. Note that if you add a bitmap stored in a file which is
1722 in a filesystem that is on the raid array being affected, the system
1723 will deadlock. The bitmap must be on a separate filesystem.
1725 .SH INCREMENTAL MODE
1729 .B mdadm --incremental
1735 .B mdadm --incremental --rebuild
1738 .B mdadm --incremental --run --scan
1742 This mode is designed to be used in conjunction with a device
1743 discovery system. As devices are found in a system, they can be
1745 .B "mdadm --incremental"
1746 to be conditionally added to an appropriate array.
1749 performs a number of tests to determine if the device is part of an
1750 array, and which array is should be part of. If an appropriate array
1751 is found, or can be created,
1753 adds the device to the array and conditionally starts the array.
1757 will only add devices to an array which were previously working
1758 (active or spare) parts of that array. It does not currently support
1759 automatic inclusion of a new drive as a spare in some array.
1761 .B "mdadm --incremental"
1762 requires a bug present in all kernels through 2.6.19, to be fixed.
1763 Hopefully this will be fixed in 2.6.20. Alternately apply the patch
1764 which is included with the mdadm source distribution. If
1766 detects that this bug is present, it will abort any attempt to use
1771 makes are as follow:
1773 Is the device permitted by
1775 That is, is it listed in a
1777 line in that file. If
1779 is absent then the default it to allow any device. Similar if
1781 contains the special word
1783 then any device is allowed. Otherwise the device name given to
1785 must match one of the names or patterns in a
1790 Does the device have a valid md superblock. If a specific metadata
1791 version is request with
1795 then only that style of metadata is accepted, otherwise
1797 finds any known version of metadata. If no
1799 metadata is found, the device is rejected.
1802 Does the metadata match an expected array?
1803 The metadata can match in two ways. Either there is an array listed
1806 which identifies the array (either by UUID, by name, by device list,
1807 or by minor-number), the array was created with a
1811 matches that which is given in
1813 or on the command line.
1816 is not able to positively identify the array as belonging to the
1817 current host, the device will be rejected.
1821 keeps a list of arrays that is has partly assembled in
1822 .B /var/run/mdadm/map
1824 .B /var/run/mdadm.map
1825 if the directory doesn't exist). If no array exists which matches
1826 the metadata on the new device,
1828 must choose a device name and unit number. It does this based on any
1831 or any name information stored in the metadata. If this name
1832 suggests a unit number, that number will be used, otherwise a free
1833 unit number will be chosen. Normally
1835 will prefer to create a partitionable array, however if the
1839 suggests that a non-partitionable array is preferred, that will be
1843 Once an appropriate array is found or created and the device is added,
1845 must decide if the array is ready to be started. It will
1846 normally compare the number of available (non-spare) devices to the
1847 number of devices that the metadata suggests need to be active. If
1848 there are at least that many, the array will be started. This means
1849 that if any devices are missing the array will not be restarted.
1855 in which case the array will be run as soon as there are enough
1856 devices present for the data to be accessible. For a raid1, that
1857 means one device will start the array. For a clean raid5, the array
1858 will be started as soon as all but one drive is present.
1860 Note that neither of these approaches is really ideal. If it is can
1861 be known that all device discovery has completed, then
1865 can be run which will try to start all arrays that are being
1866 incrementally assembled. They are started in "read-auto" mode in
1867 which they are read-only until the first write request. This means
1868 that no metadata updates are made and no attempt at resync or recovery
1869 happens. Further devices that are found before the first write can
1870 still be added safely.
1874 .B " mdadm --query /dev/name-of-device"
1876 This will find out if a given device is a raid array, or is part of
1877 one, and will provide brief information about the device.
1879 .B " mdadm --assemble --scan"
1881 This will assemble and start all arrays listed in the standard config file
1882 file. This command will typically go in a system startup file.
1884 .B " mdadm --stop --scan"
1886 This will shut down all array that can be shut down (i.e. are not
1887 currently in use). This will typically go in a system shutdown script.
1889 .B " mdadm --follow --scan --delay=120"
1891 If (and only if) there is an Email address or program given in the
1892 standard config file, then
1893 monitor the status of all arrays listed in that file by
1894 polling them ever 2 minutes.
