A block layer cache (bcache)

Say you’ve got a big slow raid 6, and an ssd or three. Wouldn’t it benice if you could use them as cache… Hence bcache.

Wiki and git repositories are at:

It’s designed around the performance characteristics of SSDs - it only allocatesin erase block sized buckets, and it uses a hybrid btree/log to track cachedextents (which can be anywhere from a single sector to the bucket size). It’sdesigned to avoid random writes at all costs; it fills up an erase blocksequentially, then issues a discard before reusing it.

Both writethrough and writeback caching are supported. Writeback defaults tooff, but can be switched on and off arbitrarily at runtime. Bcache goes togreat lengths to protect your data - it reliably handles unclean shutdown. (Itdoesn’t even have a notion of a clean shutdown; bcache simply doesn’t returnwrites as completed until they’re on stable storage).

Writeback caching can use most of the cache for buffering writes - writingdirty data to the backing device is always done sequentially, scanning from thestart to the end of the index.

Since random IO is what SSDs excel at, there generally won’t be much benefitto caching large sequential IO. Bcache detects sequential IO and skips it;it also keeps a rolling average of the IO sizes per task, and as long as theaverage is above the cutoff it will skip all IO from that task - instead ofcaching the first 512k after every seek. Backups and large file copies shouldthus entirely bypass the cache.

In the event of a data IO error on the flash it will try to recover by readingfrom disk or invalidating cache entries. For unrecoverable errors (meta dataor dirty data), caching is automatically disabled; if dirty data was presentin the cache it first disables writeback caching and waits for all dirty datato be flushed.

Getting started:You’ll need make-bcache from the bcache-tools repository. Both the cache deviceand backing device must be formatted before use:

make-bcache -B /dev/sdbmake-bcache -C /dev/sdc

make-bcache has the ability to format multiple devices at the same time - ifyou format your backing devices and cache device at the same time, you won’thave to manually attach:

make-bcache -B /dev/sda /dev/sdb -C /dev/sdc

bcache-tools now ships udev rules, and bcache devices are known to the kernelimmediately. Without udev, you can manually register devices like this:

echo /dev/sdb > /sys/fs/bcache/registerecho /dev/sdc > /sys/fs/bcache/register

Registering the backing device makes the bcache device show up in /dev; you cannow format it and use it as normal. But the first time using a new bcachedevice, it’ll be running in passthrough mode until you attach it to a cache.If you are thinking about using bcache later, it is recommended to setup all yourslow devices as bcache backing devices without a cache, and you can choose to adda caching device later.See ‘ATTACHING’ section below.

The devices show up as:

/dev/bcache<N>

As well as (with udev):

/dev/bcache/by-uuid/<uuid>/dev/bcache/by-label/<label>

To get started:

mkfs.ext4 /dev/bcache0mount /dev/bcache0 /mnt

You can control bcache devices through sysfs at /sys/block/bcache<N>/bcache .You can also control them through /sys/fs//bcache/<cset-uuid>/ .

Cache devices are managed as sets; multiple caches per set isn’t supported yetbut will allow for mirroring of metadata and dirty data in the future. Your newcache set shows up as /sys/fs/bcache/<UUID>

Attaching

After your cache device and backing device are registered, the backing devicemust be attached to your cache set to enable caching. Attaching a backingdevice to a cache set is done thusly, with the UUID of the cache set in/sys/fs/bcache:

echo <CSET-UUID> > /sys/block/bcache0/bcache/attach

This only has to be done once. The next time you reboot, just reregister allyour bcache devices. If a backing device has data in a cache somewhere, the/dev/bcache<N> device won’t be created until the cache shows up - particularlyimportant if you have writeback caching turned on.

If you’re booting up and your cache device is gone and never coming back, youcan force run the backing device:

echo 1 > /sys/block/sdb/bcache/running

(You need to use /sys/block/sdb (or whatever your backing device is called), not/sys/block/bcache0, because bcache0 doesn’t exist yet. If you’re using apartition, the bcache directory would be at /sys/block/sdb/sdb2/bcache)

The backing device will still use that cache set if it shows up in the future,but all the cached data will be invalidated. If there was dirty data in thecache, don’t expect the filesystem to be recoverable - you will have massivefilesystem corruption, though ext4’s fsck does work miracles.

