[plug] harddrive becomes read-only

[Keith Bawden]

On Sun, Sep 7, 2008 at 22:43, Gavin Chester <gavin.chester at gmail.com> wrote:
> Anyway, I had onboard several scsi utility suites, the one seeming most
> applicable being 'scsirastools'. Two utilities in this suite, 'sgdiag'
> and 'sgdefects' told me that one of the drives ahd developed 36 defects
> as per the abbreviated report below. Trouble is, what does it all mean?
>

Grown defects might be harmless, however if the number is high (I
would say 36 is might be getting there) this could indicate eminent
disk failure. IMHO, If the number is growing quickly then it is pretty
much time to swap it out. These "defects" are parts of the disk marked
as bad and as such the OS should ignore the defective areas. The
address of these areas are detailed in the "defect list".

Either way, before spending too much time on troubleshooting now is a
great time to ensure your backs ups are up to date and that you can
actually restore from them. If the disk is still under warranty I
would return it now.

Once you have ensured backups are done I would conduct a fsck on these
filesystem and see if things improve thereafter.

"shutdown -F -r now" might help out here :)

If the drives are SMART[1] capable try SMART tools[2] and see if you
can get anything that either confirms or adds confusion to the
situation.. I would also suggest you look for some diagnostic tools
that are specific to your controller (check the vendors website for
packages compatible with the distro you are using), as the data you
can pull out will be specific to your controller and often more
meaningful. Some controller tools will also allow you to retrieve
SMART data.

Regards, Keith

[1] http://en.wikipedia.org/wiki/Self-Monitoring,_Analysis,_and_Reporting_Technology
[2] http://sourceforge.net/projects/smartmontools/