Login

stm / [WIP] ZFS commands

Last active 1 month ago

Like 0

Commands to create new ZFS pool and datasets on Linux

samba zfs

stm revised this gist 1 month ago · eb76af0

1 file changed, 3 insertions, 2 deletions

zfs.txt
@@ -21,6 +21,7 @@
21 21 # https://openzfs.github.io/openzfs-docs/man/master/7/zfsprops.7.html#dnodesize
22 22 # https://openzfs.github.io/openzfs-docs/man/master/7/zfsprops.7.html#longname
23 23 # https://openzfs.github.io/openzfs-docs/man/master/7/zfsprops.7.html#pbkdf2iters
24 + # https://openzfs.github.io/openzfs-docs/man/master/7/zfsprops.7.html#xattr
24 25 #
25 26
26 27
@@ -36,7 +37,7 @@ zpool create \
36 37 -O acltype=posixacl -O xattr=sa -O dnodesize=auto \
37 38 -O compression=on \
38 39 -O normalization=formD \
39 - -O relatime=on \
40 + -O atime=off \
40 41 -O longname=on \
41 42 -O canmount=off \
42 43 -O mountpoint=/mnt/$${POOL_NAME} ${POOL_NAME} \
@@ -72,7 +73,7 @@ zfs create -o recordsize=4M ${POOL_NAME}/recordsize-test/4M
72 73 # Dataset creation (media, file server)
73 74 #
74 75 # casesensitivity=insensitive: Can speed up SAMBA sorting due to SAMBA being case-insensitive as well
75 - # recordsize=XXX Bigger recordsize can yield higher compression rations (if data is suitable; video files aren't)
76 + # recordsize=XXX Bigger recordsize can yield higher compression ratios (if data is suitable; video files aren't)
76 77 #
77 78 TODO
78 79

stm revised this gist 1 month ago · 97878a7

1 file changed, 32 insertions, 4 deletions

zfs.txt
@@ -5,7 +5,7 @@
5 5 # encryption=on: Defaults to aes-256-gcm (since 0.8.4)
6 6 # pbkdf2iters=1400000: PBKDF2-SHA1 iterations recommendation by OWASP, default is only 350000
7 7 # dnodesize=auto: Is supposed to be used on systems running SAMBA with xattr=sa
8 - # compression=on: Defaults to lz4 (when lz4_compress is enabled, see `zpool get feature@lz4_compress ${NAME}`)
8 + # compression=on: Defaults to lz4
9 9 # normalization=formD: Decomposed Unicode normalization, implies utf8only=on, so UTF-8 filenames only
10 10 # longname=on: Allows filenames up to 1023 bytes, default is off meaning 255 bytes (supported since 2.3.0)
11 11 # primarycache: Defines what is cached in ARC, defaults to all (alternative settings: none, metadata)
@@ -39,20 +39,48 @@ zpool create \
39 39 -O relatime=on \
40 40 -O longname=on \
41 41 -O canmount=off \
42 - -O mountpoint=/mnt/${NAME} ${NAME} \
42 + -O mountpoint=/mnt/$${POOL_NAME} ${POOL_NAME} \
43 43 raidz2 \
44 44 ${DISK1..N}
45 45
46 46
47 - # Dataset creation (media)
47 + # Dataset creation (benchmarking)
48 48 #
49 - # Bigger record sizes benefit from better compression ratios (in theory; video files are already highly compressed)
49 + # To evaluate which recordsize is most suitable for a certain task we want to hit the raw disk as much as possible.
50 + # Modern OpenZFS versions already allow for direct IO (direct=standard by default), which mostly bypasses ARC/L2ARC.
51 + # In addition we tell ZFS to not hold actual file data in cache, only metadata.
52 + #
53 + zfs create \
54 + -o canmount=off \
55 + -o casesensitivity=insensitive \
56 + -o primarycache=metadata \
57 + -o secondarycache=metadata \
58 + ${POOL_NAME}/recordsize-test
59 + zfs create -o recordsize=4K ${POOL_NAME}/recordsize-test/4K
60 + zfs create -o recordsize=8K ${POOL_NAME}/recordsize-test/8K
61 + zfs create -o recordsize=16K ${POOL_NAME}/recordsize-test/16K
62 + zfs create -o recordsize=32K ${POOL_NAME}/recordsize-test/32K
63 + zfs create -o recordsize=64K ${POOL_NAME}/recordsize-test/64K
64 + zfs create -o recordsize=128K ${POOL_NAME}/recordsize-test/128K
65 + zfs create -o recordsize=256K ${POOL_NAME}/recordsize-test/256K
66 + zfs create -o recordsize=512K ${POOL_NAME}/recordsize-test/512K
67 + zfs create -o recordsize=1M ${POOL_NAME}/recordsize-test/1M
68 + zfs create -o recordsize=2M ${POOL_NAME}/recordsize-test/2M
69 + zfs create -o recordsize=4M ${POOL_NAME}/recordsize-test/4M
70 +
71 +
72 + # Dataset creation (media, file server)
73 + #
74 + # casesensitivity=insensitive: Can speed up SAMBA sorting due to SAMBA being case-insensitive as well
75 + # recordsize=XXX Bigger recordsize can yield higher compression rations (if data is suitable; video files aren't)
50 76 #
51 77 TODO
52 78
79 +
53 80 # Dataset creation (databases)
54 81 #
55 82 # Common database solutions use 4K (SQLite), 8K (PostgreSQL) or 16KB (MariaDB InnoDB) page sizes.
56 83 # Adjusting recordsize to these workloads reduces write amplification and improves random IO.
84 + # Databases usually implement some form of caching themselves, so consider disabling ARC/L2ARC file caching.
57 85 #
58 86 TODO

