UNIX for Dummies Questions & Answers

hi guys

i got confused about stat output

stat manual says

File : Size in Bytes
Blocks : Number of blocks used
IO Block : Size in bytes of every block.

when i use stat command for passwd file it says

Code:

~#stat /etc/passwd   File: `/etc/passwd' Size: 999       	Blocks: 8          IO Block: 4096   regular file Device: ca00h/51712d	Inode: 845         Links: 1 Access: (0644/-rw-r--r--)  Uid: (    0/    root)   Gid: (    0/    root) Access: 2011-05-26 12:14:41.000000000 +0200 Modify: 2011-05-26 12:14:41.000000000 +0200 Change: 2011-05-26 12:14:41.000000000 +0200

size is 999 bytes and every block is 4096 bytes how can this file use 8 blocks ???

8 blocks === 8 * 4096 = 32768

file size should be 32769 bytes to use 8 blocks !!

Files can have smaller sizes than their block usage. You can have a file that's 1 byte in size, even though it'd use at least one block. The extra space is simply reserve for it.

It's been given more than one block in case someone needs to extend the file later. Less fragmentation that way.

The file might also have been larger in the past.

thx man but i think it's not rational !!

a file with 999 in bytes which can store in only 1 block use 8 blocks ???

1 block for store and 7 blocks reserve for it ????

For frequently-modified files, it's rational. I have a set of text files that gets appended to frequently, and if they had more blocks reserved for them at a time, they'd fragment less than they do. If I found a straightforward way to change that without reformatting the FS I would.

Also, like said, it may have been larger in the past, and the extra blocks got kept in case the file becomes large again.

The other thing is, it may not be going by the I/O block here.

Does your stat allow for the --printf option? If yes: %B The size in bytes of each block reported by %b
Issue

Code:

stat --printf=%B /etc/passwd
512

- this will output the disk block size. 8 * 512 = 4096, the usual disk block allocation size.