1896 .B " mdadm --create /dev/md0 --level=1 --raid-devices=2 /dev/hd[ac]1"
1898 Create /dev/md0 as a RAID1 array consisting of /dev/hda1 and /dev/hdc1.
1901 .B " echo 'DEVICE /dev/hd*[0-9] /dev/sd*[0-9]' > mdadm.conf"
1903 .B " mdadm --detail --scan >> mdadm.conf"
1905 This will create a prototype config file that describes currently
1906 active arrays that are known to be made from partitions of IDE or SCSI drives.
1907 This file should be reviewed before being used as it may
1908 contain unwanted detail.
1910 .B " echo 'DEVICE /dev/hd[a-z] /dev/sd*[a-z]' > mdadm.conf"
1912 .B " mdadm --examine --scan --config=mdadm.conf >> mdadm.conf"
1914 This will find what arrays could be assembled from existing IDE and
1915 SCSI whole drives (not partitions) and store the information is the
1916 format of a config file.
1917 This file is very likely to contain unwanted detail, particularly
1920 entries. It should be reviewed and edited before being used as an
1923 .B " mdadm --examine --brief --scan --config=partitions"
1925 .B " mdadm -Ebsc partitions"
1927 Create a list of devices by reading
1928 .BR /proc/partitions ,
1929 scan these for RAID superblocks, and printout a brief listing of all
1932 .B " mdadm -Ac partitions -m 0 /dev/md0"
1934 Scan all partitions and devices listed in
1935 .BR /proc/partitions
1938 out of all such devices with a RAID superblock with a minor number of 0.
1940 .B " mdadm --monitor --scan --daemonise > /var/run/mdadm"
1942 If config file contains a mail address or alert program, run mdadm in
1943 the background in monitor mode monitoring all md devices. Also write
1944 pid of mdadm daemon to
1945 .BR /var/run/mdadm .
1947 .B " mdadm -Iq /dev/somedevice"
1949 Try to incorporate newly discovered device into some array as
1952 .B " mdadm --incremental --rebuild --run --scan"
1954 Rebuild the array map from any current arrays, and then start any that
1957 .B " mdadm --create --help"
1959 Provide help about the Create mode.
1961 .B " mdadm --config --help"
1963 Provide help about the format of the config file.
1967 Provide general help.
1978 lists all active md devices with information about them.
1980 uses this to find arrays when
1982 is given in Misc mode, and to monitor array reconstruction
1988 The config file lists which devices may be scanned to see if
1989 they contain MD super block, and gives identifying information
1990 (e.g. UUID) about known MD arrays. See
1994 .SS /var/run/mdadm/map
1997 mode is used. this file gets a list of arrays currently being created.
2000 does not exist as a directory, then
2001 .B /var/run/mdadm.map
2006 While entries in the /dev directory can have any format you like,
2008 has an understanding of 'standard' formats which it uses to guide its
2009 behaviour when creating device files via the
2013 The standard names for non-partitioned arrays (the only sort of md
2014 array available in 2.4 and earlier) either of
2020 where NN is a number.
2021 The standard names for partitionable arrays (as available from 2.6
2028 Partition numbers should be indicated by added "pMM" to these, thus "/dev/md/d1p2".
2032 was previously known as
2036 is completely separate from the
2038 package, and does not use the
2040 configuration file at all.
2043 For information on the various levels of
2047 .UR http://ostenfeld.dk/~jakob/Software-RAID.HOWTO/
2048 http://ostenfeld.dk/~jakob/Software-RAID.HOWTO/
2051 '''for new releases of the RAID driver check out:
2054 '''.UR ftp://ftp.kernel.org/pub/linux/kernel/people/mingo/raid-patches
2055 '''ftp://ftp.kernel.org/pub/linux/kernel/people/mingo/raid-patches
2060 '''.UR http://www.cse.unsw.edu.au/~neilb/patches/linux-stable/
2061 '''http://www.cse.unsw.edu.au/~neilb/patches/linux-stable/
2064 The latest version of
2066 should always be available from
2068 .UR http://www.kernel.org/pub/linux/utils/raid/mdadm/
2069 http://www.kernel.org/pub/linux/utils/raid/mdadm/