stm revised this gist 1 month ago · 78743a7

1 file changed, 18 insertions, 5 deletions

zfs.txt
@@ -1,6 +1,4 @@
1 - # RAIDZ2 pool creation
2 - #
3 - # -o (lowercase) is a _pool_ setting, -O (uppercase) is a _root dataset_ setting
1 + # Notes
4 2 #
5 3 # ashift=12: 2^12 = 4KiB sectors, for certain NVMe drives use 13 (8KiB)
6 4 # autotrim=on: TRIM for SSDs
@@ -24,6 +22,11 @@
24 22 # https://openzfs.github.io/openzfs-docs/man/master/7/zfsprops.7.html#longname
25 23 # https://openzfs.github.io/openzfs-docs/man/master/7/zfsprops.7.html#pbkdf2iters
26 24 #
25 +
26 +
27 + # RAIDZ2 pool creation
28 + #
29 + # -o (lowercase) is a _pool_ setting, -O (uppercase) is a _root dataset_ setting
27 30 #
28 31 zpool create \
29 32 -o ashift=12 \
@@ -41,5 +44,15 @@ zpool create \
41 44 ${DISK1..N}
42 45
43 46
44 - # Dataset creation
45 - TODO
47 + # Dataset creation (media)
48 + #
49 + # Bigger record sizes benefit from better compression ratios (in theory; video files are already highly compressed)
50 + #
51 + TODO
52 +
53 + # Dataset creation (databases)
54 + #
55 + # Common database solutions use 4K (SQLite), 8K (PostgreSQL) or 16KB (MariaDB InnoDB) page sizes.
56 + # Adjusting recordsize to these workloads reduces write amplification and improves random IO.
57 + #
58 + TODO

stm revised this gist 1 month ago · a032e43

1 file changed, 1 insertion, 1 deletion

zfs.txt
@@ -5,7 +5,7 @@
5 5 # ashift=12: 2^12 = 4KiB sectors, for certain NVMe drives use 13 (8KiB)
6 6 # autotrim=on: TRIM for SSDs
7 7 # encryption=on: Defaults to aes-256-gcm (since 0.8.4)
8 - # pbkdf2iters=1400000: PBKDF2-SHA1(?) iterations recommendation by OWASP, default is only 350000
8 + # pbkdf2iters=1400000: PBKDF2-SHA1 iterations recommendation by OWASP, default is only 350000
9 9 # dnodesize=auto: Is supposed to be used on systems running SAMBA with xattr=sa
10 10 # compression=on: Defaults to lz4 (when lz4_compress is enabled, see `zpool get feature@lz4_compress ${NAME}`)
11 11 # normalization=formD: Decomposed Unicode normalization, implies utf8only=on, so UTF-8 filenames only

