Buying a used SATA disk

With the evolution of SSD drives, people are selling their old magnetic disks on Ebay or other platforms really cheap. Here are some steps to take after plugging in a bought SATA drive into your Linux system.

Keep in mind that all disk information probably can be manipulated, including model name, serial number, firmware, etc.

Check general drive information

# hdparm -I /dev/sdx
- Model Number
- Serial Number (to identify physical drive e.g. in case of replacement)
- Nominal Media Rotation Rate
- DMA: udma6
- Write cache
- SMART error logging
- SMART self-test
- SCT Error Recovery Control (if used in a RAID array)
- Security: Passwort not enabled/locked
(Enabled features are preceded by *)

Check SMART capabilities

# smartctl -i /dev/sdx
- SMART support is: Availabe
- SMART support is: Enabled (if not, enable it with "smartctl -s on /dev/sdx")

Check detailed SMART information

# smartctl -a /dev/sdx
- Model Family/Device Model
- User Capacity
- Rotation Rate
- SATA Version: current speed
- Vendor specific SMART attributes:
o Start_Stop_Count
    (Usually the same as Power_Cycle_Count)
o Reallocated_Sector_Ct
    (Bad sectors that have been marked by the disk?)
o Power_On_Hours
    (Disk has been used 24/7 as a NAS drive?)
o Power_Cycle_Count
    (Usually the same as Start_Stop_Count)
o G-Sense_Error_Rate
   (Disk has been dropped on the floor?)
o Load_Cycle_Count
   (Usually the same as Start_Stop_Count and Power_Cycle_Count)
o Temperature_Celsius
- SMART Error Log (Are there any entries?)
- SMART Self-test (Anything other than "Completed without error")

Temperature history and SCT

# smartctl -x /dev/sdx
- Temperature history
- SCT Error Recovery Control
    (Only important for use in RAID arrays, see one of my previous posts)

SMART tests

SMART tests do not degrade drive performance, they are more like collecting statistical data from the drive. Online and offline tests can be executed during normal operation.

# smartctl -t long /dev/sdx
Expected output:

root@linux:~# smartctl -t long /dev/sdx

=== START OF OFFLINE IMMEDIATE AND SELF-TEST SECTION ===
Sending command: "Execute SMART Extended self-test routine immediately in off-line mode".
Drive command "Execute SMART Extended self-test routine immediately in off-line mode" successful.
Testing has begun.
Please wait 103 minutes for test to complete.
Test will complete after Fri May 20 09:48:56 2016

Use smartctl -X to abort test.

Check test result in drive logs:

# smartctl -l selftest /dev/sdx
Expected output:

=== START OF READ SMART DATA SECTION ===
SMART Self-test log structure revision number 1
Num Test_Description Status Remaining LifeTime(hours) LBA_of_first_error
# 1 Extended offline Completed without error 00% 37143 -

# smartctl -l error /dev/sdx
Expected output:

=== START OF READ SMART DATA SECTION ===
SMART Error Log Version: 1
No Errors Logged

Conclusion

So what are you doing if some of the values are not looking right? Don't worry. The drive might still be working without problems for one or two years. But you should have an eye on it:

  • Run regular SMART tests to see if error rate / reallocated sector count increases.
  • Don't use the drive as a sole disk medium for critical high performance data. Maybe make it part of a RAID 1 or RAID 6 array or use it as a hotspare / cold standby drive. Even then you should run regular SMART tests on the drive.
  • Make sure that the temperature is not too high (should be somewhere around 40 degrees celcius and is dependent on the drive).
  • Minimize power cycles. Using a worn out disk on a PC or laptop that gets rebooted a couple of times every day is not a good idea.
  • If you can afford it, use more file system and RAID caching to minimize disk reads and writes. RAID controllers usually support writethrough and writeback. While writeback minimizes disk writes, it should only be used on battery backed or flash backed RAID controllers. Don't use software RAID or fake RAID controllers.
  • On Linux there is a tool called iotop to identify processes with heavy read/write operations. Reconfigure your system to use different disks.
  • Run the disk for a couple of days without any important data on it. Check SMART values and see if you are comfortable with it.
  • Make frequent backups of your data to prepare for disk failure. Don't use the drive as a backup medium.
  • Don't use the drive in high availability environments (with or without RAID). If you use the drive in your laptop or a PC without RAID, make sure to have a spare drive at hand and make daily backups.

Why should I go through all this and not buy a new drive? At least its more reliable and the higher price will pay off.

Even new drives might fail within a couple of days or weeks without any prior signs. There is no guarantee that a drive - new or old - will not stop working from one second to the next. Of course older drives are more likely to fail than new drives (see MTBF / load/unload cycles / power-on hours / warranty duration of your drive specification data-sheet).

But if you prepare carefully for disk failures and minimize the risk, you can save some money and spend it for that brand new SSD drive that will be out on the market in one year. SSD technology is progressing rapidly and it might be worth waiting for the prices to drop.

Share