Is SMART a reliable predictor of SSD failures?

[OrionMaster]

I ask because Windows, OSX and some third party HD health tools are bitching that my Macbook Pro Late 2013 15' SSD is going to fail (based on SMART data). So far it has been 2 months and no issue has happened. Even the benchmarks have not changed

 

[dave_the_nerd]

Depends on the error it's throwing. Post a screenshot of the smartctl info?

In my experience, it's usually pretty reliable. If smart thinks your drive is going, it'll almost always go. The flipside is, plenty of drives go without throwing a smart error. So, back up, back up, back up.

 

[Jeff7]

Similar to above:

My experience: Hard drives can fail abruptly without SMART throwing a warning.
But if a drive's own diagnostics stuff is detecting that something's wrong, and you place any value at all on what's on the drive, back up and replace it. If it's under warranty, you should be able to get a replacement due to a SMART error.

Sometimes it's because the number of new bad sectors has exceeded the internal reallocation threshold. Maybe it's done developing bad sectors and you'll get a few more years out of it. I wouldn't want to chance it though.

 

[tcsenter]

Depends on the mode of failure.

 

[Elixer]

Suffice to say that when the drive's SMART is saying something is wrong, then something is wrong.

Wish they would have more diagnostic SMART modes available though.

Backup ASAP, then get a new drive, and put the current one as a scratch drive.

 

[OrionMaster]

Thats the thing, sometimes the programs like HD Sentinal, Gsmartcontrol, Crystal Disk Info will say everything is fine, sometimes they will complain about that data. When they complain, its usually something like this

 

[Elixer]

Whoops, I missed the part where you said it was a SSD.

In the above situations, I tend to clone the SSD, then do a secure erase on it, then apply the clone back to the SSD.

Is there any indication on who makes this SSD?
With Crystaldiskinfo, it also tells you how much data you have written to it, what does that show?

 

[OrionMaster]

Crystaldiskinfo doesn't show me such info

 

[John Connor]

Crystaldiskinfo should show SMART data...

I don't understand the pre-failures because the threshold was not met.

 

[VirtualLarry]

I don't understand the pre-failures because the threshold was not met.

pre-failure doesn't mean that it anticipates a failure. Just means that SMART parameter is of type "pre-fail", meaning a failure indicator if it triggers the threshold.

 

[nanaki333]

how old is your SSD? is it from 2013 too? the error looks like it is saying your write cycles may be nearing the max. if that's the case then it WILL function but it is letting you know you are on borrowed time until you reach the finite number of cycles it supports (a drive that claims 100TB may actually go until 800TB, but there is no guarantee).

 

[OrionMaster]

2 years of use

 

[Elixer]

2 years of use

How much use we talking about though?
20 GB/ of writes a day, 100GB/day more?

Was there any utility that came with that SSD that shows that info?
Usually, that info is stored in the SMART data, but, you didn't post a screen shot from crystaldiskinfo so we can look.

 

[OrionMaster]

I'm not sure how many reads and writes, its normal business use. maybe a bit above average
here is crystal disk info

 

[Cerb]

It's vendor-specific, they don't have the standard attributes for wear, and I was unable to find any documentation for attribute AD/173. Documentation for SMART on the drive should be enough to get this figured out.

 

[OrionMaster]

It's vendor-specific, they don't have the standard attributes for wear, and I was unable to find any documentation for attribute AD/173. Documentation for SMART on the drive should be enough to get this figured out.

do you know where I can find that? this is the drive
http://www.amazon.com/gp/product/B00YWMDVF8/ref=s9_hps_bw_g147_i7

 

[Elixer]

I'm not sure how many reads and writes, its normal business use. maybe a bit above average
here is crystal disk info

Thanks, I think that provides us more info...

http://forum.hddguru.com/viewtopic.php?t=31501&f=10&start=0#p220231

So, it looks like a samsung unit, and reading that thread more, like this, http://forum.hddguru.com/viewtopic.php?t=31501&f=10&start=0#p220354
it seems that it is a firmware bug.

In other words, I wouldn't worry that much about it, just make sure you have backups.

I still would do a secure erase on it, if it acts up though.

 

[OrionMaster]

Thanks, I think that provides us more info...

http://forum.hddguru.com/viewtopic.php?t=31501&f=10&start=0#p220231

So, it looks like a samsung unit, and reading that thread more, like this, http://forum.hddguru.com/viewtopic.php?t=31501&f=10&start=0#p220354
it seems that it is a firmware bug.

In other words, I wouldn't worry that much about it, just make sure you have backups.

I still would do a secure erase on it, if it acts up though.

Yes, thats what I was thinking, that the error wasn't real