stm revised this gist 1 month ago · 43f34e1

1 file changed, 1 insertion, 1 deletion

zfs.txt
@@ -10,7 +10,7 @@
10 10 # compression=on: Defaults to lz4 (when lz4_compress is enabled, see `zpool get feature@lz4_compress ${NAME}`)
11 11 # normalization=formD: Decomposed Unicode normalization, implies utf8only=on, so UTF-8 filenames only
12 12 # longname=on: Allows filenames up to 1023 bytes, default is off meaning 255 bytes (supported since 2.3.0)
13 - # primarycache: Defines what is cached in ARC , defaults to all (alternative settings: none, metadata)
13 + # primarycache: Defines what is cached in ARC, defaults to all (alternative settings: none, metadata)
14 14 # secondarycache: Defines what is cached in L2ARC, defaults to all (alternative settings: none, metadata)
15 15 # direct: Unset, defaulting to standard (since 2.3.0)
16 16 # !!! recordsize: Unset, defaulting to 128KiB (important performance related setting!)

stm revised this gist 1 month ago · 8ab61e9

1 file changed, 3 insertions, 1 deletion

zfs.txt
@@ -3,13 +3,15 @@
3 3 # -o (lowercase) is a _pool_ setting, -O (uppercase) is a _root dataset_ setting
4 4 #
5 5 # ashift=12: 2^12 = 4KiB sectors, for certain NVMe drives use 13 (8KiB)
6 - # autotrim: TRIM for SSDs
6 + # autotrim=on: TRIM for SSDs
7 7 # encryption=on: Defaults to aes-256-gcm (since 0.8.4)
8 8 # pbkdf2iters=1400000: PBKDF2-SHA1(?) iterations recommendation by OWASP, default is only 350000
9 9 # dnodesize=auto: Is supposed to be used on systems running SAMBA with xattr=sa
10 10 # compression=on: Defaults to lz4 (when lz4_compress is enabled, see `zpool get feature@lz4_compress ${NAME}`)
11 11 # normalization=formD: Decomposed Unicode normalization, implies utf8only=on, so UTF-8 filenames only
12 12 # longname=on: Allows filenames up to 1023 bytes, default is off meaning 255 bytes (supported since 2.3.0)
13 + # primarycache: Defines what is cached in ARC , defaults to all (alternative settings: none, metadata)
14 + # secondarycache: Defines what is cached in L2ARC, defaults to all (alternative settings: none, metadata)
13 15 # direct: Unset, defaulting to standard (since 2.3.0)
14 16 # !!! recordsize: Unset, defaulting to 128KiB (important performance related setting!)
15 17 #

