Module for managing BCache sets
BCache is a block-level caching mechanism similar to ZFS L2ARC/ZIL, dm-cache and fscache. It works by formatting one block device as a cache set, then adding backend devices (which need to be formatted as such) to the set and activating them.
It's available in Linux mainline kernel since 3.10
https://www.kernel.org/doc/Documentation/bcache.txt
This module needs the bcache userspace tools to function.
New in version 2016.3.0.
Attach a backing devices to a cache set If no dev is given, all backing devices will be attached.
CLI Example:
salt '*' bcache.attach sdc
salt '*' bcache.attach /dev/bcache1
bool or None if nuttin' happened
Create a backing device for attachment to a set. Because the block size must be the same, a cache set already needs to exist.
CLI Example:
salt '*' bcache.back_make sdc cache_mode=writeback attach=True
cache_mode -- writethrough, writeback, writearound or none.
force -- Overwrite existing bcaches
attach -- Immediately attach the backing device to the set
bucket_size -- Size of a bucket (see kernel doc)
Create BCache cache on a block device. If blkdiscard is available the entire device will be properly cleared in advance.
CLI Example:
salt '*' bcache.cache_make sdb reserved=10% block_size=4096
reserved --
if dev is a full device, create a partition table with this size empty.
Note
this increases the amount of reserved space available to SSD garbage collectors, potentially (vastly) increasing performance
block_size -- Block size of the cache; defaults to devices' logical block size
force -- Overwrite existing BCache sets
attach -- Attach all existing backend devices immediately
Show or update config of a bcache device.
If no device is given, operate on the cache set itself.
CLI Example:
salt '*' bcache.config
salt '*' bcache.config bcache1
salt '*' bcache.config errors=panic journal_delay_ms=150
salt '*' bcache.config bcache1 cache_mode=writeback writeback_percent=15
config or True/False
Detach a backing device(s) from a cache set If no dev is given, all backing devices will be attached.
Detaching a backing device will flush its write cache. This should leave the underlying device in a consistent state, but might take a while.
CLI Example:
salt '*' bcache.detach sdc
salt '*' bcache.detach bcache1
Check the state of a single bcache device
CLI Example:
salt '*' bcache.device bcache0
salt '*' bcache.device /dev/sdc stats=True
stats -- include statistics
settings -- include all settings
internals -- include all internals
superblock -- include superblock info
Trigger a start of the full bcache system through udev.
CLI Example:
salt '*' bcache.start
Show the full status of the BCache system and optionally all its involved devices
CLI Example:
salt '*' bcache.status
salt '*' bcache.status stats=True
salt '*' bcache.status internals=True alldevs=True
stats -- include statistics
config -- include settings
internals -- include internals
superblock -- include superblock
Stop a bcache device If no device is given, all backing devices will be detached from the cache, which will subsequently be stopped.
Warning
'Stop' on an individual backing device means hard-stop; no attempt at flushing will be done and the bcache device will seemingly 'disappear' from the device lists
CLI Example:
salt '*' bcache.stop
Read out BCache SuperBlock
CLI Example:
salt '*' bcache.device bcache0
salt '*' bcache.device /dev/sdc
Return the bcache UUID of a block device. If no device is given, the Cache UUID is returned.
CLI Example:
salt '*' bcache.uuid
salt '*' bcache.uuid /dev/sda
salt '*' bcache.uuid bcache0