bcache.rst 24 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656
  1. ============================
  2. A block layer cache (bcache)
  3. ============================
  4. Say you've got a big slow raid 6, and an ssd or three. Wouldn't it be
  5. nice if you could use them as cache... Hence bcache.
  6. The bcache wiki can be found at:
  7. https://bcache.evilpiepirate.org
  8. This is the git repository of bcache-tools:
  9. https://git.kernel.org/pub/scm/linux/kernel/git/colyli/bcache-tools.git/
  10. The latest bcache kernel code can be found from mainline Linux kernel:
  11. https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/
  12. It's designed around the performance characteristics of SSDs - it only allocates
  13. in erase block sized buckets, and it uses a hybrid btree/log to track cached
  14. extents (which can be anywhere from a single sector to the bucket size). It's
  15. designed to avoid random writes at all costs; it fills up an erase block
  16. sequentially, then issues a discard before reusing it.
  17. Both writethrough and writeback caching are supported. Writeback defaults to
  18. off, but can be switched on and off arbitrarily at runtime. Bcache goes to
  19. great lengths to protect your data - it reliably handles unclean shutdown. (It
  20. doesn't even have a notion of a clean shutdown; bcache simply doesn't return
  21. writes as completed until they're on stable storage).
  22. Writeback caching can use most of the cache for buffering writes - writing
  23. dirty data to the backing device is always done sequentially, scanning from the
  24. start to the end of the index.
  25. Since random IO is what SSDs excel at, there generally won't be much benefit
  26. to caching large sequential IO. Bcache detects sequential IO and skips it;
  27. it also keeps a rolling average of the IO sizes per task, and as long as the
  28. average is above the cutoff it will skip all IO from that task - instead of
  29. caching the first 512k after every seek. Backups and large file copies should
  30. thus entirely bypass the cache.
  31. In the event of a data IO error on the flash it will try to recover by reading
  32. from disk or invalidating cache entries. For unrecoverable errors (meta data
  33. or dirty data), caching is automatically disabled; if dirty data was present
  34. in the cache it first disables writeback caching and waits for all dirty data
  35. to be flushed.
  36. Getting started:
  37. You'll need bcache util from the bcache-tools repository. Both the cache device
  38. and backing device must be formatted before use::
  39. bcache make -B /dev/sdb
  40. bcache make -C /dev/sdc
  41. `bcache make` has the ability to format multiple devices at the same time - if
  42. you format your backing devices and cache device at the same time, you won't
  43. have to manually attach::
  44. bcache make -B /dev/sda /dev/sdb -C /dev/sdc
  45. If your bcache-tools is not updated to latest version and does not have the
  46. unified `bcache` utility, you may use the legacy `make-bcache` utility to format
  47. bcache device with same -B and -C parameters.
  48. bcache-tools now ships udev rules, and bcache devices are known to the kernel
  49. immediately. Without udev, you can manually register devices like this::
  50. echo /dev/sdb > /sys/fs/bcache/register
  51. echo /dev/sdc > /sys/fs/bcache/register
  52. Registering the backing device makes the bcache device show up in /dev; you can
  53. now format it and use it as normal. But the first time using a new bcache
  54. device, it'll be running in passthrough mode until you attach it to a cache.
  55. If you are thinking about using bcache later, it is recommended to setup all your
  56. slow devices as bcache backing devices without a cache, and you can choose to add
  57. a caching device later.
  58. See 'ATTACHING' section below.
  59. The devices show up as::
  60. /dev/bcache<N>
  61. As well as (with udev)::
  62. /dev/bcache/by-uuid/<uuid>
  63. /dev/bcache/by-label/<label>
  64. To get started::
  65. mkfs.ext4 /dev/bcache0
  66. mount /dev/bcache0 /mnt
  67. You can control bcache devices through sysfs at /sys/block/bcache<N>/bcache .
  68. You can also control them through /sys/fs//bcache/<cset-uuid>/ .
  69. Cache devices are managed as sets; multiple caches per set isn't supported yet
  70. but will allow for mirroring of metadata and dirty data in the future. Your new
  71. cache set shows up as /sys/fs/bcache/<UUID>
  72. Attaching
  73. ---------
  74. After your cache device and backing device are registered, the backing device
  75. must be attached to your cache set to enable caching. Attaching a backing
  76. device to a cache set is done thusly, with the UUID of the cache set in
  77. /sys/fs/bcache::
  78. echo <CSET-UUID> > /sys/block/bcache0/bcache/attach
  79. This only has to be done once. The next time you reboot, just reregister all
  80. your bcache devices. If a backing device has data in a cache somewhere, the
  81. /dev/bcache<N> device won't be created until the cache shows up - particularly
  82. important if you have writeback caching turned on.
