bdf values do not add up

[Shahril M]

Hi folks,

Brief background:Data in a file system on a server needs to be replicated to another server.

FS of same size created on the other server. However, the bdf values do not tally

server 1:
/dev/vg01/lvburnbuf 983040 925203 54869 94% /ixos/burnbuf

server 2:
/dev/vg01/lvburnbuf 983040 1341 920350 0% /ixos/burnbuf

The total sizes tally, but the sum of the used and avail do not. This causes the replication to fail as the avail space (920350) is insufficient

Appreciate possible explanations for this.


Rgds,
Shahril

[Michael Tully]

What type of filesystems are you using?

# fstyp dev/vg01/lvburnbuf

[Shahril M]

Hi Michael,

Both are JFS.


Rgds,
Shahril

[Sridhar Bhaskarla]

Hi Shahril,

When you create a filesystem, 10% is alloted for minfree area meant for performance and another 5% for system overhead. You will not see atleast 5% in the usage field of your bdf output.

So if you want 100MB, create a logical volume size of around 110 MB as after filesystem creation, you will get approximately 100MB.

-Sri

[Rajesh D L]

Hi Shahril,

The discrepancy could be due to different metastructure of the disks on both the servers, if the disks are from a storage box. Just the compare the disk sizes and number of disks in the volume groups. And as a practice, create/extend logical volumes to specific disks only.

regards,
RDL.

[Michael D'Aulerio]

Shahril,

As Sri points out, a certain amount of free space is reserved on each file system. It looks like the 2 file systems you list have a different % set aside for free space. I am not that familiar with JFS file systems, but with HFS file systems there is a tunefs command that changes the %minfree on a file system. e.g.
tunefs -m 3 /dev/vg01/lvburnbuf
I would umount the file system before running tunefs.

[Jose Mosquera]

Hi,

Pls umount in server 2 /ixos/burnbuf and check if after this files exist inside, in this case you remove them a re-mount the FS.

Rgds.

[Geoff Wild]

Try it with
df -k
instead.

From bdf man page:

The bdf command does not account for any disk space reserved for swap
space, or used for the HFS boot block (8 KB, 1 per file system), HFS
superblocks (8 KB each, 1 per disk cylinder), HFS cylinder group
blocks (1 KB - 8 KB each, 1 per cylinder group), and inodes (currently
128 bytes reserved for each inode). Non-HFS file systems may have
other items not accounted for by this command.


Rgds...Geoff