S.M.A.R.T. ... is it?

[Fraser Campbell]

Hi,

I've been playing with smart monitoring of my disks for about a year and I'm 
not sure yet if it's at all useful.  For those that don't know smart is a 
technology for querying disks as to their overall health.  It can 
(supposedly) predict failures before they happen.

I recently had a drive in my server fail (fortunately raided).  The first sign 
of failure was that one of the partitions in the raid array failed on me.  
Almost every time I rebooted another partition would drop out of the raid 
(software raid, probably 6 total partitions).

I took my time replacing the failing disk (perhaps 2 months) but as my disk 
slowly but surely was dying the smart utilities never reported a failure or 
impending failure.  What I did notice was frequent log messages like those 
below:

    smartd: Device: /dev/hdc, S.M.A.R.T. Attribute: 1 Changed 1 
    smartd: Device: /dev/hdc, S.M.A.R.T. Attribute: 195 Changed

Sometimes attributes change +1 (good I think), sometimes they change -1 ... 
occassionally the increments would be larger than 1.

I replaced the disk about a little over week ago and tried to do a fresh 
install on it, I/O errors ... S.M.A.R.T.  still reports that the disk is 
A-OK.

Has anyone ever seen, or heard, of smart actually predicting a failure or is 
it just useless noise in my logfiles?

-- 
Fraser Campbell <fraser-Txk5XLRqZ6CsTnJN9+BGXg at public.gmane.org>                 http://www.wehave.net/
Georgetown, Ontario, Canada                               Debian GNU/Linux

--
The Toronto Linux Users Group.      Meetings: http://tlug.ss.org
TLUG requests: Linux topics, No HTML, wrap text below 80 columns
How to UNSUBSCRIBE: http://tlug.ss.org/subscribe.shtml