'Some Contents Unreadable" error still exists after checking HDD

Robert Nichols rnichols42 at comcast.net
Mon Feb 21 16:50:24 UTC 2005


Rajev Mhasawade wrote:
> Hi,
> I did use shutdown now -Fr command and it checked filesystem after
> reboot without detecting any errors and stating all drives are
> 'non-contagious'.Yet when i try to find out space used by each
> drive(via right click),information displayed there does not match with
> that of df -h.And the error (some contents cannot be read) is still
> there.
> What does this mean,is my HDD problematic?What should i do?Plz help!

The responses you are reporting do not mean that anything is wrong
with your disk drive.  The "some contents cannot be read" message
simply means that the program is not running with root permissions
and has encountered some directories that it does not have permission
to access.

The disk usage calculated by walking through the file system directory
tree will often not be an exact match for what is reported by 'df'.
There are several reasons for this:

  1. Files that have been unlinked ('rm'ed) from the directory tree
     still occupy space on the disk but cannot be seen for the usage
     summary.

  2. Files that are open for writing will have some data blocks
     pre-allocated.  A program that sums file sizes (st_size in the
     inodes) won't see these blocks, but they are still considered
     "used" by 'df'.  A program that sums block counts (st_blocks
     in the inodes) will see these pe-allocated blocks.

  3. Files with more data blocks than can be listed in the inode
     need additional "indirect" blocks that contain lists of block
     numbers.  As above, these blocks are included in st_blocks but
     not in st_size, so a program that sums file sizes won't see them.

  4. Files with multiple hard links need to be treated specially.
     Programs that walk the directory tree commonly add (size/N),
     where N is the number of hard links, to the usage total each
     time a multiply-linked file is encountered.  Because of integer
     truncation, N*(size/N) may not equal the actual size.

--
Bob Nichols         rnichols42 at comcast.net




More information about the fedora-list mailing list