SMART (was Re: [clug] forget RAID?)

Alex Satrapa grail at goldweb.com.au
Thu Feb 22 04:46:50 GMT 2007


On 22/02/2007, at 15:26 , Michael Cohen wrote:

>   I find that monitoring smart really helps to let you know when a  
> disk is
>   about to die, which helps you plan for replacing it.

If anyone can give a presentation on how to monitor SMART properly,  
I'd absolutely make sure I could get to that CLUG meeting.

My experience with SMART is that it smartmon will continually send me  
messages about junk that doesn't matter ("The temperature of your  
hard drive dropped from 35 to 34!"), and that disk failure just  
suddenly occurs without warning. Not even "hey! the number of read  
errors on this drive, which was stable at 1 for the last few weeks  
has suddenly climbed to 4!"

The only time SMART has been of any use to me was when one drive  
suddenly launched itself into a pyroclasm, starting the week at  
temperature 64 (degrees? some arbitrary level on a scale of 42-69?)  
and heading to 110 before failing.


More information about the linux mailing list