最後更新: 2017-11-20

目錄

• Debian package
• truncate (極快)
• fallocate (快)
• dd (慢)

 

---

Debian package

 

/usr/bin/fallocate

• package: util-linux

/usr/bin/truncate

• package: coreutils

/bin/dd

• package: coreutils

 

---

ls Opts

 

link

 

---

truncate

 

說明

shrink or extend the size of a file

 For filesystems which support the fallocate system call (btrfs, ext4, ocfs2, and xfs)

 * no IO to the data blocks (much faster than creating a file by filling it with zeros("dd"))

 * shrink: If a FILE is larger than the specified size, the extra data is lost.

 * extend: If a FILE is shorter, it extended part (hole) reads as zero bytes

Opts

-s[+/-]N(K,M,G)            # size

i.e.

truncate -s 1g iscsi.img

# "ls -s" print the allocated size of each file, in blocks

ls -slh iscsi.img

0 -rw-r--r--. 1 root root 1.0G Nov 19 23:33 iscsi.img

 

---

fallocate

 

fallocate is used to manipulate the allocated disk space for a file, either to deallocate or preallocate it.

說明

Preallocate space to a file            # 須要 fs (ext4, btrfs ...) 支援 fallocate system call

preallocation is done quickly by allocating blocks and marking them as uninitialized

no IO to the data blocks =>  faster than creating a file by filling it with zeros.

fallocate 沒有 overcommit 效果 (因為 allocating blocks and marking them as uninitialized)

fallocate: test.bin: fallocate failed: No space left on device

Opts

-l, --length length           # Specifies the length of the range, in bytes.

• KiB (=1024), MiB (=1024*1024)         # The "iB" is optional
• KB (=1000), MB (=1000*1000)              

no overcommit test

# fallocate 前

df -h /ssd

Filesystem      Size  Used Avail Use% Mounted on
/dev/sdc1       230G   96G  122G  45% /ssd

# fallocate 1G iscsi.img

fallocate -l 1G 1G.img

# df -h 會見到 "Used" 多了 1G

df -h /ssd

Filesystem      Size  Used Avail Use% Mounted on
/dev/sdc1       230G   97G  121G  45% /ssd

ls -sl *.img    # "-s" print the allocated size of each file, in blocks

1025M -rw-r--r-- 1 root root 1024M Sep 20 14:42 1G.img

holes & zero opts

支持的FS: EXT4, XFS, BTRFS

punch a hole

a portion of the file can be marked as unwanted and the associated storage released

-d, --dig-holes

Detect and dig holes. This makes the file sparse in-place, without using extra disk space.

You can think of this option as doing a "cp --sparse" and then renaming the destination file to the original,

  without the need for extra disk space.

If no range is specified by --offset and --length, then the entire file is analyzed for holes

The minimum size of the hole depends on filesystem I/O block size  (usually  4096  bytes)

ie.

dd if=/dev/zero of=zero.bin bs=1M count=100

100+0 records in
100+0 records out
104857600 bytes (105 MB, 100 MiB) copied, 0.0447315 s, 2.3 GB/s

ll -sh zero.bin

100M -rw-r--r--  1 root root 100M Feb 21 14:39 zero.bin

fallocate -d zero.bin

ll -sh zero.bin

   0 -rw-r--r--  1 root root 100M Feb 21 14:39 zero.bin

-p, --punch-hole

Deallocates space (i.e., creates a hole) in the byte range starting at offset and continuing for length bytes.

Within the specified range, partial filesystem blocks are zeroed, and whole filesystem blocks are removed from the file.

 * regardless of what the file contains at that offset: it works even if the file contains non-zeroes there

ie.

echo -n 1111111111111111 > test.txt

hexdump -C test.txt

00000000  31 31 31 31 31 31 31 31  31 31 31 31 31 31 31 31  |1111111111111111|
00000010

fallocate -p -o 8 -l 4  zero.bin

fallocate -p -o 8 -l 4  test.txt

hexdump -C test.txt

00000000  31 31 31 31 31 31 31 31  00 00 00 00 31 31 31 31  |11111111....1111|
00000010

-z, --zero-range

Zeroes space in the byte range starting at offset and continuing for length bytes.

Zeroing is done within the filesystem preferably by converting the range into unwritten extents.

I/O is required only to update metadata. (will not be physically zeroed out on the device)

 

---

dd

 

# Create 1G file (fill with zero)

dd bs=1M count=1024 if=/dev/zero of=test.bin

 

---