Hard drives. What does SMART not detect?

[Dos_Box]

I have a drive in my desktop here that is about to fail. I know it's going to fail as for the past couple of weeks it's been sluggish 'starting' when trying to get files from it and when it does so it does it with an audible click after a delayed start (and no, no power magement in play there).

Not it's started clicking every 30 seconds or so however, SMART reports that all is good! To be fair, It's almost 7 years old and has been running for quite a bit (see pic).

I'm just wondering though as to what things SMART does not detect as some kinds of mechanical failure seem to be on the list perhaps?

 

Edited February 25, 2016 by Dos_Box

[Danp]

300 days? That's nothing really, I would expect that to run much longer.

[Michael]

Smart can't predict failure (with the motor), only highlight there's a potential issue as such. I guess it also cannot verify the file system or data integrity, but more so transfers through the system bus. It's a good question, never really thought about it.

[dhicks]

I'm just wondering though as to what things SMART does not detect as some kinds of mechanical failure seem to be on the list perhaps?

 

It might rather depend on the make of drive. Backblaze recently reported some long-term reliability data from their datacentres:

 

https://www.backblaze.com/blog/hard-drive-reliability-q4-2015/

 

The relevant quote would seem to be:

 

A relevant observation from our Operations team on the Seagate drives is that they generally signal their impending failure via their SMART stats. Since we monitor several SMART stats, we are often warned of trouble before a pending failure and can take appropriate action. Drive failures from the other manufacturers appear to be less predictable via SMART stats.

[sted]

Smart can't predict failure (with the motor), only highlight there's a potential issue as such. I guess it also cannot verify the file system or data integrity, but more so transfers through the system bus. It's a good question, never really thought about it.

id of thought a failing motor would have different power draw stats than a healthy one?

[Michael]

id of thought a failing motor would have different power draw stats than a healthy one?

 

Hmm possibly, but I can't see how Smart could monitor this. If you view it as an integral part of the BIOS or Chipset, then no, but in saying that I know some BIOSes do accurately measure voltages, especially on overclocked systems that allow you to monitor such data.

 

I suppose if the voltage was variable, how could it tell whether it was a PSU or hard drive issue? Maybe this is just it, it very much depends on the motherboard and its features?

[Dos_Box]

300 days? That's nothing really, I would expect that to run much longer.

 

Apparently the Samsung F1 has a MTBF of 600k hours. This one has 8880.

But lets face it. Who here would honestly (in the real world) expect a drive to last 7 years of daily use without it needing repalcing?

Has anyone here run, or is runnung, a drive (non-enterprise) for that kind of time period and not got a little worried about the need to perhaps swap it out?

Edited February 25, 2016 by Dos_Box

[Dos_Box]

Hmm possibly, but I can't see how Smart could monitor this. If you view it as an integral part of the BIOS or Chipset, then no, but in saying that I know some BIOSes do accurately measure voltages, especially on overclocked systems that allow you to monitor such data.

 

I suppose if the voltage was variable, how could it tell whether it was a PSU or hard drive issue? Maybe this is just it, it very much depends on the motherboard and its features?

 

And let's be honest. Who here really monitors their HD voltages?

For some strange reason I get the impression at least one of you is going to go 'But doesn't everyone?'!

At least with a spinning disk you get a little notification (usually) before the drive finally carks it

Once high-capacity low-cost SSD's become the norm there will be an infinately greater amount of wailing and knashing of teeth with sudden drive deaths!

[sted]

Apparently the Samsung F! has a MTBF of 600k hours. This one has 8880.

But lets face it. Who here would honestly (in the real world) expect a drive to last 7 years of daily use without it needing repalcing?

Has anyone here run, or is runnung, a drive (non-enterprise) for that kind of time period and not got a little worried about the need to perhaps swap it out?

 

i think the raid drives in my microserver are getting on for 5 year+ and im not that worried granted the only non raid drive the os failed (typical) but that was a drive i dont even remember where it came from and the drives in the raid array are standard 2tb seagates in 1 array and 2tb wd's in the other iirc it its been constantly on for years. Granted im not worried as they are in 2 raid one arrays so id have to be pretty unlucky to lose data

[sted]

And let's be honest. Who here really monitors their HD voltages?

For some strange reason I get the impression at least one of you is going to go 'But doesn't everyone?'!

At least with a spinning disk you get a little notification (usually) before the drive finally carks it

Once high-capacity low-cost SSD's become the norm there will be an infinately greater amount of wailing and knashing of teeth with sudden drive deaths!

 

fair point nobody monitors hdd voltages but it dosent seem like its something impossible for the smart firmware to do

[dhicks]

Apparently the Samsung F1 has a MTBF of 600k hours. This one has 8880.

 

Again, going on BackBlaze's statistics (if I remember correctly, I haven't looked this one up), they tended to get most drive failures either in the first year or sometime after the fith - you either get a drive with manufacturing defects that conks out after a bit of usage, or you don't and it keeps going for the best part of a decade.

[difinity]

SMART does not detect corrosion on the PCB, which causes all sorts of issues.

 

No cleaning.

 

Half cleaned

 

Cleaned

 

Use a normal pencil type rubber.

[Michael]

And let's be honest. Who here really monitors their HD voltages?

For some strange reason I get the impression at least one of you is going to go 'But doesn't everyone?'!

At least with a spinning disk you get a little notification (usually) before the drive finally carks it

Once high-capacity low-cost SSD's become the norm there will be an infinately greater amount of wailing and knashing of teeth with sudden drive deaths!

 

VoltageGeek!

 

I'm not one, just for the record

 

I got past the 'overclock make my PC all flashy phase' a very long time ago!