Hard Disk Monitoring

Hello everybody

We have replaced some Hitachi drives from a machine with Seagate ones. Apparently the Hitachi's have a different way to count the Raw_Read_Error_Rate parameter. While on Hitachi's the raw value is zero or maybe 1 (value=100), on Seagate's the raw value is huge all the time and the "value" reported by SMART table is 80ish.

I have been searching on the Internet and it seems that Seagate's have a different approach to the matter. The same, if I'm not mistaken, happens with Seek_Error_Rate and Hardware_ECC_Recovered, that show huge numbers.

Is there any official paper saying something? A manufacturer is used to considering 75 as a limit for a "good" drive. Seagate's have all 80ish as starting point so, following this approach, all Seagate's drives have to be replaced!

Could someone shed some light on the matter?

Many thanks in advance!

Antonio

Hi Antonio,

Manufacturers do not provide such "official" information about S.M.A.R.T. feature of the drive.
If you read that a manufacturer says 75 for the threshold value - it means ONLY for that manufacturer - or even more specifically to that particulat model because manufacturers may use different methods to set up the values/thresholds for their drives.

This is why this approach (checking value/thresholds) is wrong because manufacturers sometimes use "tricks" to show their drives better than others. So a manufacturer-independent method is better (which is used in Hard Disk Sentinel) to verify the real disk status. More information: www.hdsentinel.com/smart

You are correct: Seagate drives use these attributes (Raw Read Error Rate, Seek Error Rate, Hardware ECC Recovered) as counters for ALL such operations/events and the "value" field reflects the RATE of errors compared to the total count of such operations/events.
That's why the raw counter is very huge (and constantly increasing) while the "value" field remains stable (until there is a problem when the "value" field drops down).

If you can use the Report menu Send test report to developer option, we can examine the complete status of the drive.
You may use the "Verify online" button in Hard Disk Sentinel (a hard disk with a globe) to compare the drive status with other drives of the same model to check if that drive is better/worse than the average.

Hi HdSentinel

Thanks for your help, it's great to have such an experienced help.

I've read your page about SMART. There is still something that is unclear for me. First, where the "threshold" value comes from? Is it stored on the HD by the manufacturer? Secondly, is the DATA value calculated from the RAW value with a formula that comes from the manufacturer and is stored on the HD?

If I correctly understand, the RAW data is the unprocessed date from the HD; Value is calculated from RAW with a formula provided by the manufacturer; Worst is just the worst "value" data.

And: is the "value" always decreasing? Have a look at this
ID# ATTRIBUTE_NAME FLAG VALUE WORST THRESH TYPE UPDATED WHEN_FAILED RAW_VALUE
7 Seek_Error_Rate 0x000f 100 253 030 Pre-fail Always - 129645

This comes from a different system, but I've checked on HDSentinel and it says the same: the worst value is HIGHER than the VALUE. Does it make sense? What does it mean?

Thanks for your precious help.

Antonio

You're welcome, I'm glad to help.

Yes, you are correct: the "threshold" number is stored on the HDD, it is specified by the manufacturer. Different manufacturers uses different such numbers (even different models of the same manufacturer may have different thresholds for a specific attribute).

Yes, you are also correct: the "value" field is calculated by a special formula also made by the manufacturer (which may change on different models and even different firmware versions for a particular model). The DATA (RAW number) field is the source of the "value".

tony359 wrote:If I correctly understand, the RAW data is the unprocessed date from the HD; Value is calculated from RAW with a formula provided by the manufacturer; Worst is just the worst "value" data.

Absolutely correct.

tony359 wrote:And: is the "value" always decreasing? Have a look at this
ID# ATTRIBUTE_NAME FLAG VALUE WORST THRESH TYPE UPDATED WHEN_FAILED RAW_VALUE
7 Seek_Error_Rate 0x000f 100 253 030 Pre-fail Always - 129645

In theory yes, the "value" field should decrease as the drive is getting worse and worse.
However, the "value" field may increase back (that's why the "worst" field is designed, to keep the the lowest (worst) "value").

You pointed on one of the numerous problems with this aspect: some drives do not update the "worst" and even the "value" numbers as should. This way the user would miss some problems as these numbers do not react quickly for new problems with the drives.

Another example is shown at the end of http://www.hdsentinel.com/smart/ where the (failed) drive would not show any problems in the "value" and "worst" fields, but the problem could be detected well by checking the RAW (data) field of each attributes.

Thanks again.

It seems that to understand properly SMART attributes you definitely need to have some knowledge about the drive you are going to analise.
Thanksfully we manage just two types of drives at the moment so it should be easy to spot weird behaviors.

I've just realise that comparing the drives online I contribute to your database, I will compare all the drives I come across in the future to improve your data.

Again, thanks

Antonio