Drive Health Calculation

[jg00d3]

I use HDSentinel on all my clients computers. Usually when a drive starts to drop in health I order a replacement and return the drive w/ poor health to my vendor for warranty replacement. Lately he has been complaining about the drives coming back so I started digging into how HD Sentinel comes up w/ the drive health percentage.

I have a drive in a server right now that has these SMART Values, hex has been converted to decimal:

1,Raw Read Error Rate,6,119,95,OK,218344771,0,Enabled
3,Spin Up Time,0,94,94,OK (Always passing),0,0,Enabled
4,Start/Stop Count,20,100,100,OK,20,0,Enabled
5,Reallocated Sectors Count,36,100,100,OK,483,0,Enabled
7,Seek Error Rate,30,46,45,OK,62583001,0,Enabled
9,Power On Time Count,0,68,68,OK (Always passing),28671,0,Enabled
10,Spin Retry Count,97,100,100,OK,0,0,Enabled
12,Drive Power Cycle Count,20,100,100,OK,20,0,Enabled
184,End-to-End Error Count,99,100,100,OK,0,0,Enabled
187,Reported Uncorrectable Errors,0,100,100,OK (Always passing),0,0,Enabled
188,Command Timeout,0,100,100,OK (Always passing),0,0,Enabled
189,High Fly Writes,0,1,1,OK (Always passing),204,0,Enabled
190,Airflow Temperature,45,70,64,OK,30,0,Enabled
194,Disk Temperature,0,30,40,OK (Always passing),18; 30,0,Enabled
195,Hardware ECC Recovered,0,26,24,OK (Always passing),218344771,0,Enabled
197,Current Pending Sector Count,0,100,100,OK (Always passing),0,0,Enabled
198,Off-Line Uncorrectable Sector Count,0,100,100,OK (Always passing),0,0,Enabled
199,Ultra ATA CRC Error Count,0,200,200,OK (Always passing),0,0,Enabled

HDS has a yellow warning icon next to reallocated sectors w/ a data value of 483 reallocated sectors. HDS is showing a health of 9% for this drive, we are using the more strict option. I read through this webpage on your site http://www.hdsentinel.com/help/en/52_cond.html and can't figure out how the math works here...

If the only thing wrong w/ this drive is the 483 reallocated sectors, should the drive have a health status of 40% since the limit for the more strict reallocated sectors event is capped @ 60%? I don't even know how to plug in the seek error rate value of 62583001... What I am hoping to do is present the formula that we are using to my vendor to show him how we are predicting that these drives have problems and will possibly fail.

I made a quick excel spreadsheet and plugged in the values:
No. Data Value Weight Max % Calc'd % Used %
5 483 6 60 2898 60%
7 62583001 2 20 125166002 20%
10 0 6 40 0 0%
196 0 4 30 0 0%
197 0 4 30 0 0%
198 0 6 30 0 0%
80% Total Bad
- 100% Total Possible
Drive Health -20%
________________________________________

Another example is here:

1,Raw Read Error Rate,44,71,63,OK,000000DD828C,0,Enabled
3,Spin Up Time,0,100,100,OK (Always passing),000000000000,0,Enabled
4,Start/Stop Count,20,100,100,OK,00000000000A,0,Enabled
5,Reallocated Sectors Count,36,100,100,OK,000000000004,0,Enabled
7,Seek Error Rate,30,68,60,OK,00000063710F,0,Enabled
9,Power On Time Count,0,90,90,OK (Always passing),0000000024B0,0,Enabled
10,Spin Retry Count,97,100,100,OK,000000000000,0,Enabled
12,Drive Power Cycle Count,20,100,100,OK,00000000000A,0,Enabled
184,End-to-End Error Count,99,100,100,OK,000000000000,0,Enabled
187,Reported Uncorrectable Errors,0,100,100,OK (Always passing),000000000000,0,Enabled
188,Command Timeout,0,100,100,OK (Always passing),000000000000,0,Enabled
189,High Fly Writes,0,100,100,OK (Always passing),000000000000,0,Enabled
190,Airflow Temperature,45,77,67,OK,000021130017,0,Enabled
191,G-Sense Error Rate,0,100,100,OK (Always passing),000000000000,0,Enabled
192,Power off Retract Cycle Count,0,100,100,OK (Always passing),000000000006,0,Enabled
193,Load/Unload Cycle Count,0,100,100,OK (Always passing),00000000000A,0,Enabled
194,Disk Temperature,0,23,40,OK (Always passing),001300000017,0,Enabled
195,Hardware ECC Recovered,0,29,23,OK (Always passing),000000DD828C,0,Enabled
197,Current Pending Sector Count,0,100,100,OK (Always passing),000000000000,0,Enabled
198,Off-Line Uncorrectable Sector Count,0,100,100,OK (Always passing),000000000000,0,Enabled
199,Ultra ATA CRC Error Count,0,200,200,OK (Always passing),000000000000,0,Enabled

