Tool to create and manage device-mapper cache devices.
Go to file
Forza beb6320df8 Update README.md 2023-11-26 16:30:06 +01:00
mdev Update mdev/dmcache.mdev 2023-11-24 10:11:37 +01:00
openrc Update openrc/dmcache.initd 2023-11-24 09:34:14 +01:00
udev Upload files to "udev" 2023-11-24 08:33:29 +01:00
LICENSE Initial commit 2023-11-24 07:04:26 +01:00
README.md Update README.md 2023-11-26 16:30:06 +01:00
cachestats.sh Upload files to "/" 2023-11-26 15:36:18 +01:00
dm-cache.jpg Upload files to "/" 2023-11-26 16:23:36 +01:00
dmcache.sh Upload files to "/" 2023-11-24 09:08:46 +01:00

README.md

dm-cache Device Mapper Cache, is a Linux kernel feature designed to enhance storage performance by implementing a block-level cache on a separate cache device. dm-cache is a tool that helps the user setup cache devices.

The goal with dm-cache is to improve random read/write performance of a slow HDD by using a small but fast SSD or NVME device.

The main advantage of dm-cache over LVM and Bcache is that it is possible to setup on devices that already have a filesystem with data on them. Both LVM and Bcache requires unformatted, empty devices, to setup (there are ways to get around, but can be risky).

Requirements

dm-cache utilises the dmsetup utility which usually can be found in lvm2 or device-mapper packages.

dm-cache requires three devices

  • origin: The slow device.
  • cache: A fast SSD or NVME device. Can be of any size.
  • meta: A small device that holds dm-cache metadata.

The metadata device size depends on how many cache blocks fit on the cache device. With default setting it should be a least 0.01% of the cache device size. If the cache device is 50GiB, and a cache block size of 128KiB, a metadata device of 5MiB is enough. Smaller block sizes requires more metadata and memory, while larger block sizes may reduce effectiveness of the cache by storing cold data. Check metadata usage with cachestats.sh to ensure you are within limits.

It is important to mount the filesystem on the dm-cache using the /dev/mapper/dmname path and not with the filesystem UUID as is commonly done. This is because the kernel might still see the UUID from the origin device, and this can cause data loss!

If you're using Btrfs, the following message can be seen in the kernel log:

# dmesg
BTRFS warning: duplicate device /dev/sdj1 devid 1 generation 182261 scanned by mount (13706)

There is a udev rule that prevents this issue by removing the /dev/disk/by-uuid/ symlink to the origin device.

Documentation

Full documentation is available at https://wiki.tnonline.net/w/Linux/dm-cache

Cache Statistics

cachestats.sh is used to see current stats of the dm-cache device.

# cachestats.sh data2
DEVICE
========
Device-mapper name:       /dev/mapper/data2
Origin size:              9 TiB
Discards:                 no_discard_passdown
CACHE
========
Size / Usage:             100 GiB / 100 GiB (99 %)
Read Hit Rate:            335117403 / 520337199 (64 %)
Write Hit Rate:           24747253 / 31885008 (77 %)
Dirty:                    0 bytes
Block Size:               128 KiB
Promotions / Demotions:   648844 / 648844
Migration Threshold:      1 MiB
Read-Write mode:          rw
Type:                     writeback
Policy:                   smq
Status:                   OK
METADATA
========
Size / Usage:             256 MiB / 10 MiB (3 %)