[SOLVED] Which SMART infos to watch for?

littlebigman · 06-03-2014, 03:56 AM

Hello

Provided SMART is a reliable way to tell how healthy a disk is... hard disks can return several infos through SMART.

Here's what I get from the two HDs on my workstation:

Code:

HD #1
01	Read Error Rate
03	Spin-Up Time
04	Start/Stop Count
05	Reallocated Sectors Count
07	Seek Error Rate
09	Power-On Hours (POH)
0A	Spin Retry Count
0B	Recalibration Retries
0C	Device Power Cycle Count
C0	Power-off Retract Count
C1	Load/Unload Cycle Count
C2	Temperature
C4	Reallocation Event Count
C5	Current Pending Sector Count
C6	Uncorrectable Sector Count
C7	UltraDMA CRC Error Count
C8	Write Error Rate / Multi-Zone Error Rate

HD #2
01	Read Error Rate
03	Spin-Up Time
04	Start/Stop Count
05	Reallocated Sectors Count
07	Seek Error Rate
09	Power-On Hours (POH)
0A	Spin Retry Count
0C	Device Power Cycle Count
B7	SATA Downshift Error Count
B8	End-to-End error / IOEDC
BB	Reported Uncorrectable Errors
BC	Command Timeout
BD	High Fly Writes (WDC)
BE	Airflow Temperature
C2	Temperature
C3	Hardware ECC Recovered
C5	Current Pending Sector Count
C6	Uncorrectable Sector Count
C7	UltraDMA CRC Error Count
F0	Head Flying Hours
F1	Total LBAs Written
F2	Total LBAs Read

Which infos are really important to watch for?

Thank you.

Ser Olmy · 06-03-2014, 10:34 PM

"Current Pending Sector Count" is the number of unreadable sectors scheduled for reallocation. The next time such a sector is written to, it becomes part of "Reallocated Sectors Count". A growing number of bad/reallocated sectors is a sure sign that the drive is going bad, so you should definitely watch both those counters.

You may want to watch the temperature as well. As a drive ages, the bearings will generate increasingly more heat due to wear. Of course, you may only be able see the trend If the ambient temperature is reasonably stable.

littlebigman · 06-04-2014, 07:46 AM

Thanks. I'll see how to read the Reallocated Sector Count from the two hard disks, and send an e-mail if it gets higher.

Ser Olmy · 06-04-2014, 08:02 AM

smartmontools can do this for you. smartd monitors selected S.M.A.R.T. attributes and send an e-mail if the value(s) change(s).

A package is most likely available for your distribution.

metaschima · 06-04-2014, 12:10 PM

All "pre-fail" attributes are important, even "old age" ones are of some importance. It is also important to run a SMART long test once in a while to check for bad blocks.

Also note that the disk can fail with normal SMART attributes so backup often.

littlebigman · 06-05-2014, 08:01 PM

Thanks for the infos.

Smartmontools doesn't work with the drives I have: It returns "Function not implemented" (although it's on Windows; Maybe it's a driver issue and they'd work find under Linux).

Emerson · 06-05-2014, 08:27 PM

Below is what I run to filter out most important data.

Code:

smartctl --all /dev/sda | grep -e "Reallocated_Sector_Ct" -e "Current_Pending_Sector" -e "Offline_Uncorrectable" -e "UDMA_CRC_Error_Count" -e "Hardware_ECC_Recovered"

metaschima · 06-05-2014, 09:17 PM

Make sure SMART is enabled in the BIOS. Sometimes it is disabled by default for some unknown reason.

littlebigman · 06-06-2014, 05:40 AM

Thanks for the tip.

Yes, it's enabled in the BIOS. For some reason, other Windows apps (SpeedFan, CrystalDiskInfo, etc.) can read SMART infos, so it's strange that smartmontools doesn't work. Oh well.