[KLUG Members] ext3 size limits

Adam Williams members@kalamazoolinux.org
04 Oct 2002 09:22:43 -0400


>>If memory serves.. ext2 will handle a 2TB partition... I'd imagine ext3 has the exact same limit.
>>I don't know if your partition size depends on the inode size though.
>umm, file system newbie here ;) What about inode size?

On a "traditional" UNIX file system (Minix, ext2, ext3, etc...) when you
create a file system a certain number of "i-nodes" are created.  These
are informational nodes.  Every file, directory, socket, pipe, etc... in
the file system consumes exactly 1 i-node.  By default ext2/3 mkfs
utility sets up the number of i-nodes in a proportion to the size of the
file system.  If you have 100Gb free and you run out of inodes, tough
luck, you can't create any more files, directories,....  Typically your
fine, as the number of i-nodes created is pretty large.  But if you
create lots of deep directories and small files in a REALLY LARGE file
system you can get into trouble.

Do a "df -i" to see inode availability.

File systems like xfs and jfs create i-node blocks on the fly, so you'll
never have to worry about this.  Although "df -i" will still display
information, the "inodes" column can change constantly, and thus the
"ifree" column as well.

These aren't your father's filesystems!  :) Assuming of course that your
father didn't work on mainframes,  they've had such things forever.