md.rst 27 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765
  1. RAID arrays
  2. ===========
  3. Boot time assembly of RAID arrays
  4. ---------------------------------
  5. Tools that manage md devices can be found at
  6. https://www.kernel.org/pub/linux/utils/raid/
  7. You can boot with your md device with the following kernel command
  8. lines:
  9. for old raid arrays without persistent superblocks::
  10. md=<md device no.>,<raid level>,<chunk size factor>,<fault level>,dev0,dev1,...,devn
  11. for raid arrays with persistent superblocks::
  12. md=<md device no.>,dev0,dev1,...,devn
  13. or, to assemble a partitionable array::
  14. md=d<md device no.>,dev0,dev1,...,devn
  15. ``md device no.``
  16. +++++++++++++++++
  17. The number of the md device
  18. ================= =========
  19. ``md device no.`` device
  20. ================= =========
  21. 0 md0
  22. 1 md1
  23. 2 md2
  24. 3 md3
  25. 4 md4
  26. ================= =========
  27. ``raid level``
  28. ++++++++++++++
  29. level of the RAID array
  30. =============== =============
  31. ``raid level`` level
  32. =============== =============
  33. -1 linear mode
  34. 0 striped mode
  35. =============== =============
  36. other modes are only supported with persistent super blocks
  37. ``chunk size factor``
  38. +++++++++++++++++++++
  39. (raid-0 and raid-1 only)
  40. Set the chunk size as 4k << n.
  41. ``fault level``
  42. +++++++++++++++
  43. Totally ignored
  44. ``dev0`` to ``devn``
  45. ++++++++++++++++++++
  46. e.g. ``/dev/hda1``, ``/dev/hdc1``, ``/dev/sda1``, ``/dev/sdb1``
  47. A possible loadlin line (Harald Hoyer <HarryH@Royal.Net>) looks like this::
  48. e:\loadlin\loadlin e:\zimage root=/dev/md0 md=0,0,4,0,/dev/hdb2,/dev/hdc3 ro
  49. Boot time autodetection of RAID arrays
  50. --------------------------------------
  51. When md is compiled into the kernel (not as module), partitions of
  52. type 0xfd are scanned and automatically assembled into RAID arrays.
  53. This autodetection may be suppressed with the kernel parameter
  54. ``raid=noautodetect``. As of kernel 2.6.9, only drives with a type 0
  55. superblock can be autodetected and run at boot time.
  56. The kernel parameter ``raid=partitionable`` (or ``raid=part``) means
  57. that all auto-detected arrays are assembled as partitionable.
  58. Boot time assembly of degraded/dirty arrays
  59. -------------------------------------------
  60. If a raid5 or raid6 array is both dirty and degraded, it could have
  61. undetectable data corruption. This is because the fact that it is
  62. ``dirty`` means that the parity cannot be trusted, and the fact that it
  63. is degraded means that some datablocks are missing and cannot reliably
  64. be reconstructed (due to no parity).
  65. For this reason, md will normally refuse to start such an array. This
  66. requires the sysadmin to take action to explicitly start the array
  67. despite possible corruption. This is normally done with::
  68. mdadm --assemble --force ....
  69. This option is not really available if the array has the root
  70. filesystem on it. In order to support this booting from such an
  71. array, md supports a module parameter ``start_dirty_degraded`` which,
  72. when set to 1, bypassed the checks and will allows dirty degraded
  73. arrays to be started.
  74. So, to boot with a root filesystem of a dirty degraded raid 5 or 6, use::
  75. md-mod.start_dirty_degraded=1
  76. Superblock formats
  77. ------------------
  78. The md driver can support a variety of different superblock formats.
  79. Currently, it supports superblock formats ``0.90.0`` and the ``md-1`` format
  80. introduced in the 2.5 development series.
  81. The kernel will autodetect which format superblock is being used.
  82. Superblock format ``0`` is treated differently to others for legacy
  83. reasons - it is the original superblock format.
  84. General Rules - apply for all superblock formats
  85. ------------------------------------------------
  86. An array is ``created`` by writing appropriate superblocks to all
  87. devices.
  88. It is ``assembled`` by associating each of these devices with an
  89. particular md virtual device. Once it is completely assembled, it can
  90. be accessed.
  91. An array should be created by a user-space tool. This will write
  92. superblocks to all devices. It will usually mark the array as
  93. ``unclean``, or with some devices missing so that the kernel md driver
  94. can create appropriate redundancy (copying in raid 1, parity
  95. calculation in raid 4/5).