  83. If you're booting up and your cache device is gone and never coming back, you
  84. can force run the backing device::
  85. echo 1 > /sys/block/sdb/bcache/running
  86. (You need to use /sys/block/sdb (or whatever your backing device is called), not
  87. /sys/block/bcache0, because bcache0 doesn't exist yet. If you're using a
  88. partition, the bcache directory would be at /sys/block/sdb/sdb2/bcache)
  89. The backing device will still use that cache set if it shows up in the future,
  90. but all the cached data will be invalidated. If there was dirty data in the
  91. cache, don't expect the filesystem to be recoverable - you will have massive
  92. filesystem corruption, though ext4's fsck does work miracles.
  93. Error Handling
  94. --------------
  95. Bcache tries to transparently handle IO errors to/from the cache device without
  96. affecting normal operation; if it sees too many errors (the threshold is
  97. configurable, and defaults to 0) it shuts down the cache device and switches all
  98. the backing devices to passthrough mode.
  99. - For reads from the cache, if they error we just retry the read from the
  100. backing device.
  101. - For writethrough writes, if the write to the cache errors we just switch to
  102. invalidating the data at that lba in the cache (i.e. the same thing we do for
  103. a write that bypasses the cache)
  104. - For writeback writes, we currently pass that error back up to the
  105. filesystem/userspace. This could be improved - we could retry it as a write
  106. that skips the cache so we don't have to error the write.
  107. - When we detach, we first try to flush any dirty data (if we were running in
  108. writeback mode). It currently doesn't do anything intelligent if it fails to
  109. read some of the dirty data, though.
  110. Howto/cookbook
  111. --------------
  112. A) Starting a bcache with a missing caching device
  113. If registering the backing device doesn't help, it's already there, you just need
  114. to force it to run without the cache::
  115. host:~# echo /dev/sdb1 > /sys/fs/bcache/register
  116. [ 119.844831] bcache: register_bcache() error opening /dev/sdb1: device already registered
  117. Next, you try to register your caching device if it's present. However
  118. if it's absent, or registration fails for some reason, you can still
  119. start your bcache without its cache, like so::
  120. host:/sys/block/sdb/sdb1/bcache# echo 1 > running
  121. Note that this may cause data loss if you were running in writeback mode.
  122. B) Bcache does not find its cache::
  123. host:/sys/block/md5/bcache# echo 0226553a-37cf-41d5-b3ce-8b1e944543a8 > attach
  124. [ 1933.455082] bcache: bch_cached_dev_attach() Couldn't find uuid for md5 in set
  125. [ 1933.478179] bcache: __cached_dev_store() Can't attach 0226553a-37cf-41d5-b3ce-8b1e944543a8
  126. [ 1933.478179] : cache set not found
  127. In this case, the caching device was simply not registered at boot
  128. or disappeared and came back, and needs to be (re-)registered::
  129. host:/sys/block/md5/bcache# echo /dev/sdh2 > /sys/fs/bcache/register
  130. C) Corrupt bcache crashes the kernel at device registration time:
  131. This should never happen. If it does happen, then you have found a bug!
  132. Please report it to the bcache development list: linux-bcache@vger.kernel.org
  133. Be sure to provide as much information that you can including kernel dmesg
  134. output if available so that we may assist.
  135. D) Recovering data without bcache:
  136. If bcache is not available in the kernel, a filesystem on the backing
  137. device is still available at an 8KiB offset. So either via a loopdev
  138. of the backing device created with --offset 8K, or any value defined by
  139. --data-offset when you originally formatted bcache with `bcache make`.
  140. For example::
  141. losetup -o 8192 /dev/loop0 /dev/your_bcache_backing_dev
  142. This should present your unmodified backing device data in /dev/loop0
  143. If your cache is in writethrough mode, then you can safely discard the
  144. cache device without loosing data.
