brintos

brintos / linux-shallow public Read only

0
0
Text · 4.2 KiB · 60c16b7 Raw
115 lines · plain
1=================2Writecache target3=================4 5The writecache target caches writes on persistent memory or on SSD. It6doesn't cache reads because reads are supposed to be cached in page cache7in normal RAM.8 9When the device is constructed, the first sector should be zeroed or the10first sector should contain valid superblock from previous invocation.11 12Constructor parameters:13 141. type of the cache device - "p" or "s"15	- p - persistent memory16	- s - SSD172. the underlying device that will be cached183. the cache device194. block size (4096 is recommended; the maximum block size is the page20   size)215. the number of optional parameters (the parameters with an argument22   count as two)23 24	start_sector n		(default: 0)25		offset from the start of cache device in 512-byte sectors26	high_watermark n	(default: 50)27		start writeback when the number of used blocks reach this28		watermark29	low_watermark x		(default: 45)30		stop writeback when the number of used blocks drops below31		this watermark32	writeback_jobs n	(default: unlimited)33		limit the number of blocks that are in flight during34		writeback. Setting this value reduces writeback35		throughput, but it may improve latency of read requests36	autocommit_blocks n	(default: 64 for pmem, 65536 for ssd)37		when the application writes this amount of blocks without38		issuing the FLUSH request, the blocks are automatically39		committed40	autocommit_time ms	(default: 1000)41		autocommit time in milliseconds. The data is automatically42		committed if this time passes and no FLUSH request is43		received44	fua			(by default on)45		applicable only to persistent memory - use the FUA flag46		when writing data from persistent memory back to the47		underlying device48	nofua49		applicable only to persistent memory - don't use the FUA50		flag when writing back data and send the FLUSH request51		afterwards52 53		- some underlying devices perform better with fua, some54		  with nofua. The user should test it55	cleaner56		when this option is activated (either in the constructor57		arguments or by a message), the cache will not promote58		new writes (however, writes to already cached blocks are59		promoted, to avoid data corruption due to misordered60		writes) and it will gradually writeback any cached61		data. The userspace can then monitor the cleaning62		process with "dmsetup status". When the number of cached63		blocks drops to zero, userspace can unload the64		dm-writecache target and replace it with dm-linear or65		other targets.66	max_age n67		specifies the maximum age of a block in milliseconds. If68		a block is stored in the cache for too long, it will be69		written to the underlying device and cleaned up.70	metadata_only71		only metadata is promoted to the cache. This option72		improves performance for heavier REQ_META workloads.73	pause_writeback n	(default: 3000)74		pause writeback if there was some write I/O redirected to75		the origin volume in the last n milliseconds76 77Status:78 791. error indicator - 0 if there was no error, otherwise error number802. the number of blocks813. the number of free blocks824. the number of blocks under writeback835. the number of read blocks846. the number of read blocks that hit the cache857. the number of write blocks868. the number of write blocks that hit uncommitted block879. the number of write blocks that hit committed block8810. the number of write blocks that bypass the cache8911. the number of write blocks that are allocated in the cache9012. the number of write requests that are blocked on the freelist9113. the number of flush requests9214. the number of discarded blocks93 94Messages:95	flush96		Flush the cache device. The message returns successfully97		if the cache device was flushed without an error98	flush_on_suspend99		Flush the cache device on next suspend. Use this message100		when you are going to remove the cache device. The proper101		sequence for removing the cache device is:102 103		1. send the "flush_on_suspend" message104		2. load an inactive table with a linear target that maps105		   to the underlying device106		3. suspend the device107		4. ask for status and verify that there are no errors108		5. resume the device, so that it will use the linear109		   target110		6. the cache device is now inactive and it can be deleted111	cleaner112		See above "cleaner" constructor documentation.113	clear_stats114		Clear the statistics that are reported on the status line115