  96. When an array is assembled, it is first initialized with the
  97. SET_ARRAY_INFO ioctl. This contains, in particular, a major and minor
  98. version number. The major version number selects which superblock
  99. format is to be used. The minor number might be used to tune handling
  100. of the format, such as suggesting where on each device to look for the
  101. superblock.
  102. Then each device is added using the ADD_NEW_DISK ioctl. This
  103. provides, in particular, a major and minor number identifying the
  104. device to add.
  105. The array is started with the RUN_ARRAY ioctl.
  106. Once started, new devices can be added. They should have an
  107. appropriate superblock written to them, and then be passed in with
  108. ADD_NEW_DISK.
  109. Devices that have failed or are not yet active can be detached from an
  110. array using HOT_REMOVE_DISK.
  111. Specific Rules that apply to format-0 super block arrays, and arrays with no superblock (non-persistent)
  112. --------------------------------------------------------------------------------------------------------
  113. An array can be ``created`` by describing the array (level, chunksize
  114. etc) in a SET_ARRAY_INFO ioctl. This must have ``major_version==0`` and
  115. ``raid_disks != 0``.
  116. Then uninitialized devices can be added with ADD_NEW_DISK. The
  117. structure passed to ADD_NEW_DISK must specify the state of the device
  118. and its role in the array.
  119. Once started with RUN_ARRAY, uninitialized spares can be added with
  120. HOT_ADD_DISK.
  121. MD devices in sysfs
  122. -------------------
  123. md devices appear in sysfs (``/sys``) as regular block devices,
  124. e.g.::
  125. /sys/block/md0
  126. Each ``md`` device will contain a subdirectory called ``md`` which
  127. contains further md-specific information about the device.
  128. All md devices contain:
  129. level
  130. a text file indicating the ``raid level``. e.g. raid0, raid1,
  131. raid5, linear, multipath, faulty.
  132. If no raid level has been set yet (array is still being
  133. assembled), the value will reflect whatever has been written
  134. to it, which may be a name like the above, or may be a number
  135. such as ``0``, ``5``, etc.
  136. raid_disks
  137. a text file with a simple number indicating the number of devices
  138. in a fully functional array. If this is not yet known, the file
  139. will be empty. If an array is being resized this will contain
  140. the new number of devices.
  141. Some raid levels allow this value to be set while the array is
  142. active. This will reconfigure the array. Otherwise it can only
  143. be set while assembling an array.
  144. A change to this attribute will not be permitted if it would
  145. reduce the size of the array. To reduce the number of drives
  146. in an e.g. raid5, the array size must first be reduced by
  147. setting the ``array_size`` attribute.
  148. chunk_size
  149. This is the size in bytes for ``chunks`` and is only relevant to
  150. raid levels that involve striping (0,4,5,6,10). The address space
  151. of the array is conceptually divided into chunks and consecutive
  152. chunks are striped onto neighbouring devices.
  153. The size should be at least PAGE_SIZE (4k) and should be a power
  154. of 2. This can only be set while assembling an array
  155. layout
  156. The ``layout`` for the array for the particular level. This is
  157. simply a number that is interpretted differently by different
  158. levels. It can be written while assembling an array.
  159. array_size
  160. This can be used to artificially constrain the available space in
  161. the array to be less than is actually available on the combined
  162. devices. Writing a number (in Kilobytes) which is less than
  163. the available size will set the size. Any reconfiguration of the
  164. array (e.g. adding devices) will not cause the size to change.
  165. Writing the word ``default`` will cause the effective size of the
  166. array to be whatever size is actually available based on
  167. ``level``, ``chunk_size`` and ``component_size``.
  168. This can be used to reduce the size of the array before reducing
  169. the number of devices in a raid4/5/6, or to support external
  170. metadata formats which mandate such clipping.
  171. reshape_position
  172. This is either ``none`` or a sector number within the devices of
  173. the array where ``reshape`` is up to. If this is set, the three
  174. attributes mentioned above (raid_disks, chunk_size, layout) can
  175. potentially have 2 values, an old and a new value. If these
  176. values differ, reading the attribute returns::
  177. new (old)
  178. and writing will effect the ``new`` value, leaving the ``old``
  179. unchanged.