  145. E) Wiping a cache device
  146. ::
  147. host:~# wipefs -a /dev/sdh2
  148. 16 bytes were erased at offset 0x1018 (bcache)
  149. they were: c6 85 73 f6 4e 1a 45 ca 82 65 f5 7f 48 ba 6d 81
  150. After you boot back with bcache enabled, you recreate the cache and attach it::
  151. host:~# bcache make -C /dev/sdh2
  152. UUID: 7be7e175-8f4c-4f99-94b2-9c904d227045
  153. Set UUID: 5bc072a8-ab17-446d-9744-e247949913c1
  154. version: 0
  155. nbuckets: 106874
  156. block_size: 1
  157. bucket_size: 1024
  158. nr_in_set: 1
  159. nr_this_dev: 0
  160. first_bucket: 1
  161. [ 650.511912] bcache: run_cache_set() invalidating existing data
  162. [ 650.549228] bcache: register_cache() registered cache device sdh2
  163. start backing device with missing cache::
  164. host:/sys/block/md5/bcache# echo 1 > running
  165. attach new cache::
  166. host:/sys/block/md5/bcache# echo 5bc072a8-ab17-446d-9744-e247949913c1 > attach
  167. [ 865.276616] bcache: bch_cached_dev_attach() Caching md5 as bcache0 on set 5bc072a8-ab17-446d-9744-e247949913c1
  168. F) Remove or replace a caching device::
  169. host:/sys/block/sda/sda7/bcache# echo 1 > detach
  170. [ 695.872542] bcache: cached_dev_detach_finish() Caching disabled for sda7
  171. host:~# wipefs -a /dev/nvme0n1p4
  172. wipefs: error: /dev/nvme0n1p4: probing initialization failed: Device or resource busy
  173. Ooops, it's disabled, but not unregistered, so it's still protected
  174. We need to go and unregister it::
  175. host:/sys/fs/bcache/b7ba27a1-2398-4649-8ae3-0959f57ba128# ls -l cache0
  176. lrwxrwxrwx 1 root root 0 Feb 25 18:33 cache0 -> ../../../devices/pci0000:00/0000:00:1d.0/0000:70:00.0/nvme/nvme0/nvme0n1/nvme0n1p4/bcache/
  177. host:/sys/fs/bcache/b7ba27a1-2398-4649-8ae3-0959f57ba128# echo 1 > stop
  178. kernel: [ 917.041908] bcache: cache_set_free() Cache set b7ba27a1-2398-4649-8ae3-0959f57ba128 unregistered
  179. Now we can wipe it::
  180. host:~# wipefs -a /dev/nvme0n1p4
  181. /dev/nvme0n1p4: 16 bytes were erased at offset 0x00001018 (bcache): c6 85 73 f6 4e 1a 45 ca 82 65 f5 7f 48 ba 6d 81
  182. G) dm-crypt and bcache
  183. First setup bcache unencrypted and then install dmcrypt on top of
  184. /dev/bcache<N> This will work faster than if you dmcrypt both the backing
  185. and caching devices and then install bcache on top. [benchmarks?]
  186. H) Stop/free a registered bcache to wipe and/or recreate it
  187. Suppose that you need to free up all bcache references so that you can
  188. fdisk run and re-register a changed partition table, which won't work
  189. if there are any active backing or caching devices left on it:
  190. 1) Is it present in /dev/bcache* ? (there are times where it won't be)
  191. If so, it's easy::
  192. host:/sys/block/bcache0/bcache# echo 1 > stop
  193. 2) But if your backing device is gone, this won't work::
  194. host:/sys/block/bcache0# cd bcache
  195. bash: cd: bcache: No such file or directory
  196. In this case, you may have to unregister the dmcrypt block device that
  197. references this bcache to free it up::
  198. host:~# dmsetup remove oldds1
  199. bcache: bcache_device_free() bcache0 stopped
  200. bcache: cache_set_free() Cache set 5bc072a8-ab17-446d-9744-e247949913c1 unregistered
  201. This causes the backing bcache to be removed from /sys/fs/bcache and
  202. then it can be reused. This would be true of any block device stacking
  203. where bcache is a lower device.