This drive has a value of 4 in the reallocated sectors count and has a value of 6517041 in the seek error rate, but seems to also only use the reallocated sectors count to come up w/ a health of 76% using the strict option. The strict formula on your website make sense for this drive which is under the cap if we only use the reallocated sector count to get the drive health.

Here is my excel formula which comes up w/ 76% if we don’t enter anything into the Seek Error Rate, which matches what HDS is displaying as the health for the drive:
No. Data Value Weight Max % Calc'd % Used %
5 4 6 60 24 24%
7 0 2 20 0 0%
10 0 6 40 0 0%
196 0 4 30 0 0%
197 0 4 30 0 0%
198 0 6 30 0 0%
24% Total Bad
- 100% Total Possible
Drive Health -76%

________________________________________

The last example here is a drive w/ a strict health of 7%, here are the values:

1,Raw Read Error Rate,6,118,67,OK,196004477,0,Enabled
3,Spin Up Time,0,95,95,OK (Always passing),0,0,Enabled
4,Start/Stop Count,20,100,100,OK,316,0,Enabled
5,Reallocated Sectors Count,36,97,97,OK,127,0,Enabled
7,Seek Error Rate,30,86,60,OK,425947493,0,Enabled
9,Power On Time Count,0,68,68,OK (Always passing),28666,0,Enabled
10,Spin Retry Count,97,100,100,OK,0,0,Enabled
12,Drive Power Cycle Count,20,100,100,OK,327,0,Enabled
187,Reported Uncorrectable Errors,0,1,1,OK (Always passing),17361,0,Enabled
189,High Fly Writes,0,100,100,OK (Always passing),0,0,Enabled
190,Airflow Temperature,45,57,54,OK,63; 43,0,Enabled
194,Disk Temperature,0,43,46,OK (Always passing),15; 43,0,Enabled
195,Hardware ECC Recovered,0,59,44,OK (Always passing),118285825,0,Enabled
197,Current Pending Sector Count,0,100,100,OK (Always passing),2,0,Enabled
198,Off-Line Uncorrectable Sector Count,0,100,100,OK (Always passing),2,0,Enabled
199,Ultra ATA CRC Error Count,0,200,200,OK (Always passing),0,0,Enabled
200,Write Error Rate,0,100,253,OK (Always passing),0,0,Enabled
202,Data Address Mark Errors,0,94,247,OK (Always passing),6,0,Enabled

The Reallocated Sectors Count is 127 which is over the 60% cap. The Seek Error Rate value is 425947493 which HDS says is OK? Spin Retry is 0, Drive doesn’t have the Reallocation event count value, Current Pending Sector Count is 2, and the Off-Line uncorrectable sectors count value is 2.

I made a quick excel spreadsheet and plugged in the values:
No. Data Value Weight Max % Calc'd % Used %
5 127 6 60 762 60%
7 425947493 2 20 851894986 20%
10 0 6 40 0 0%
196 0 4 30 0 0%
197 2 4 30 8 8%
198 2 6 30 12 12%
100% Total Bad
- 100% Total Possible
Drive Health 0%

Again, I can’t figure out how the math works, according to the way I am interpreting the formula this drive should be dead, but HDS says it has 7% health…

Thanks in advance for you help with this, I am sure it is just something I am overlooking….

[hdsentinel]

Thanks for your message and questions.

First of all, you need to know that the vendor / reseller / manufacturer should replace the drive (in terms of warranty) only if the number of problems are high enough and reaches the threshold set by the manufacturer. That point the health value is 0 %.
Before that point, you may see "At this point, warranty replacement of the disk is not yet possible, only if the health drops further." in the text description.
Of course, you may try to ask for warranty replacement before that, but in general, your vendor may refuse replacement.