  180. component_size
  181. For arrays with data redundancy (i.e. not raid0, linear, faulty,
  182. multipath), all components must be the same size - or at least
  183. there must a size that they all provide space for. This is a key
  184. part or the geometry of the array. It is measured in sectors
  185. and can be read from here. Writing to this value may resize
  186. the array if the personality supports it (raid1, raid5, raid6),
  187. and if the component drives are large enough.
  188. metadata_version
  189. This indicates the format that is being used to record metadata
  190. about the array. It can be 0.90 (traditional format), 1.0, 1.1,
  191. 1.2 (newer format in varying locations) or ``none`` indicating that
  192. the kernel isn't managing metadata at all.
  193. Alternately it can be ``external:`` followed by a string which
  194. is set by user-space. This indicates that metadata is managed
  195. by a user-space program. Any device failure or other event that
  196. requires a metadata update will cause array activity to be
  197. suspended until the event is acknowledged.
  198. resync_start
  199. The point at which resync should start. If no resync is needed,
  200. this will be a very large number (or ``none`` since 2.6.30-rc1). At
  201. array creation it will default to 0, though starting the array as
  202. ``clean`` will set it much larger.
  203. new_dev
  204. This file can be written but not read. The value written should
  205. be a block device number as major:minor. e.g. 8:0
  206. This will cause that device to be attached to the array, if it is
  207. available. It will then appear at md/dev-XXX (depending on the
  208. name of the device) and further configuration is then possible.
  209. safe_mode_delay
  210. When an md array has seen no write requests for a certain period
  211. of time, it will be marked as ``clean``. When another write
  212. request arrives, the array is marked as ``dirty`` before the write
  213. commences. This is known as ``safe_mode``.
  214. The ``certain period`` is controlled by this file which stores the
  215. period as a number of seconds. The default is 200msec (0.200).
  216. Writing a value of 0 disables safemode.
  217. array_state
  218. This file contains a single word which describes the current
  219. state of the array. In many cases, the state can be set by
  220. writing the word for the desired state, however some states
  221. cannot be explicitly set, and some transitions are not allowed.
  222. Select/poll works on this file. All changes except between
  223. Active_idle and active (which can be frequent and are not
  224. very interesting) are notified. active->active_idle is
  225. reported if the metadata is externally managed.
  226. clear
  227. No devices, no size, no level
  228. Writing is equivalent to STOP_ARRAY ioctl
  229. inactive
  230. May have some settings, but array is not active
  231. all IO results in error
  232. When written, doesn't tear down array, but just stops it
  233. suspended (not supported yet)
  234. All IO requests will block. The array can be reconfigured.
  235. Writing this, if accepted, will block until array is quiessent
  236. readonly
  237. no resync can happen. no superblocks get written.
  238. Write requests fail
  239. read-auto
  240. like readonly, but behaves like ``clean`` on a write request.
  241. clean
  242. no pending writes, but otherwise active.
  243. When written to inactive array, starts without resync
  244. If a write request arrives then
  245. if metadata is known, mark ``dirty`` and switch to ``active``.
  246. if not known, block and switch to write-pending
  247. If written to an active array that has pending writes, then fails.
  248. active
  249. fully active: IO and resync can be happening.
  250. When written to inactive array, starts with resync
  251. write-pending
  252. clean, but writes are blocked waiting for ``active`` to be written.
  253. active-idle
  254. like active, but no writes have been seen for a while (safe_mode_delay).
  255. bitmap/location
  256. This indicates where the write-intent bitmap for the array is
  257. stored.
  258. It can be one of ``none``, ``file`` or ``[+-]N``.
  259. ``file`` may later be extended to ``file:/file/name``
  260. ``[+-]N`` means that many sectors from the start of the metadata.
  261. This is replicated on all devices. For arrays with externally
  262. managed metadata, the offset is from the beginning of the
  263. device.
  264. bitmap/chunksize
  265. The size, in bytes, of the chunk which will be represented by a
  266. single bit. For RAID456, it is a portion of an individual
  267. device. For RAID10, it is a portion of the array. For RAID1, it
  268. is both (they come to the same thing).
  269. bitmap/time_base
  270. The time, in seconds, between looking for bits in the bitmap to
  271. be cleared. In the current implementation, a bit will be cleared
  272. between 2 and 3 times ``time_base`` after all the covered blocks
  273. are known to be in-sync.