  204. 3) In other cases, you can also look in /sys/fs/bcache/::
  205. host:/sys/fs/bcache# ls -l */{cache?,bdev?}
  206. lrwxrwxrwx 1 root root 0 Mar 5 09:39 0226553a-37cf-41d5-b3ce-8b1e944543a8/bdev1 -> ../../../devices/virtual/block/dm-1/bcache/
  207. lrwxrwxrwx 1 root root 0 Mar 5 09:39 0226553a-37cf-41d5-b3ce-8b1e944543a8/cache0 -> ../../../devices/virtual/block/dm-4/bcache/
  208. lrwxrwxrwx 1 root root 0 Mar 5 09:39 5bc072a8-ab17-446d-9744-e247949913c1/cache0 -> ../../../devices/pci0000:00/0000:00:01.0/0000:01:00.0/ata10/host9/target9:0:0/9:0:0:0/block/sdl/sdl2/bcache/
  209. The device names will show which UUID is relevant, cd in that directory
  210. and stop the cache::
  211. host:/sys/fs/bcache/5bc072a8-ab17-446d-9744-e247949913c1# echo 1 > stop
  212. This will free up bcache references and let you reuse the partition for
  213. other purposes.
  214. Troubleshooting performance
  215. ---------------------------
  216. Bcache has a bunch of config options and tunables. The defaults are intended to
  217. be reasonable for typical desktop and server workloads, but they're not what you
  218. want for getting the best possible numbers when benchmarking.
  219. - Backing device alignment
  220. The default metadata size in bcache is 8k. If your backing device is
  221. RAID based, then be sure to align this by a multiple of your stride
  222. width using `bcache make --data-offset`. If you intend to expand your
  223. disk array in the future, then multiply a series of primes by your
  224. raid stripe size to get the disk multiples that you would like.
  225. For example: If you have a 64k stripe size, then the following offset
  226. would provide alignment for many common RAID5 data spindle counts::
  227. 64k * 2*2*2*3*3*5*7 bytes = 161280k
  228. That space is wasted, but for only 157.5MB you can grow your RAID 5
  229. volume to the following data-spindle counts without re-aligning::
  230. 3,4,5,6,7,8,9,10,12,14,15,18,20,21 ...
  231. - Bad write performance
  232. If write performance is not what you expected, you probably wanted to be
  233. running in writeback mode, which isn't the default (not due to a lack of
  234. maturity, but simply because in writeback mode you'll lose data if something
  235. happens to your SSD)::
  236. # echo writeback > /sys/block/bcache0/bcache/cache_mode
  237. - Bad performance, or traffic not going to the SSD that you'd expect
  238. By default, bcache doesn't cache everything. It tries to skip sequential IO -
  239. because you really want to be caching the random IO, and if you copy a 10
  240. gigabyte file you probably don't want that pushing 10 gigabytes of randomly
  241. accessed data out of your cache.
  242. But if you want to benchmark reads from cache, and you start out with fio
  243. writing an 8 gigabyte test file - so you want to disable that::
  244. # echo 0 > /sys/block/bcache0/bcache/sequential_cutoff
  245. To set it back to the default (4 mb), do::
  246. # echo 4M > /sys/block/bcache0/bcache/sequential_cutoff
  247. - Traffic's still going to the spindle/still getting cache misses
  248. In the real world, SSDs don't always keep up with disks - particularly with
  249. slower SSDs, many disks being cached by one SSD, or mostly sequential IO. So
  250. you want to avoid being bottlenecked by the SSD and having it slow everything
  251. down.
  252. To avoid that bcache tracks latency to the cache device, and gradually
  253. throttles traffic if the latency exceeds a threshold (it does this by
  254. cranking down the sequential bypass).
  255. You can disable this if you need to by setting the thresholds to 0::
  256. # echo 0 > /sys/fs/bcache/<cache set>/congested_read_threshold_us
  257. # echo 0 > /sys/fs/bcache/<cache set>/congested_write_threshold_us
  258. The default is 2000 us (2 milliseconds) for reads, and 20000 for writes.
  259. - Still getting cache misses, of the same data
  260. One last issue that sometimes trips people up is actually an old bug, due to
  261. the way cache coherency is handled for cache misses. If a btree node is full,
  262. a cache miss won't be able to insert a key for the new data and the data
  263. won't be written to the cache.
  264. In practice this isn't an issue because as soon as a write comes along it'll
  265. cause the btree node to be split, and you need almost no write traffic for
  266. this to not show up enough to be noticeable (especially since bcache's btree
  267. nodes are huge and index large regions of the device). But when you're
  268. benchmarking, if you're trying to warm the cache by reading a bunch of data
  269. and there's no other traffic - that can be a problem.