Error Handling

Bcache tries to transparently handle IO errors to/from the cache device withoutaffecting normal operation; if it sees too many errors (the threshold isconfigurable, and defaults to 0) it shuts down the cache device and switches allthe backing devices to passthrough mode.

  • For reads from the cache, if they error we just retry the read from thebacking device.
  • For writethrough writes, if the write to the cache errors we just switch toinvalidating the data at that lba in the cache (i.e. the same thing we do fora write that bypasses the cache)
  • For writeback writes, we currently pass that error back up to thefilesystem/userspace. This could be improved - we could retry it as a writethat skips the cache so we don’t have to error the write.
  • When we detach, we first try to flush any dirty data (if we were running inwriteback mode). It currently doesn’t do anything intelligent if it fails toread some of the dirty data, though.

Howto/cookbook

  1. Starting a bcache with a missing caching device

If registering the backing device doesn’t help, it’s already there, you just needto force it to run without the cache:

host:~# echo /dev/sdb1 > /sys/fs/bcache/register[  119.844831] bcache: register_bcache() error opening /dev/sdb1: device already registered

Next, you try to register your caching device if it’s present. Howeverif it’s absent, or registration fails for some reason, you can stillstart your bcache without its cache, like so:

host:/sys/block/sdb/sdb1/bcache# echo 1 > running

Note that this may cause data loss if you were running in writeback mode.

  1. Bcache does not find its cache:

    host:/sys/block/md5/bcache# echo 0226553a-37cf-41d5-b3ce-8b1e944543a8 > attach[ 1933.455082] bcache: bch_cached_dev_attach() Couldn't find uuid for md5 in set[ 1933.478179] bcache: __cached_dev_store() Can't attach 0226553a-37cf-41d5-b3ce-8b1e944543a8[ 1933.478179] : cache set not found

In this case, the caching device was simply not registered at bootor disappeared and came back, and needs to be (re-)registered:

host:/sys/block/md5/bcache# echo /dev/sdh2 > /sys/fs/bcache/register
  1. Corrupt bcache crashes the kernel at device registration time:

This should never happen. If it does happen, then you have found a bug!Please report it to the bcache development list:linux-bcache@vger.kernel.org

Be sure to provide as much information that you can including kernel dmesgoutput if available so that we may assist.

  1. Recovering data without bcache:

If bcache is not available in the kernel, a filesystem on the backingdevice is still available at an 8KiB offset. So either via a loopdevof the backing device created with –offset 8K, or any value defined by–data-offset when you originally formatted bcache withmake-bcache.

For example:

losetup -o 8192 /dev/loop0 /dev/your_bcache_backing_dev

This should present your unmodified backing device data in /dev/loop0

If your cache is in writethrough mode, then you can safely discard thecache device without loosing data.

  1. Wiping a cache device
host:~# wipefs -a /dev/sdh216 bytes were erased at offset 0x1018 (bcache)they were: c6 85 73 f6 4e 1a 45 ca 82 65 f5 7f 48 ba 6d 81

After you boot back with bcache enabled, you recreate the cache and attach it:

host:~# make-bcache -C /dev/sdh2UUID:                   7be7e175-8f4c-4f99-94b2-9c904d227045Set UUID:               5bc072a8-ab17-446d-9744-e247949913c1version:                0nbuckets:               106874block_size:             1bucket_size:            1024nr_in_set:              1nr_this_dev:            0first_bucket:           1[  650.511912] bcache: run_cache_set() invalidating existing data[  650.549228] bcache: register_cache() registered cache device sdh2

start backing device with missing cache:

host:/sys/block/md5/bcache# echo 1 > running

attach new cache:

host:/sys/block/md5/bcache# echo 5bc072a8-ab17-446d-9744-e247949913c1 > attach[  865.276616] bcache: bch_cached_dev_attach() Caching md5 as bcache0 on set 5bc072a8-ab17-446d-9744-e247949913c1

  1. Remove or replace a caching device:

    host:/sys/block/sda/sda7/bcache# echo 1 > detach[  695.872542] bcache: cached_dev_detach_finish() Caching disabled for sda7host:~# wipefs -a /dev/nvme0n1p4wipefs: error: /dev/nvme0n1p4: probing initialization failed: Device or resource busyOoops, it's disabled, but not unregistered, so it's still protected

