[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: RE : dealing with mke2fs -T option



GARDAIS Ionel wrote:
> But as stated in the mke2fs man page, the -T option is for 1 inode 
> per X bytes. Except a lower number of available inodes, if 1 file ==
> 1 inode whatever the -T option used, what is this option for ?

The filesystem has a fixed number of inodes available, which are
allocated in one table when the FS is formatted.  After that, you can't
add more inodes if you run out.  (And if you run out of inodes, you
can't create more files on the FS, even if there's disk space available.
I'm not quite sure why the inode table can't be resized, but it may need
to be contiguous on the disk, or something.)

The -T option exists to tweak the number of entries available in the
inode table, based on the size of the filesystem.

(From your second message:)

GARDAIS Ionel wrote:
> I'm gonna write files of 32Mb each on a Xserve RAID with two "arrays"
> of 4.1Tb (sdd and sde).

You would be able to fit 131072 files on the RAID array before running
out of space if the array was 4 terabytes.  But since it's slightly
larger than 4 terabytes, you'll be able to fit a few more files on it.
Anyway, this means you will only need that many slots in the FS's inode
table.

If there's a variant of -T that takes an explicit size-per-created-inode
value, then I'd try that, with 32Mb as the size-per-created-inode.  (Or
possibly 16Mb, to give yourself some wiggle room.)  OTOH, the
"largefile4" option is probably fairly close to what you want, too: it
leaves some space available if you create some smaller files, and
(AFAIK) having 8 times as many inodes available doesn't waste much disk
space.

Attachment: signature.asc
Description: OpenPGP digital signature


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]