  270. Solution: warm the cache by doing writes, or use the testing branch (there's
  271. a fix for the issue there).
  272. Sysfs - backing device
  273. ----------------------
  274. Available at /sys/block/<bdev>/bcache, /sys/block/bcache*/bcache and
  275. (if attached) /sys/fs/bcache/<cset-uuid>/bdev*
  276. attach
  277. Echo the UUID of a cache set to this file to enable caching.
  278. cache_mode
  279. Can be one of either writethrough, writeback, writearound or none.
  280. clear_stats
  281. Writing to this file resets the running total stats (not the day/hour/5 minute
  282. decaying versions).
  283. detach
  284. Write to this file to detach from a cache set. If there is dirty data in the
  285. cache, it will be flushed first.
  286. dirty_data
  287. Amount of dirty data for this backing device in the cache. Continuously
  288. updated unlike the cache set's version, but may be slightly off.
  289. label
  290. Name of underlying device.
  291. readahead
  292. Size of readahead that should be performed. Defaults to 0. If set to e.g.
  293. 1M, it will round cache miss reads up to that size, but without overlapping
  294. existing cache entries.
  295. running
  296. 1 if bcache is running (i.e. whether the /dev/bcache device exists, whether
  297. it's in passthrough mode or caching).
  298. sequential_cutoff
  299. A sequential IO will bypass the cache once it passes this threshold; the
  300. most recent 128 IOs are tracked so sequential IO can be detected even when
  301. it isn't all done at once.
  302. sequential_merge
  303. If non zero, bcache keeps a list of the last 128 requests submitted to compare
  304. against all new requests to determine which new requests are sequential
  305. continuations of previous requests for the purpose of determining sequential
  306. cutoff. This is necessary if the sequential cutoff value is greater than the
  307. maximum acceptable sequential size for any single request.
  308. state
  309. The backing device can be in one of four different states:
  310. no cache: Has never been attached to a cache set.
  311. clean: Part of a cache set, and there is no cached dirty data.
  312. dirty: Part of a cache set, and there is cached dirty data.
  313. inconsistent: The backing device was forcibly run by the user when there was
  314. dirty data cached but the cache set was unavailable; whatever data was on the
  315. backing device has likely been corrupted.
  316. stop
  317. Write to this file to shut down the bcache device and close the backing
  318. device.
  319. writeback_delay
  320. When dirty data is written to the cache and it previously did not contain
  321. any, waits some number of seconds before initiating writeback. Defaults to
  322. 30.
  323. writeback_percent
  324. If nonzero, bcache tries to keep around this percentage of the cache dirty by
  325. throttling background writeback and using a PD controller to smoothly adjust
  326. the rate.
  327. writeback_rate
  328. Rate in sectors per second - if writeback_percent is nonzero, background
  329. writeback is throttled to this rate. Continuously adjusted by bcache but may
  330. also be set by the user.
  331. writeback_running
  332. If off, writeback of dirty data will not take place at all. Dirty data will
  333. still be added to the cache until it is mostly full; only meant for
  334. benchmarking. Defaults to on.
  335. Sysfs - backing device stats
  336. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  337. There are directories with these numbers for a running total, as well as
  338. versions that decay over the past day, hour and 5 minutes; they're also
  339. aggregated in the cache set directory as well.
  340. bypassed
  341. Amount of IO (both reads and writes) that has bypassed the cache
  342. cache_hits, cache_misses, cache_hit_ratio
  343. Hits and misses are counted per individual IO as bcache sees them; a
  344. partial hit is counted as a miss.
  345. cache_bypass_hits, cache_bypass_misses
  346. Hits and misses for IO that is intended to skip the cache are still counted,
  347. but broken out here.
  348. cache_miss_collisions
  349. Counts instances where data was going to be inserted into the cache from a
  350. cache miss, but raced with a write and data was already present (usually 0
  351. since the synchronization for cache misses was rewritten)
  352. cache_readaheads
  353. Count of times readahead occurred.
  354. Sysfs - cache set
  355. ~~~~~~~~~~~~~~~~~
  356. Available at /sys/fs/bcache/<cset-uuid>
  357. average_key_size
  358. Average data per key in the btree.
  359. bdev<0..n>
  360. Symlink to each of the attached backing devices.
  361. block_size
  362. Block size of the cache devices.
  363. btree_cache_size
  364. Amount of memory currently used by the btree cache
  365. bucket_size
  366. Size of buckets
  367. cache<0..n>
  368. Symlink to each of the cache devices comprising this cache set.
  369. cache_available_percent
  370. Percentage of cache device which doesn't contain dirty data, and could
  371. potentially be used for writeback. This doesn't mean this space isn't used
  372. for clean cached data; the unused statistic (in priority_stats) is typically
  373. much lower.
