Linux command
du 命令
文件
复制后可按需替换文件名、目录或参数。
常用示例
Show total size of a directory
du -sh [directory]
Show sizes in human-readable format
du -h [directory]
Show sizes of subdirectories
du -h --max-depth=1 [directory]
Show largest subdirectories first
du -h --max-depth=1 [directory] | sort -hr
Show size of all files and directories
du -ah [directory]
Stay on one filesystem
du -shx [directory]
说明
du (disk usage) estimates and reports file and directory space usage by recursively examining directory trees. Unlike df which shows filesystem-level free space, du focuses on individual files and directories, making it essential for identifying what's consuming disk space and where cleanup efforts should focus. The tool walks directory hierarchies and sums the disk space used by each file, reporting sizes at each directory level. By default, it shows sizes for all directories in the tree, but options like --max-depth allow limiting output to higher-level summaries. The -h flag converts raw block counts to human-readable formats (KB, MB, GB). du is commonly combined with sort to identify the largest space consumers. Patterns like "du -h | sort -hr | head" reveal the top disk space users, essential for troubleshooting full filesystems or planning cleanup operations. The command reports disk usage (actual blocks consumed) rather than apparent file size, which matters for sparse files and filesystems with compression. Hard links to the same inode are counted only once by default (use -l to count each link separately). The -x option prevents crossing filesystem boundaries, useful for analyzing specific filesystems without including mounted subdirectories.
参数
- -h, --human-readable
- Human-readable sizes (K, M, G)
- -s, --summarize
- Display total only
- -c, --total
- Produce grand total
- -a, --all
- Include files, not just directories
- -d _N_, --max-depth=_N_
- Maximum directory depth
- -b, --bytes
- Print sizes in bytes
- -k, --kilobytes
- Print sizes in kilobytes
- -m, --megabytes
- Print sizes in megabytes
- -x, --one-file-system
- Skip different filesystems
- -l, --count-links
- Count hard links multiple times
- --apparent-size
- Print apparent sizes rather than disk usage
- --si
- Like -h but use powers of 1000 (not 1024)
- -L, --dereference
- Follow symbolic links
- --exclude=_pattern_
- Exclude files matching pattern
- --time
- Show last modification time
FAQ
What is the du command used for?
du (disk usage) estimates and reports file and directory space usage by recursively examining directory trees. Unlike df which shows filesystem-level free space, du focuses on individual files and directories, making it essential for identifying what's consuming disk space and where cleanup efforts should focus. The tool walks directory hierarchies and sums the disk space used by each file, reporting sizes at each directory level. By default, it shows sizes for all directories in the tree, but options like --max-depth allow limiting output to higher-level summaries. The -h flag converts raw block counts to human-readable formats (KB, MB, GB). du is commonly combined with sort to identify the largest space consumers. Patterns like "du -h | sort -hr | head" reveal the top disk space users, essential for troubleshooting full filesystems or planning cleanup operations. The command reports disk usage (actual blocks consumed) rather than apparent file size, which matters for sparse files and filesystems with compression. Hard links to the same inode are counted only once by default (use -l to count each link separately). The -x option prevents crossing filesystem boundaries, useful for analyzing specific filesystems without including mounted subdirectories.
How do I run a basic du example?
Run `du -sh [directory]` in a terminal, then adjust file names, paths, flags, or remote targets for your system.
What does -h, --human-readable do in du?
Human-readable sizes (K, M, G)