filesystem overhead

[cheers4asc]

I am creating a filesystem with vxfs and larfiles option . I need to know is there any specific formula to calculate the overhead that is assined to filesystme for inode and some other stuff .

[Patrick Wallek]

Nope. There is actually very little overhead with vxfs filesystems. Inodes are assigned on an as needed basis. You also don't have the 'minfree' parameter like you do with HFS filesystems.

If you create a 4GB vxfs filesystem you will have very close to 4GB available for you to use.

[RAC_1]

A very little overhead.

This thread is helpful.
http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=98595

[Trond Haugen]

I would say that the overhead of a vxfs filesystem is neglectable.
Whith HFS there is some, but nowadays not even /stand has to be HFS.
So an additional advise is to keep ninode low (512).

Regards,
Trond

[Jeff Schussele]

Hi,

The standard "overhead" for a VG is one PE per VG. The default PE size is 4MB, but can & in some cases must be larger to accomodate very large VG sizes.

Then at the FS level you have the space for the metadata - basically the superblock & inode table which contains the intent log. The default intent log size is 1024 blocks. Blocks can be 1,2,4 or 8K in size and the block sized defines the intent log size & block sizes can be specified when the FS is created. Block size default is determined by FS size. The larger the FS size the larger the block size default, but can be overridden with the -b option of the newfs command.

With the mkfs command you can use -o bsize= to define the block size, -o inosize= to define the inode structure size, and -o logsize= to define the intent log size in X blocks with a minimum of 32 & max of 16384 depending on block size.

Man newfs_vxfs & mkfs_vxfs for greater detail.

HTH,
Jeff