  374. clear_stats
  375. Clears the statistics associated with this cache
  376. dirty_data
  377. Amount of dirty data is in the cache (updated when garbage collection runs).
  378. flash_vol_create
  379. Echoing a size to this file (in human readable units, k/M/G) creates a thinly
  380. provisioned volume backed by the cache set.
  381. io_error_halflife, io_error_limit
  382. These determines how many errors we accept before disabling the cache.
  383. Each error is decayed by the half life (in # ios). If the decaying count
  384. reaches io_error_limit dirty data is written out and the cache is disabled.
  385. journal_delay_ms
  386. Journal writes will delay for up to this many milliseconds, unless a cache
  387. flush happens sooner. Defaults to 100.
  388. root_usage_percent
  389. Percentage of the root btree node in use. If this gets too high the node
  390. will split, increasing the tree depth.
  391. stop
  392. Write to this file to shut down the cache set - waits until all attached
  393. backing devices have been shut down.
  394. tree_depth
  395. Depth of the btree (A single node btree has depth 0).
  396. unregister
  397. Detaches all backing devices and closes the cache devices; if dirty data is
  398. present it will disable writeback caching and wait for it to be flushed.
  399. Sysfs - cache set internal
  400. ~~~~~~~~~~~~~~~~~~~~~~~~~~
  401. This directory also exposes timings for a number of internal operations, with
  402. separate files for average duration, average frequency, last occurrence and max
  403. duration: garbage collection, btree read, btree node sorts and btree splits.
  404. active_journal_entries
  405. Number of journal entries that are newer than the index.
  406. btree_nodes
  407. Total nodes in the btree.
  408. btree_used_percent
  409. Average fraction of btree in use.
  410. bset_tree_stats
  411. Statistics about the auxiliary search trees
  412. btree_cache_max_chain
  413. Longest chain in the btree node cache's hash table
  414. cache_read_races
  415. Counts instances where while data was being read from the cache, the bucket
  416. was reused and invalidated - i.e. where the pointer was stale after the read
  417. completed. When this occurs the data is reread from the backing device.
  418. trigger_gc
  419. Writing to this file forces garbage collection to run.
  420. Sysfs - Cache device
  421. ~~~~~~~~~~~~~~~~~~~~
  422. Available at /sys/block/<cdev>/bcache
  423. block_size
  424. Minimum granularity of writes - should match hardware sector size.
  425. btree_written
  426. Sum of all btree writes, in (kilo/mega/giga) bytes
  427. bucket_size
  428. Size of buckets
  429. cache_replacement_policy
  430. One of either lru, fifo or random.
  431. discard
  432. Boolean; if on a discard/TRIM will be issued to each bucket before it is
  433. reused. Defaults to off, since SATA TRIM is an unqueued command (and thus
  434. slow).
  435. freelist_percent
  436. Size of the freelist as a percentage of nbuckets. Can be written to to
  437. increase the number of buckets kept on the freelist, which lets you
  438. artificially reduce the size of the cache at runtime. Mostly for testing
  439. purposes (i.e. testing how different size caches affect your hit rate), but
  440. since buckets are discarded when they move on to the freelist will also make
  441. the SSD's garbage collection easier by effectively giving it more reserved
  442. space.
  443. io_errors
  444. Number of errors that have occurred, decayed by io_error_halflife.
  445. metadata_written
  446. Sum of all non data writes (btree writes and all other metadata).
  447. nbuckets
  448. Total buckets in this cache
  449. priority_stats
  450. Statistics about how recently data in the cache has been accessed.
  451. This can reveal your working set size. Unused is the percentage of
  452. the cache that doesn't contain any data. Metadata is bcache's
  453. metadata overhead. Average is the average priority of cache buckets.
  454. Next is a list of quantiles with the priority threshold of each.
  455. written
  456. Sum of all data that has been written to the cache; comparison with
  457. btree_written gives the amount of write inflation in bcache.