inode in filesystem

[Eli_pok]

hi,
Out of interest....
I am working on a HP-UX 11i os with a vxfs filesystem.
The questation is if an inode table size reaches the maximum will it take some of the free filesystem sapce ?

[Peter Godron]

Hi,
if the inode table becomes full, you should start getting "Inode Table full" messages.
You can increase with ninode kernel param ( or maxusers), as I believe the inode cache is in-core.

[Jaime Bolanos Rojas. · Accepted solution]

Eli, the vxfs will allocate inodes dinamically until there is no more inodes available in the hard drive.

Regards,

Jaime.

[Steven E. Protter]

Shalom,

No, it will stop working.

SEP

[Jaime Bolanos Rojas.]

Changes are that I understood the question wrong but wasn't it suppord to be dinamic?

The addition of filesets and dynamic allocation of inodes has affected the disk layout in various ways. In particular, many of the file system structures are now located in files (referred to as structural files) rather than in fixed disk areas. This provides a simple mechanism for dynamic growth of structures. For example, inodes are now stored in structural files and allocated as needed. In general, file system structures that deal with space allocation are still in fixed disk locations, while most other structures are dynamically allocated and have become clients of the file system's disk space allocation scheme.

http://docs.hp.com/en/B3929-90011/ch02s04.html

Regards,

Jaime.

[James R. Ferguson]

Hi:

For a VxFS filesystem, inodes will grow in number dynamically as needed as long as there is space available within the filesystem.

A largefiles filesystem is necessary for more than about 8 million inodes.

Regards!

...JRF...

[A. Clay Stephenson]

By default, vxfs filesystems grow the inode list dynamically BUT if mkfs is called with -o ninode=n specified with a non-zero value the the number of inodes is fixed and will not increase "on the fly". There is almost never a good reason for doing this but vxfs filesystems can be made just as static as hfs/ufs filesystems.

[Dave Olker]

Hi Eli,

I know you've gotten a number of correct answers to this question - assuming you're referring to on-disk inodes - but I want to also point out that there is a difference between on-disk inodes, which do occupy space in the filesystem, and in-memory inodes. The HP-UX kernel maintains a memory-based cache of recently used and open VxFS inodes.

Are you referring to the on-disk inodes or the in-memory inode cache? Those are two completely different mechanisms controlled by different tunables.

Regards,

Dave

[Eli_pok]

I ment to the inodes i see by typng "bdf -i".
Te problem i ran into was a filesysten with a lot files (almost 30000) that was created by somenes script ran by cron and genarating an empty log file for the last year or so.
So the quastion came from that direction if the amount of files would reach the maximum number per filesystem but not taking any "real" filesystem space would the filesystem become "full" ?
Also i didnt undestood the diferebce between the two types of inode tables....

[A. Clay Stephenson]

Run this command:
mkfs -m /dev/vgxx/lvoln

The -m option will display the options used to create an existing filesystem but will not create a new filesystem --- so it is a safe command to run. Man mkfs for details.

As long as there is free space in a vxfs filesystem AND if the filesystem was created with ninode=unlimited (the default) then the inode list will expand as needed. When it expands, it doesn't add just 1 more but instead adds a chunk of available inodes. The number reported by bdf -i is the current inode capacity but if the above conditions are met the that number can increase with no intervention on your part. There is 1 more constraint and that is, if largefiles are not enabled, and the underlying inode data structure exceeds 2GiB then no more inodes can be allocated even if ninode=unlimited.

The two types of inode tables are very different things; the one you are talking about is the total number of inodes the filesystem itself can handle; the other one is the number of inodes in the kernel at any one time. Because it would be extremely unlikely that the system would have that many different files open at any one time, the kernel table can be much, much smaller.