We need to go and unregister it:

host:/sys/fs/bcache/b7ba27a1-2398-4649-8ae3-0959f57ba128# ls -l cache0lrwxrwxrwx 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/host:/sys/fs/bcache/b7ba27a1-2398-4649-8ae3-0959f57ba128# echo 1 > stopkernel: [  917.041908] bcache: cache_set_free() Cache set b7ba27a1-2398-4649-8ae3-0959f57ba128 unregistered

Now we can wipe it:

host:~# wipefs -a /dev/nvme0n1p4/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
  1. dm-crypt and bcache

First setup bcache unencrypted and then install dmcrypt on top of/dev/bcache<N> This will work faster than if you dmcrypt both the backingand caching devices and then install bcache on top. [benchmarks?]

  1. Stop/free a registered bcache to wipe and/or recreate it

Suppose that you need to free up all bcache references so that you canfdisk run and re-register a changed partition table, which won’t workif there are any active backing or caching devices left on it:

  1. Is it present in /dev/bcache* ? (there are times where it won’t be)

    If so, it’s easy:

    host:/sys/block/bcache0/bcache# echo 1 > stop

  2. But if your backing device is gone, this won’t work:

    host:/sys/block/bcache0# cd bcachebash: cd: bcache: No such file or directory

    In this case, you may have to unregister the dmcrypt block device thatreferences this bcache to free it up:

    host:~# dmsetup remove oldds1bcache: bcache_device_free() bcache0 stoppedbcache: cache_set_free() Cache set 5bc072a8-ab17-446d-9744-e247949913c1 unregistered

    This causes the backing bcache to be removed from /sys/fs/bcache andthen it can be reused. This would be true of any block device stackingwhere bcache is a lower device.

  3. In other cases, you can also look in /sys/fs/bcache/:

    host:/sys/fs/bcache# ls -l */{cache?,bdev?}lrwxrwxrwx 1 root root 0 Mar  5 09:39 0226553a-37cf-41d5-b3ce-8b1e944543a8/bdev1 -> ../../../devices/virtual/block/dm-1/bcache/lrwxrwxrwx 1 root root 0 Mar  5 09:39 0226553a-37cf-41d5-b3ce-8b1e944543a8/cache0 -> ../../../devices/virtual/block/dm-4/bcache/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/

    The device names will show which UUID is relevant, cd in that directoryand stop the cache:

    host:/sys/fs/bcache/5bc072a8-ab17-446d-9744-e247949913c1# echo 1 > stop

    This will free up bcache references and let you reuse the partition forother purposes.

Troubleshooting performance

Bcache has a bunch of config options and tunables. The defaults are intended tobe reasonable for typical desktop and server workloads, but they’re not what youwant for getting the best possible numbers when benchmarking.

  • Backing device alignment

    The default metadata size in bcache is 8k. If your backing device isRAID based, then be sure to align this by a multiple of your stridewidth usingmake-bcache –data-offset. If you intend to expand yourdisk array in the future, then multiply a series of primes by yourraid stripe size to get the disk multiples that you would like.

    For example: If you have a 64k stripe size, then the following offsetwould provide alignment for many common RAID5 data spindle counts:

    64k * 2*2*2*3*3*5*7 bytes = 161280k

    That space is wasted, but for only 157.5MB you can grow your RAID 5volume to the following data-spindle counts without re-aligning:

    3,4,5,6,7,8,9,10,12,14,15,18,20,21 ...

  • Bad write performance

    If write performance is not what you expected, you probably wanted to berunning in writeback mode, which isn’t the default (not due to a lack ofmaturity, but simply because in writeback mode you’ll lose data if somethinghappens to your SSD):

    # echo writeback > /sys/block/bcache0/bcache/cache_mode

  • Bad performance, or traffic not going to the SSD that you’d expect

    By default, bcache doesn’t cache everything. It tries to skip sequential IO -because you really want to be caching the random IO, and if you copy a 10gigabyte file you probably don’t want that pushing 10 gigabytes of randomlyaccessed data out of your cache.