  274. bitmap/backlog
  275. When write-mostly devices are active in a RAID1, write requests
  276. to those devices proceed in the background - the filesystem (or
  277. other user of the device) does not have to wait for them.
  278. ``backlog`` sets a limit on the number of concurrent background
  279. writes. If there are more than this, new writes will by
  280. synchronous.
  281. bitmap/metadata
  282. This can be either ``internal`` or ``external``.
  283. ``internal``
  284. is the default and means the metadata for the bitmap
  285. is stored in the first 256 bytes of the allocated space and is
  286. managed by the md module.
  287. ``external``
  288. means that bitmap metadata is managed externally to
  289. the kernel (i.e. by some userspace program)
  290. bitmap/can_clear
  291. This is either ``true`` or ``false``. If ``true``, then bits in the
  292. bitmap will be cleared when the corresponding blocks are thought
  293. to be in-sync. If ``false``, bits will never be cleared.
  294. This is automatically set to ``false`` if a write happens on a
  295. degraded array, or if the array becomes degraded during a write.
  296. When metadata is managed externally, it should be set to true
  297. once the array becomes non-degraded, and this fact has been
  298. recorded in the metadata.
  299. consistency_policy
  300. This indicates how the array maintains consistency in case of unexpected
  301. shutdown. It can be:
  302. none
  303. Array has no redundancy information, e.g. raid0, linear.
  304. resync
  305. Full resync is performed and all redundancy is regenerated when the
  306. array is started after unclean shutdown.
  307. bitmap
  308. Resync assisted by a write-intent bitmap.
  309. journal
  310. For raid4/5/6, journal device is used to log transactions and replay
  311. after unclean shutdown.
  312. ppl
  313. For raid5 only, Partial Parity Log is used to close the write hole and
  314. eliminate resync.
  315. The accepted values when writing to this file are ``ppl`` and ``resync``,
  316. used to enable and disable PPL.
  317. uuid
  318. This indicates the UUID of the array in the following format:
  319. xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx
  320. As component devices are added to an md array, they appear in the ``md``
  321. directory as new directories named::
  322. dev-XXX
  323. where ``XXX`` is a name that the kernel knows for the device, e.g. hdb1.
  324. Each directory contains:
  325. block
  326. a symlink to the block device in /sys/block, e.g.::
  327. /sys/block/md0/md/dev-hdb1/block -> ../../../../block/hdb/hdb1
  328. super
  329. A file containing an image of the superblock read from, or
  330. written to, that device.
  331. state
  332. A file recording the current state of the device in the array
  333. which can be a comma separated list of:
  334. faulty
  335. device has been kicked from active use due to
  336. a detected fault, or it has unacknowledged bad
  337. blocks
  338. in_sync
  339. device is a fully in-sync member of the array
  340. writemostly
  341. device will only be subject to read
  342. requests if there are no other options.
  343. This applies only to raid1 arrays.
  344. blocked
  345. device has failed, and the failure hasn't been
  346. acknowledged yet by the metadata handler.
  347. Writes that would write to this device if
  348. it were not faulty are blocked.
  349. spare
  350. device is working, but not a full member.
  351. This includes spares that are in the process
  352. of being recovered to
  353. write_error
  354. device has ever seen a write error.
  355. want_replacement
  356. device is (mostly) working but probably
  357. should be replaced, either due to errors or
  358. due to user request.
  359. replacement
  360. device is a replacement for another active
  361. device with same raid_disk.
  362. This list may grow in future.
  363. This can be written to.
  364. Writing ``faulty`` simulates a failure on the device.
  365. Writing ``remove`` removes the device from the array.
  366. Writing ``writemostly`` sets the writemostly flag.
  367. Writing ``-writemostly`` clears the writemostly flag.
  368. Writing ``blocked`` sets the ``blocked`` flag.
  369. Writing ``-blocked`` clears the ``blocked`` flags and allows writes
  370. to complete and possibly simulates an error.
  371. Writing ``in_sync`` sets the in_sync flag.
  372. Writing ``write_error`` sets writeerrorseen flag.
  373. Writing ``-write_error`` clears writeerrorseen flag.
  374. Writing ``want_replacement`` is allowed at any time except to a
  375. replacement device or a spare. It sets the flag.
  376. Writing ``-want_replacement`` is allowed at any time. It clears
  377. the flag.
  378. Writing ``replacement`` or ``-replacement`` is only allowed before
  379. starting the array. It sets or clears the flag.