(most hard disks do not reach this value and fail before that, while all data may be lost - but it is an other question. See http://www.hdsentinel.com/smart for more details).

Regardless of the actual health value, it may be important to verify if the hard disk status is stable. There are numeruous tests available in Hard Disk Sentinel to verify if the disk is stable or if there is an error which require further actions to repair or if the recommended action is to ask for replacement.
Please check http://www.hdsentinel.com/faq.php#health and http://www.hdsentinel.com/faq.php#tests for more information.


Back to Drive Health Calculation:

You did not mention the hard disk type, model, revision for the drives. Without seeing the complete information about the drives (the listed S.M.A.R.T. attributes are partial information), it is hard to say more about the actual situation.

I agree, that in the Help the most general health calculation method was described. That help topic was written in 2007. Since that, the fundamentals are still the same but really lots of model-specific calculations are added (and also different calculation method is used for different SSD devices - but it is an other question).

Let's see the drives with more details:

Drive #1:

If the only thing wrong w/ this drive is the 483 reallocated sectors, should the drive have a health status of 40% since the limit for the more strict reallocated sectors event is capped @ 60%?

I agree. However, due to the higher number of problems (more reallocated sectors), the health drops further. There is a bit more complicated (non-linear) calculation applied if the number of errors is higher.

I don't even know how to plug in the seek error rate value of 62583001...

This value is not an error counter for this drive (this is why I wrote that it would be nice to know more details about the type/model).
HDSentinel only uses this attribute to decrease health if it would actually reflect error count. See the "Flags" in the bottom right area when this attribute is selected. If this attribute would really reflect an error, then Hard Disk Sentinel would display that in the text description also.

What I am hoping to do is present the formula that we are using to my vendor to show him how we are predicting that these drives have problems and will possibly fail.

The problem is that even if you can show the formula, the formula is used in Hard Disk Sentinel to warn you to prevent data loss and take the required actions. However, this may not be enough for the vendor to replace the drive as the vendor and manufacturer does not care about your important data.


Drive #2:

This drive has a value of 4 in the reallocated sectors count and has a value of 6517041 in the seek error rate, but seems to also only use the reallocated sectors count to come up w/ a health of 76% using the strict option. The strict formula on your website make sense for this drive which is under the cap if we only use the reallocated sector count to get the drive health.

Absolutely correct.
Even if you see 6517041 in "seek error rate", this does NOT mean that there are so much errors reported because on this drive that value is not an error counter (this is why called "error rate", instead of "error count"). If there would be problems with this attribute, Hard Disk Sentinel would mention in the text description and the health would be much lower.


Drive #3:

The Reallocated Sectors Count is 127 which is over the 60% cap. The Seek Error Rate value is 425947493 which HDS says is OK?

Yes - as this high number is also not the count of errors.

Again, I can’t figure out how the math works, according to the way I am interpreting the formula this drive should be dead, but HDS says it has 7% health…

Excuse me, but as I see, you ADDED the (negative) values which decrease the health. However, this is not true as the health values need to be multiplied to get the total health value. For example, if you'd have an attribute which (alone) results in 90 % and an other which (alone) results in 80% (10 and 20% decrease), then the total health is 90% x 80% = 72% (not 70% if you simply add the negative values). Of course you can't get negative health in any ways.
Please check the example at the end of www.hdsentinel.com/smart , it describes the used formula to calculate the health (describing the values and the formula also).

Excuse me if something is confusing. I can make sure the health value calculation is constantly improved as new and new hard disk (and SSD) models are released by manufacturers to make the proper calculation. This means the use of all known and unknown attributes, so the information in the Help is only describing the basic calculation for simple situations (for example, if the drive has less problems).

Please let me know if you have any questions or I can help or clarify. If you prefer, you may use Report -> Send test report to developer option on any drive(s) so I can check the complete information and may advise.

[celavey]

I don't quite understand how to configure and identify the health of hard drives via HD Sentinel.

[hdsentinel]

Hard Disk Sentinel automatically calculage the health based on various factors. Please check Help -> Appendix -> Health Calculation for complete information about how the health value is determined.