OT: predicting harddisk failure 24 hours ahead, did you know about this utility?

Jan Panteltje skrev:

utility?

Yes since quite a long time ago.

This is what I've got the past year.

smartctl -H /dev/ad0 smartctl version 5.36 [i386-portbld-freebsd6.2] Copyright (C) 2002-6 Bruce Allen Home page is

=== START OF READ SMART DATA SECTION === SMART overall-health self-assessment test result: FAILED! Drive failure expected in less than 24 hours. SAVE ALL DATA.

On a sunny day (Thu, 18 Sep 2008 22:06:38 +0200) it happened Sven Svensson wrote in :

utility?

I must have had it on my system for years, only found out recently about it through some online article.

So, I suppose you took that advice, and saved all data.... I have to admit I would look a bit closer at smartctl -a ... before I bought a new HD. But this is utility is a very cool thing, even tells you the number of hours your hd has been on. Divided by 3600 x 24 that gave me 4.5 years... for a Seagate hda,

I often wondered about the state of the old drive, this gives great info.

On a sunny day (Thu, 18 Sep 2008 21:05:39 GMT) it happened Jan Panteltje wrote in :

oops 365x24 of course.

Very nice. The Windoze version runs under Cygwin. It's all part of the utilities that read S.M.A.R.T. data from the drive. Some BIOS versions also do that, but only give a warning that the drive is having a bad day.

For Windoze, I use SpeedFan:

(scroll down to S.M.A.R.T. window) It may sound like just a fan control utility, but buried in the menus is one of the few free S.M.A.R.T. reporting systems for Windoze. If you click on the "analyze" button, it sends your data to their web server and returns a human readable (and understandable) report on the condition of the drive.

How S.M.A.R.T. works:

There's also HDAT2 (free):

which offers the same reports, but also the ability to manipulate the S.M.A.R.T. registers and do some drive testing.

I've had erratic luck with these programs and others over the years. I've caught some drives just before a crash. In general, I can get about a 1 to 5 day warning before impending disaster. However, many drives simply don't report S.M.A.R.T. numbers or worse, report useless numbers showing perfect operation. Some old Maxtor drives did exactly that. Similarly, some IBM drive firmware would fail to report some important errors, such errors booting (reading from the drive firmware load tracks). Some vendor ship drives that have survived an extensive final test, and are showing errors on a brand new drive. Those are fine, but watch out if they change value in any way.

In general, if the S.M.A.R.T. report shows problems, the drive is going to fail shortly. If the report shows perfection over and extended period of operation, it's lying.