    But if you want to benchmark reads from cache, and you start out with fiowriting an 8 gigabyte test file - so you want to disable that:

    # echo 0 > /sys/block/bcache0/bcache/sequential_cutoff

    To set it back to the default (4 mb), do:

    # echo 4M > /sys/block/bcache0/bcache/sequential_cutoff

  • Traffic’s still going to the spindle/still getting cache misses

    In the real world, SSDs don’t always keep up with disks - particularly withslower SSDs, many disks being cached by one SSD, or mostly sequential IO. Soyou want to avoid being bottlenecked by the SSD and having it slow everythingdown.

    To avoid that bcache tracks latency to the cache device, and graduallythrottles traffic if the latency exceeds a threshold (it does this bycranking down the sequential bypass).

    You can disable this if you need to by setting the thresholds to 0:

    # echo 0 > /sys/fs/bcache/<cache set>/congested_read_threshold_us# echo 0 > /sys/fs/bcache/<cache set>/congested_write_threshold_us

    The default is 2000 us (2 milliseconds) for reads, and 20000 for writes.

  • Still getting cache misses, of the same data

    One last issue that sometimes trips people up is actually an old bug, due tothe way cache coherency is handled for cache misses. If a btree node is full,a cache miss won’t be able to insert a key for the new data and the datawon’t be written to the cache.

    In practice this isn’t an issue because as soon as a write comes along it’llcause the btree node to be split, and you need almost no write traffic forthis to not show up enough to be noticeable (especially since bcache’s btreenodes are huge and index large regions of the device). But when you’rebenchmarking, if you’re trying to warm the cache by reading a bunch of dataand there’s no other traffic - that can be a problem.

    Solution: warm the cache by doing writes, or use the testing branch (there’sa fix for the issue there).

Sysfs - backing device

Available at /sys/block/<bdev>/bcache, /sys/block/bcache*/bcache and(if attached) /sys/fs/bcache/<cset-uuid>/bdev*

attach
Echo the UUID of a cache set to this file to enable caching.
cache_mode
Can be one of either writethrough, writeback, writearound or none.
clear_stats
Writing to this file resets the running total stats (not the day/hour/5 minutedecaying versions).
detach
Write to this file to detach from a cache set. If there is dirty data in thecache, it will be flushed first.
dirty_data
Amount of dirty data for this backing device in the cache. Continuouslyupdated unlike the cache set’s version, but may be slightly off.
label
Name of underlying device.
readahead
Size of readahead that should be performed. Defaults to 0. If set to e.g.1M, it will round cache miss reads up to that size, but without overlappingexisting cache entries.
running
1 if bcache is running (i.e. whether the /dev/bcache device exists, whetherit’s in passthrough mode or caching).
sequential_cutoff
A sequential IO will bypass the cache once it passes this threshold; themost recent 128 IOs are tracked so sequential IO can be detected even whenit isn’t all done at once.
sequential_merge
If non zero, bcache keeps a list of the last 128 requests submitted to compareagainst all new requests to determine which new requests are sequentialcontinuations of previous requests for the purpose of determining sequentialcutoff. This is necessary if the sequential cutoff value is greater than themaximum acceptable sequential size for any single request.
state

The backing device can be in one of four different states:

no cache: Has never been attached to a cache set.

clean: Part of a cache set, and there is no cached dirty data.

dirty: Part of a cache set, and there is cached dirty data.

inconsistent: The backing device was forcibly run by the user when there wasdirty data cached but the cache set was unavailable; whatever data was on thebacking device has likely been corrupted.

stop
Write to this file to shut down the bcache device and close the backingdevice.
writeback_delay
When dirty data is written to the cache and it previously did not containany, waits some number of seconds before initiating writeback. Defaults to30.
writeback_percent
If nonzero, bcache tries to keep around this percentage of the cache dirty bythrottling background writeback and using a PD controller to smoothly adjustthe rate.
writeback_rate
Rate in sectors per second - if writeback_percent is nonzero, backgroundwriteback is throttled to this rate. Continuously adjusted by bcache but mayalso be set by the user.
writeback_running
If off, writeback of dirty data will not take place at all. Dirty data willstill be added to the cache until it is mostly full; only meant forbenchmarking. Defaults to on.