  380. This file responds to select/poll. Any change to ``faulty``
  381. or ``blocked`` causes an event.
  382. errors
  383. An approximate count of read errors that have been detected on
  384. this device but have not caused the device to be evicted from
  385. the array (either because they were corrected or because they
  386. happened while the array was read-only). When using version-1
  387. metadata, this value persists across restarts of the array.
  388. This value can be written while assembling an array thus
  389. providing an ongoing count for arrays with metadata managed by
  390. userspace.
  391. slot
  392. This gives the role that the device has in the array. It will
  393. either be ``none`` if the device is not active in the array
  394. (i.e. is a spare or has failed) or an integer less than the
  395. ``raid_disks`` number for the array indicating which position
  396. it currently fills. This can only be set while assembling an
  397. array. A device for which this is set is assumed to be working.
  398. offset
  399. This gives the location in the device (in sectors from the
  400. start) where data from the array will be stored. Any part of
  401. the device before this offset is not touched, unless it is
  402. used for storing metadata (Formats 1.1 and 1.2).
  403. size
  404. The amount of the device, after the offset, that can be used
  405. for storage of data. This will normally be the same as the
  406. component_size. This can be written while assembling an
  407. array. If a value less than the current component_size is
  408. written, it will be rejected.
  409. recovery_start
  410. When the device is not ``in_sync``, this records the number of
  411. sectors from the start of the device which are known to be
  412. correct. This is normally zero, but during a recovery
  413. operation it will steadily increase, and if the recovery is
  414. interrupted, restoring this value can cause recovery to
  415. avoid repeating the earlier blocks. With v1.x metadata, this
  416. value is saved and restored automatically.
  417. This can be set whenever the device is not an active member of
  418. the array, either before the array is activated, or before
  419. the ``slot`` is set.
  420. Setting this to ``none`` is equivalent to setting ``in_sync``.
  421. Setting to any other value also clears the ``in_sync`` flag.
  422. bad_blocks
  423. This gives the list of all known bad blocks in the form of
  424. start address and length (in sectors respectively). If output
  425. is too big to fit in a page, it will be truncated. Writing
  426. ``sector length`` to this file adds new acknowledged (i.e.
  427. recorded to disk safely) bad blocks.
  428. unacknowledged_bad_blocks
  429. This gives the list of known-but-not-yet-saved-to-disk bad
  430. blocks in the same form of ``bad_blocks``. If output is too big
  431. to fit in a page, it will be truncated. Writing to this file
  432. adds bad blocks without acknowledging them. This is largely
  433. for testing.
  434. ppl_sector, ppl_size
  435. Location and size (in sectors) of the space used for Partial Parity Log
  436. on this device.
  437. An active md device will also contain an entry for each active device
  438. in the array. These are named::
  439. rdNN
  440. where ``NN`` is the position in the array, starting from 0.
  441. So for a 3 drive array there will be rd0, rd1, rd2.
  442. These are symbolic links to the appropriate ``dev-XXX`` entry.
  443. Thus, for example::
  444. cat /sys/block/md*/md/rd*/state
  445. will show ``in_sync`` on every line.
  446. Active md devices for levels that support data redundancy (1,4,5,6,10)
  447. also have
  448. sync_action
  449. a text file that can be used to monitor and control the rebuild
  450. process. It contains one word which can be one of:
  451. resync
  452. redundancy is being recalculated after unclean
  453. shutdown or creation
  454. recover
  455. a hot spare is being built to replace a
  456. failed/missing device
  457. idle
  458. nothing is happening
  459. check
  460. A full check of redundancy was requested and is
  461. happening. This reads all blocks and checks
  462. them. A repair may also happen for some raid
  463. levels.
  464. repair
  465. A full check and repair is happening. This is
  466. similar to ``resync``, but was requested by the
  467. user, and the write-intent bitmap is NOT used to
  468. optimise the process.
  469. This file is writable, and each of the strings that could be
  470. read are meaningful for writing.
  471. ``idle`` will stop an active resync/recovery etc. There is no
  472. guarantee that another resync/recovery may not be automatically
  473. started again, though some event will be needed to trigger
  474. this.
  475. ``resync`` or ``recovery`` can be used to restart the
  476. corresponding operation if it was stopped with ``idle``.
  477. ``check`` and ``repair`` will start the appropriate process
  478. providing the current state is ``idle``.
