On Mon, Mar 03, 2003 at 07:46:23PM +0200, Beni Cherniavsky wrote:
>
> > In addition, don't under estimate the option to return it directly
> > to the manufacturer if there is one. This would probably lead you to
> > verify your suspicion for a broken HD by using the manufacturer
> > tools. Just to demonstrate my attitude to the return to the
> > manufacturer option, if this is a valid option indeed but you, for
> > some reason, won't carry then I will be glad to be given such a HD.
> > It could be that other people on the list would be willing to
> > actually pay for it (for myself I used the word given).
> >
> I'll consider it.
>
In case you are shipping it abroad, do fill the custom forms as
accurately as you can. Do that even if some customs worker tells you it
is redundant.
> > In case there is a some correlation between damaged fs and power
> > failures you might consider a UPS.
>
> The failure that I actaully experienced was apparently triggered by
> reseting the computer with the reset button, perhaps while the disk
> was accessed. There is a chance the failure is only in the data (e.g.
> bad writes => bad error-correction data) so that when I re-write it,
> the disk will be fine. I hope so but I think it's really damaged. I
> suspect it was accumulating fialures before. From time to time the
> system would hang for long, apparently accessing the disk, without
> raising CPU usage. I wonder how that can be, can the low-level disk
> drivers hang the kernel? Perhaps it was just bloking some critical
> application read (e.g. swapping in some part needed by X server should
> give the appearance of hanging).
>
I believe the kernel puts a lot of effort to recover. Maybe it just
get so busy that it looks like it hang but it is not? On the other hand,
on a system with a single HD it could be that the HD controller gets
confused so no real work can continue.
> > As far as I know modern HD have self recovery feature by exploiting
> > reserved sectors which are put by the manufacturer just for that
> > purpose. Once the reserved sectors run out you start to get non
> > repairable HD failures.
> >
> Any way to know how many have been used so far? Early warning
> would be almost a perfect solution.
>
Don't SCSI disks have some utilities which claim to give early
warnings? I don't know about IDE though. In addition, it is hard to give
a reliable early warnings since this is mostly a matter of forecasting
the future.
--
Shaul Karl, [EMAIL PROTECTED] e t
=================================================================
To unsubscribe, send mail to [EMAIL PROTECTED] with
the word "unsubscribe" in the message body, e.g., run the command
echo unsubscribe | mail [EMAIL PROTECTED]
