>>> On 11/27/2008 at  8:10 AM, "van Sleeuwen, Berry"
<[EMAIL PROTECTED]> wrote: 
> Hello list,
>  
> Last night we had to recover the root disk of one of our SLES9 machines.
> For some unknown reason part of the root disk has disappeared. In the
> end we have restored the disk from our backups but we still have no idea
> why this had happend. It looks like someone has removed selected files
> or directories. Some directories are stil available, such as /bin or
> /boot but other have been partially or entirly removed. /etc contains
> some 40 files, var/log is gone, /var/spool still exists.
>  
> Would it be possible to recover the data from the disk? Are there any
> tools available that can analyze and/or recover files? For instance the
> /var/log has vanished so if we could recover that perhaps it would give
> us some clue as to what happend.

The fact that you restored a bunch of files from backup makes that far less 
likely than if you had left the disk alone.  However, you can do things like 
this:
dd if=/dev/dasda1 bs=512 count=20 | hexdump -C
dd if=/dev/dasda1 bs=512 count=20 skip=20 | hexdump -C
dd if=/dev/dasda1 bs=512 count=20 skip=40 | hexdump -C

and you'll get a hexidecimal dump with ASCII translation to the side.  You can 
choose whatever blocksize you want, and whatever count you want.  If you leave 
the count off, you'll dump the whole partition.

> Do you have any thoughts as to why a part of a filesystem can just
> disappear?

My first thought, based on your description, was some kind of human error on 
the part of a Linux sysadmin, or a z/VM systems programmer.  Florian's 
suggestion that _part_ of the minidisk may have been overlaid could produce the 
results you're seeing.  Since there's no way to know what parts of the file 
system are located on what cylinders (unless you've mapped them out 
beforehand), you could have files from any number of directories located in the 
same cylinder of disk space.  If you have some sort of directory management 
tool, then this becomes less likely a cause, although not impossible.  Human 
error on the part of the Linux systems administrators would have been more 
likely to cause entire contents of directories to be deleted, not just parts of 
some, and all of others.  Unless there is only one directory that has partial 
contents, and all the rest are entirely gone.

When the system was rebooted and the problem discovered, did you see any 
console messages about file system checks being run, or journals being 
recovered?  If not, that would indicate the file systems had been unmounted 
during an orderly shutdown.  If you did see them, it's possible someone forced 
the guest off and you were just unlucky enough to lose data.


Mark Post

----------------------------------------------------------------------
For LINUX-390 subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit
http://www.marist.edu/htbin/wlvindex?LINUX-390

Reply via email to