Re: Volume management on Linux with the ext2fs.

Miguel de Icaza (
Wed, 23 Apr 1997 11:06:29 -0500

> > So, when a user adds a new disk to an existing file system, the
> >ext2fs code will get the number of available blocks on this extra disk
> >and the number of block groups, and add this information to the number
> >of blocks/block groups he knows about.
> The one fly in the ointment is that you would *probably* want
> to be able to grow the inode table when you expand the filesystem. I
> am not sure - would it make sense to give each disk an inode table in such
> a way that the complete inode table would consist of the sum of the inode
> tables on each disk.

I believe this is the case: The device is divided in a set of block
groups. Every block group comes with his copy of the superblock, all
the bitmaps it needs, a bunch of inodes to use all by itself and its
data blocks. So, this means that we can just use all of the inodes
that are part of the new groups in the new disk as soon as the disk is
plugged in.

As a first implementation we could put the restriction that the new
file system being added shared some properties with the original file
system: block size, blocks per group and inodes per group. As soon as
I finish reading the ext2 source I will comment more on the options we
have here and how many restrictions we need to make.