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

[Byron Sonne]

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

Bit of both I'd say; anything based on statistical analysis of some kind 
is really just an educated maybe.

Case in point: I ran a box for 2 years with SMART indicators going off 
and never had a problem or performance issue; I retired the box later 
for unrelated reasons. Last job I had I ran my box with SMART indicators 
for 3 months, then one day the drive choked and died. Not a failed 
partition or anything that simple, the friggin' thing wouldn't even 
autodetect.

On a side note, working with Compaq servers we'd sometimes get the 
orange light going off indicating that the drive had performance issues, 
was dying, or had error rates past thresholds. Disks were SCSI, so I 
don't know if it was SMART or some equivalent. Replace the drive with a 
new one, same thing. WTF we're thinking... upgrade the firmware on the 
RAID controller and the new drive orange light goes off. Just for laughs 
we stuff the old 'bad' drive back in. Wouldn't you know it, now its fine :)

I guess I'd look at it this way: if the SMART is going off, start doing 
backups religiously but otherwise don't panic. If it's a business 
production box or you got people with lawyers watching the box, then 
book your downtime and get it replaced ASAP.

All in all, it's more useful at helping you than the Canadian government 
is at protecting you from torture at the hands of corrupt governments 
like Saudi Arabia, but then again, what isn't?

-- 

	For good, return good. For evil, return justice.

--
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