Sysfs - backing device stats

There are directories with these numbers for a running total, as well asversions that decay over the past day, hour and 5 minutes; they’re alsoaggregated in the cache set directory as well.

bypassed
Amount of IO (both reads and writes) that has bypassed the cache
cache_hits, cache_misses, cache_hit_ratio
Hits and misses are counted per individual IO as bcache sees them; apartial hit is counted as a miss.
cache_bypass_hits, cache_bypass_misses
Hits and misses for IO that is intended to skip the cache are still counted,but broken out here.
cache_miss_collisions
Counts instances where data was going to be inserted into the cache from acache miss, but raced with a write and data was already present (usually 0since the synchronization for cache misses was rewritten)
cache_readaheads
Count of times readahead occurred.

Sysfs - cache set

Available at /sys/fs/bcache/<cset-uuid>

average_key_size
Average data per key in the btree.
bdev<0..n>
Symlink to each of the attached backing devices.
block_size
Block size of the cache devices.
btree_cache_size
Amount of memory currently used by the btree cache
bucket_size
Size of buckets
cache<0..n>
Symlink to each of the cache devices comprising this cache set.
cache_available_percent
Percentage of cache device which doesn’t contain dirty data, and couldpotentially be used for writeback. This doesn’t mean this space isn’t usedfor clean cached data; the unused statistic (in priority_stats) is typicallymuch lower.
clear_stats
Clears the statistics associated with this cache
dirty_data
Amount of dirty data is in the cache (updated when garbage collection runs).
flash_vol_create
Echoing a size to this file (in human readable units, k/M/G) creates a thinlyprovisioned volume backed by the cache set.
io_error_halflife, io_error_limit
These determines how many errors we accept before disabling the cache.Each error is decayed by the half life (in # ios). If the decaying countreaches io_error_limit dirty data is written out and the cache is disabled.
journal_delay_ms
Journal writes will delay for up to this many milliseconds, unless a cacheflush happens sooner. Defaults to 100.
root_usage_percent
Percentage of the root btree node in use. If this gets too high the nodewill split, increasing the tree depth.
stop
Write to this file to shut down the cache set - waits until all attachedbacking devices have been shut down.
tree_depth
Depth of the btree (A single node btree has depth 0).
unregister
Detaches all backing devices and closes the cache devices; if dirty data ispresent it will disable writeback caching and wait for it to be flushed.

Sysfs - cache set internal

This directory also exposes timings for a number of internal operations, withseparate files for average duration, average frequency, last occurrence and maxduration: garbage collection, btree read, btree node sorts and btree splits.

active_journal_entries
Number of journal entries that are newer than the index.
btree_nodes
Total nodes in the btree.
btree_used_percent
Average fraction of btree in use.
bset_tree_stats
Statistics about the auxiliary search trees
btree_cache_max_chain
Longest chain in the btree node cache’s hash table
cache_read_races
Counts instances where while data was being read from the cache, the bucketwas reused and invalidated - i.e. where the pointer was stale after the readcompleted. When this occurs the data is reread from the backing device.
trigger_gc
Writing to this file forces garbage collection to run.

Sysfs - Cache device

Available at /sys/block/<cdev>/bcache

block_size
Minimum granularity of writes - should match hardware sector size.
btree_written
Sum of all btree writes, in (kilo/mega/giga) bytes
bucket_size
Size of buckets
cache_replacement_policy
One of either lru, fifo or random.
discard
Boolean; if on a discard/TRIM will be issued to each bucket before it isreused. Defaults to off, since SATA TRIM is an unqueued command (and thusslow).
freelist_percent
Size of the freelist as a percentage of nbuckets. Can be written to toincrease the number of buckets kept on the freelist, which lets youartificially reduce the size of the cache at runtime. Mostly for testingpurposes (i.e. testing how different size caches affect your hit rate), butsince buckets are discarded when they move on to the freelist will also makethe SSD’s garbage collection easier by effectively giving it more reservedspace.
io_errors
Number of errors that have occurred, decayed by io_error_halflife.
metadata_written
Sum of all non data writes (btree writes and all other metadata).
nbuckets
Total buckets in this cache
priority_stats
Statistics about how recently data in the cache has been accessed.This can reveal your working set size. Unused is the percentage ofthe cache that doesn’t contain any data. Metadata is bcache’smetadata overhead. Average is the average priority of cache buckets.Next is a list of quantiles with the priority threshold of each.
written
Sum of all data that has been written to the cache; comparison withbtree_written gives the amount of write inflation in bcache.