How do I interpret SMART threshold values?

[PeterSteele]

I've been doing some analysis with a SMART monitoring tool. When I run the tool against one of my drives it indicates that the SMART defined threshold for the internal temperature (attribute 194) is 0. Shouldn't it be something larger? According to what I've read, SMART attribute values can range from 1-253, with 1 meaning bad and 253 being the best possible value. So what does a threshold of 0 for temperature mean? The raw value usually shows something around 25-28, and the normalized value is usually around 100-125. With a threshold of 0 though, how can I tell if a drive has crossed a temperature threshold?

Likewise, how do I interpret the offline_uncorrectable attribute (# 198 )? I currently have a drive with some bad sectors on it, where smartctl shows the raw value for this attribute as 691. The threshold for this attribute is 0, and the current normalized smart value is 182. However, another drive with no bad sectors shows the raw value as 0, the threshold as 0, and the normalized value as 200.

Based on this, I would think that the threshold should be 200, meaning that anything under 200 would indicate the drive has bad sectors. If the threshold is 0, what does this mean in terms in interpreting the current value of the offline_uncorrectable attribute?

Maybe I'm missing something obvious...

[Michael (Passmark)]

All three values (raw, normalised and threshold) come from the drive itself. This means that the manufacturer of the drive gets to decide what they actually mean, and what the threshold should be. In the case where a threshold has been set to 0 it seems the manufacturer has decided not to set a threshold and leave it up to the person viewing the data to make a judgement call.

Unfortunately drive manufacturers tend to not document these values very well, and have been unresponsive when we've tried to ask them.

http://en.wikipedia.org/wiki/Self-Mo...ing_Technology

This is a good page to start at, although keep in mind that the meanings of values can change between drives. In nearly all cases (I think temperature is the only exception we know of) lower normalised values are worse and higher are better. If a normalised value is decreasing over time then the drive is likely degrading.

Hope this helps.

[PeterSteele]

Thanks for the reply. I was beginning to wonder if my findings were a manufacturer specific thing. The temperature for example in the case of our drives does show a lower normalized value as the temperature increases, but as you indicate this is sometimes not the case for some manufacturers (the wikipedia article mentions this too).

You gotta love "standards."