  479. This file responds to select/poll. Any important change in the value
  480. triggers a poll event. Sometimes the value will briefly be
  481. ``recover`` if a recovery seems to be needed, but cannot be
  482. achieved. In that case, the transition to ``recover`` isn't
  483. notified, but the transition away is.
  484. degraded
  485. This contains a count of the number of devices by which the
  486. arrays is degraded. So an optimal array will show ``0``. A
  487. single failed/missing drive will show ``1``, etc.
  488. This file responds to select/poll, any increase or decrease
  489. in the count of missing devices will trigger an event.
  490. mismatch_count
  491. When performing ``check`` and ``repair``, and possibly when
  492. performing ``resync``, md will count the number of errors that are
  493. found. The count in ``mismatch_cnt`` is the number of sectors
  494. that were re-written, or (for ``check``) would have been
  495. re-written. As most raid levels work in units of pages rather
  496. than sectors, this may be larger than the number of actual errors
  497. by a factor of the number of sectors in a page.
  498. bitmap_set_bits
  499. If the array has a write-intent bitmap, then writing to this
  500. attribute can set bits in the bitmap, indicating that a resync
  501. would need to check the corresponding blocks. Either individual
  502. numbers or start-end pairs can be written. Multiple numbers
  503. can be separated by a space.
  504. Note that the numbers are ``bit`` numbers, not ``block`` numbers.
  505. They should be scaled by the bitmap_chunksize.
  506. sync_speed_min, sync_speed_max
  507. This are similar to ``/proc/sys/dev/raid/speed_limit_{min,max}``
  508. however they only apply to the particular array.
  509. If no value has been written to these, or if the word ``system``
  510. is written, then the system-wide value is used. If a value,
  511. in kibibytes-per-second is written, then it is used.
  512. When the files are read, they show the currently active value
  513. followed by ``(local)`` or ``(system)`` depending on whether it is
  514. a locally set or system-wide value.
  515. sync_completed
  516. This shows the number of sectors that have been completed of
  517. whatever the current sync_action is, followed by the number of
  518. sectors in total that could need to be processed. The two
  519. numbers are separated by a ``/`` thus effectively showing one
  520. value, a fraction of the process that is complete.
  521. A ``select`` on this attribute will return when resync completes,
  522. when it reaches the current sync_max (below) and possibly at
  523. other times.
  524. sync_speed
  525. This shows the current actual speed, in K/sec, of the current
  526. sync_action. It is averaged over the last 30 seconds.
  527. suspend_lo, suspend_hi
  528. The two values, given as numbers of sectors, indicate a range
  529. within the array where IO will be blocked. This is currently
  530. only supported for raid4/5/6.
  531. sync_min, sync_max
  532. The two values, given as numbers of sectors, indicate a range
  533. within the array where ``check``/``repair`` will operate. Must be
  534. a multiple of chunk_size. When it reaches ``sync_max`` it will
  535. pause, rather than complete.
  536. You can use ``select`` or ``poll`` on ``sync_completed`` to wait for
  537. that number to reach sync_max. Then you can either increase
  538. ``sync_max``, or can write ``idle`` to ``sync_action``.
  539. The value of ``max`` for ``sync_max`` effectively disables the limit.
  540. When a resync is active, the value can only ever be increased,
  541. never decreased.
  542. The value of ``0`` is the minimum for ``sync_min``.
  543. Each active md device may also have attributes specific to the
  544. personality module that manages it.
  545. These are specific to the implementation of the module and could
  546. change substantially if the implementation changes.
  547. These currently include:
  548. stripe_cache_size (currently raid5 only)
  549. number of entries in the stripe cache. This is writable, but
  550. there are upper and lower limits (32768, 17). Default is 256.
  551. strip_cache_active (currently raid5 only)
  552. number of active entries in the stripe cache
  553. preread_bypass_threshold (currently raid5 only)
  554. number of times a stripe requiring preread will be bypassed by
  555. a stripe that does not require preread. For fairness defaults
  556. to 1. Setting this to 0 disables bypass accounting and
  557. requires preread stripes to wait until all full-width stripe-
  558. writes are complete. Valid values are 0 to stripe_cache_size.
  559. journal_mode (currently raid5 only)
  560. The cache mode for raid5. raid5 could include an extra disk for
  561. caching. The mode can be "write-throuth" and "write-back". The
  562. default is "write-through".
  563. ppl_write_hint
  564. NVMe stream ID to be set for each PPL write request.