Federico Sevilla III wrote:
>
> On Sun, 22 Jul 2001 at 08:03, Steve Lord wrote:
> > If you create a large number of files in XFS, inodes are allocated
> > dynamically for them. If you remove all the files, the inodes are not
> > returned to free space. On something like ext2 inodes are allocated at
> > mkfs time and the space is never available for anything else. This I
> > think is what you were referring to.
>
> Yes, this is what I was referring to. I was under the impression that this
> was "bad", though, and one effectively "lost space". Are inodes allocated
> dynamically for large files different from those allocated dynamically for
> a bunch of small files? Meaning, if inodes are allocated for a large file,
> then the large file is deleted, and the inodes are not returned to free
> space, will filling up these allocated inodes with small files be "bad" or
> would filling up the free space with small files (thus creating inodes for
> them) have had the same effect anyway?
I had a minor, nagging worry, that if there is no cap on inode density,
that something like a fork bomb could become a denial of service
attack...allocate till much of the system is inode, and no way to get it
back. It would be good to have a cap on inode density, does something
like that exist now?
D. Stimits, stimits@xxxxxxxxxx
>
> > XFS will slow down doing allocations when it is really full, you are
> > no where near full, 99.x% full is nearly full. Basically XFS chops the
> > filesystem into allocation groups (1 to 4 Gbytes each), free space is
> > managed independently in each of these. The slowdown happens when you
> > have to scan through lots of allocation groups looking for space to
> > extend a file. There is an in memory summary structure which tells you
> > if it is worth even looking in an allocation group, so it is not a
> > major slowdown - unless you have lots of parallel allocation calls
> > going on at the time.
>
> This is great! Perhaps we could have this information in the XFS FAQ under
> something like "when is my disk nearly full and what happens when it's
> nearly full"?
>
> Thanks a lot for the help! :)
>
> --> Jijo
>
> --
> Federico Sevilla III :: jijo@xxxxxxxxxxxxxxxxxxxx
> Network Administrator :: The Leather Collection, Inc.
|