how to increse inode numbers in a existing file system

[Susovan]

inode is full in the file system "/data/oracle " in a hp-ux server.without removing any file or directory how to increse the inode numbers,in the existing file system.

os version is--hp-ux 11iv2

[Jose Mosquera]

Hi,

Please provide this command output:
#fstyp /dev/vgname/lvolname

Rgds.

[Jose Mosquera]

Hi again,

Omit previos post!
Please provide this command output:
#fstyp -v /dev/vgname/lvolname

Rgds

[Dennis Handly]

What does "bdf -i /data/oracle" show?
If this was vxfs, you shouldn't run out.

[Matti_Kurkela]

On a VxFS filesystem, you *cannot* run out of inodes because the filesystem can automatically create more inodes as long as there is disk space available. (For this reason, if you setup monitoring on VxFS for number of inodes, you may get a lot of meaningless alarms.)

Is this a HFS filesystem?

Please run this command and paste the output to this thread, to verify the filesystem type and version:

fstyp -v $(grep /data/oracle /etc/fstab | awk '{print $1};')

On a HFS filesystem, the number of inodes is related to the filesystem size. This relation is determined at filesystem creation time and cannot be changed later without backing up all data and completely recreating the filesystem. The normal ratio is one inode for every 6144 bytes of disk space.

So, to get more inodes, you must extend the filesystem: as the filesystem is extended, new inodes are created to keep the ratio of inodes per unit of disk space the same as before.

Unfortunately, a HFS filesystem cannot be extended on-line: you must stop Oracle and temporarily unmount the filesystem to extend it using the extendfs command. To extend a filesystem, you must first extend the LV it's placed on; the LV extension can happen while the filesystem is still mounted and Oracle is running.

Example procedure: (assuming the filesystem is located on /dev/vgoracle/lvdata)

lvextend -L /dev/vgoracle/lvdata

umount /data/oracle
extendfs -F hfs /dev/vgoracle/rlvdata
mount /data/oracle


MK

[Ismail Azad]

Hi,

Adding on, on a v3 operating system having vxfs filesystem ofcourse if you execute the following command...

mkfs -F vxfs -m /dev/vgname/lvolname, you will see the number of inodes is unlimited {ninodes=unlimited}. The answer as to why it is so is explained in Matti's post.


Regards
Ismail Azad

[Dennis Handly]

>MK: On a VxFS filesystem, you *cannot* run out of inodes

Provided you created it with the default ninode=unlimited. :-)