Is SMART a smart idea?

[no_name]

I have a GA-K8VT800 motherboard with 2x SATA 160Gb Maxtor HDDs... Is it
worth while enabling SMART monitoring in the BIOS? How would it affect
system performance?

 

[Arno Wagner]

I have a GA-K8VT800 motherboard with 2x SATA 160Gb Maxtor HDDs... Is it
worth while enabling SMART monitoring in the BIOS?

Yes. But likely to give you warnings too late, since usually only
tested at power-up. Better to do SMART monitoring in the OS regularly.
This will not give you perfect reliability, but SMART can actually
diagnose some problems (depending on the concrete drive) before
you loose data.

It is also a good idea to run a long SMART self-test once a month
or so to find sectors going bad in seldom-used areas of the disk.

How would it affect system performance?

Insignificantly.

Arno

 

[Ian Roberts]

Previously no_name <[email protected]> wrote:

big snip....

Better to do SMART monitoring in the OS regularly.
This will not give you perfect reliability, but SMART can actually
diagnose some problems (depending on the concrete drive) before
you loose data.

It is also a good idea to run a long SMART self-test once a month
or so to find sectors going bad in seldom-used areas of the disk.

Hi Arno

This thread caught my eye.

Just wondered how do we implement SMART monitoring in the OS? Is this
something built into Windows XP or does it require a third party app?

Similarly, how do I run a long SMART self-test?

Thanks a lot.

Ian

 

[Arno Wagner]

big snip....

Hi Arno

This thread caught my eye.

Just wondered how do we implement SMART monitoring in the OS? Is this
something built into Windows XP or does it require a third party app?

Since I do not use Windows for anything except gaming, I can only offer
you the Linux-view. On Linux what you do it to run a monitoring
programm like "smartctl" periodically and look whether something
critical has changed. This can be done manually with, e.g., a
Perl-script that is run from a cron-job (periodic sheduler).
I run this every 5 minutes on my servers to log and do temperature
monitoring. There is also a related tool called "smartd", that
sends email when important fields (like reallocated_sector_count)
change.

Smartctl is also available for Wondows, as far as I understand.
I have no idea (and do not care) whether Windows has a periodic
execution facility. I also have no idea whether smartd was
ported to Windows.

Perl is available on Windows iw you want to roll something yourself.

I don't think that Windows has integrated SMART monitoring.

Similarly, how do I run a long SMART self-test?

smartctl -t long <device>

under Linux (something similar under Windows I would guess).
Wait until finished, look into the log or check whether SMART values
have changed. Takes, e.g. about 80 minutes on a 200GB Maxtor.

smartctl -a <device>

will tell you in the header what the disk thinks its "Extended self
test time" is. If the disk has problems it may take longer.
This command also shows you all the values (e.g. for use from
Perl) and the last few self-test results.

Arno