stm revised this gist 1 month ago · d03bf0b

1 file changed, 3 insertions, 2 deletions

zfs.txt
@@ -5,7 +5,7 @@
5 5 # ashift=12: 2^12 = 4KiB sectors, for certain NVMe drives use 13 (8KiB)
6 6 # autotrim: TRIM for SSDs
7 7 # encryption=on: Defaults to aes-256-gcm (since 0.8.4)
8 - # pbkdf2iters=700000: Double the default PBKDF2-SHA1(?) iterations, default is 350000
8 + # pbkdf2iters=1400000: PBKDF2-SHA1(?) iterations recommendation by OWASP, default is only 350000
9 9 # dnodesize=auto: Is supposed to be used on systems running SAMBA with xattr=sa
10 10 # compression=on: Defaults to lz4 (when lz4_compress is enabled, see `zpool get feature@lz4_compress ${NAME}`)
11 11 # normalization=formD: Decomposed Unicode normalization, implies utf8only=on, so UTF-8 filenames only
@@ -14,6 +14,7 @@
14 14 # !!! recordsize: Unset, defaulting to 128KiB (important performance related setting!)
15 15 #
16 16 # Sources:
17 + # https://cheatsheetseries.owasp.org/cheatsheets/Password_Storage_Cheat_Sheet.html#pbkdf2
17 18 # https://openzfs.github.io/openzfs-docs/Getting%20Started/Debian/Debian%20Trixie%20Root%20on%20ZFS.html#step-2-disk-formatting
18 19 # https://openzfs.github.io/openzfs-docs/man/master/7/zfsprops.7.html#compression
19 20 # https://openzfs.github.io/openzfs-docs/man/master/7/zfsprops.7.html#direct
@@ -26,7 +27,7 @@ zpool create \
26 27 -o ashift=12 \
27 28 -o autotrim=on \
28 29 -O encryption=on -O keylocation=prompt -O keyformat=passphrase \
29 - -O pbkdf2iters=700000 \
30 + -O pbkdf2iters=1400000 \
30 31 -O acltype=posixacl -O xattr=sa -O dnodesize=auto \
31 32 -O compression=on \
32 33 -O normalization=formD \

stm revised this gist 1 month ago · 0bff9fd

1 file changed, 1 insertion, 1 deletion

zfs.txt
@@ -10,7 +10,7 @@
10 10 # compression=on: Defaults to lz4 (when lz4_compress is enabled, see `zpool get feature@lz4_compress ${NAME}`)
11 11 # normalization=formD: Decomposed Unicode normalization, implies utf8only=on, so UTF-8 filenames only
12 12 # longname=on: Allows filenames up to 1023 bytes, default is off meaning 255 bytes (supported since 2.3.0)
13 - # direct: Unset, defaulting to always (since 2.2.0)
13 + # direct: Unset, defaulting to standard (since 2.3.0)
14 14 # !!! recordsize: Unset, defaulting to 128KiB (important performance related setting!)
15 15 #
16 16 # Sources:

stm revised this gist 2 months ago · 9071edc

1 file changed, 4 insertions, 2 deletions

zfs.txt
@@ -32,8 +32,10 @@ zpool create \
32 32 -O normalization=formD \
33 33 -O relatime=on \
34 34 -O longname=on \
35 - -O canmount=off -O mountpoint=/${NAME} \
36 - ${NAME} ${DISK1..N} raidz2
35 + -O canmount=off \
36 + -O mountpoint=/mnt/${NAME} ${NAME} \
37 + raidz2 \
38 + ${DISK1..N}
37 39
38 40
39 41 # Dataset creation

stm revised this gist 2 months ago · f19b43e

1 file changed, 5 insertions, 3 deletions

zfs.txt
@@ -1,4 +1,7 @@
1 1 # RAIDZ2 pool creation
2 + #
3 + # -o (lowercase) is a _pool_ setting, -O (uppercase) is a _root dataset_ setting
4 + #
2 5 # ashift=12: 2^12 = 4KiB sectors, for certain NVMe drives use 13 (8KiB)
3 6 # autotrim: TRIM for SSDs
4 7 # encryption=on: Defaults to aes-256-gcm (since 0.8.4)
@@ -8,10 +11,9 @@
8 11 # normalization=formD: Decomposed Unicode normalization, implies utf8only=on, so UTF-8 filenames only
9 12 # longname=on: Allows filenames up to 1023 bytes, default is off meaning 255 bytes (supported since 2.3.0)
10 13 # direct: Unset, defaulting to always (since 2.2.0)
11 - # !!! recordsize: Unset, defaulting to 128KiB
12 - #
14 + # !!! recordsize: Unset, defaulting to 128KiB (important performance related setting!)
13 15 #
14 - # Sources:
16 + # Sources:
15 17 # https://openzfs.github.io/openzfs-docs/Getting%20Started/Debian/Debian%20Trixie%20Root%20on%20ZFS.html#step-2-disk-formatting
16 18 # https://openzfs.github.io/openzfs-docs/man/master/7/zfsprops.7.html#compression
17 19 # https://openzfs.github.io/openzfs-docs/man/master/7/zfsprops.7.html#direct