On every OS I ever create the snapshots of size bigger than the occupied fs space. For example: if I had a fs called /try of 250GB but the space used is 100GB I will create a snapshot of 100GB or a little more. On AIX I create a snapshot with a size under the fs occupied space..and works. On a logical volume of 2GB I create a file of 590MB

dd if=/dev/zero of=hello bs=1m count=590
590+0 records in
590+0 records out

then I create the snapshot of 256mb instead of 590MB or more

snapshot -o snapfrom=/try -o size=256m 
Snapshot for file system /try created on /dev/fslv00

df report the correct size of fs "try"

df -hT /try
/dev/try     jfs2    2,0G  591M    1,5G  29% /try

I can mount the snapshot

mount -o snapshot /dev/fslv00 /mnt/

df report size of 256mb for the snapshot but with 1,5mb used!

df -hT /mnt
/dev/fslv00    jfs2  256M  1,5M    255M   1% /mnt

For the final try I have make a md5sum check of a file..and works for both snapshot and real fs.

b7fb2207c641085a9cf02064edf1742e  /mnt/hello
b7fb2207c641085a9cf02064edf1742e  /try/hello

The questions are simple: how can the snapshot size being under the size of the real fs? Why the space on the snapshot (1,5mb) is under the space used(591mb) of the real fs?

1 Answer 1


Solution found. Is "possible" to create a snapshot with a size under the used fs size of real fs but is totally useless. When I delete the original file /try/hello, the snapshot size become big(because of transfer of the file blocks) but because I have created it under the size of real fs, the fs become full, and the file "hello" became broken, and return i/o error.

df -hT /mnt
/dev/fslv00    jfs2    256M  256M     0 100% /mnt

md5sum /mnt/hello 
md5sum_64: /mnt/hello: There is an input or output error.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .