Deeper research of broken HDD
Date: November 09, 2020
Introduction
In the previous article, I got an access to data on failing HDD and even temporarily fixed NTFS table. This HDD was gifted for tech experiments, and it was a nice occasion to make a deeper diagnostics of the device. Drive SMART before the research looked like this.
Preparing tools
I was surprised to notice, that there aren’t many tools for HDD testing, especially for Linux. Some of them (Victoria or MHDD) are based on MS-DOS and were written over 15 years ago, but still usable. In order to use MHDD, PC should support running hard drives (even SATA ones) in IDE legacy mode. As a testing platform, MS-DOS is recommended by the software author. However, I managed to run it from FreeDOS, an open-source DOS-compatible environment.
In order to create bootable pen drive, I used the command sudo dd if=PATH_TO_FD12LITE.img of=/dev/sda status=progress && sync. The first partition is a bootable one from the image, and the second partition was created manually to copy MHDD files and to store its logs. Since the program was written many years ago, last release dates back to 2005, I made a small FAT32 partition for data storage.
Starting the research
After the bootable device is created and FreeDOS is launched, your should choose the language and select option to run without installation “No, return to DOS”.
The interface of MHDD is quite scary for new users and looks like this. However, there is an instruction how to use it.
Step numer one is connecting HDD to MHDD. First your should type Shift + F3 to select HDD for scanning.
Another step was simple scan of HDD surface to check for bad blocks, without parameters. The result of SCAN command is visible on HDD screen and is written to LOG/MHDD.LOG file.
6.03.2012 19:15:47 MHDD>SCAN
6.03.2012 19:15:53 Scan started
6.03.2012 19:15:53 MODE: IDE
6.03.2012 19:15:53 Device: ST1000LM024 HN-M101MBB
6.03.2012 19:15:53 -------------------------------
6.03.2012 19:15:53 Lap : 1
6.03.2012 19:15:53 LBA scan: 0 to 1953525167
6.03.2012 19:18:46 þ LBA warning: 37671660
6.03.2012 19:19:16 þ LBA warning: 44103015
6.03.2012 19:19:16 þ LBA Warning: 44103270
6.03.2012 19:19:38 þ LBA Error: 48350664
6.03.2012 19:19:41 þ LBA Error: 48482768
...
6.03.2012 22:47:42 Time spent: 03:31:46
6.03.2012 22:47:42 Blocks < 3ms = 7560949
6.03.2012 22:47:42 Blocks < 10ms = 749635
6.03.2012 22:47:42 Blocks < 50ms = 1382
6.03.2012 22:47:42 Blocks < 150ms = 36
6.03.2012 22:47:42 Blocks < 500ms = 19
6.03.2012 22:47:42 Blocks > 500ms = 6
6.03.2012 22:47:42 Errors: 32, Warnings: 25
6.03.2012 22:47:42 Done
After the scan I fully erased the HDD via command ERASE. It was necessary since sometimes HDD generates soft bad blocks, which were caused by OS errors, but not hardware fault of those sectors. It wasn’t quite my story (since tested HDD had 0 bad blocks, but continuos read errors), still it was worth trying.
SMART (F8 shortcut) before HDD erase.
6.03.2012 19:04:12 MHDD>SMART ATT
6.03.2012 19:04:12 HDD: ST1000LM024 HN-M101MBB
6.03.2012 19:04:12 --------------------------------------------------------
6.03.2012 19:04:12 SMART attributes:
6.03.2012 19:04:12 Name Val Worst Raw
6.03.2012 19:04:12 Att # 1 : Read error rate : 100 100 4927
6.03.2012 19:04:12 Att # 5 : Reallocated sectors count : 252 252 0
6.03.2012 19:04:12 Att # 9 : Power-on time : 100 100 10604
6.03.2012 19:04:12 Att # 194 : HDA Temperature : 60 46 40
6.03.2012 19:04:12 Att # 195 : Hardware ECC recovered : 100 100 0
6.03.2012 19:04:12 Att # 196 : Reallocate event count : 252 252 0
6.03.2012 19:04:12 Att # 197 : Current pending sectors : 99 99 230
6.03.2012 19:04:36 Att # 200 : Write error rate : 100 100 25719
6.03.2012 19:06:27 MHDD>ERASE
6.03.2012 19:06:28 ST1000LM024 HN-M101MBB LBA: 1,953,525,168
SMART #200 Write error rate didn’t change after the erase, as well as other parameters, and I started scan procedure with erase option. This option tries to fill in with zeros those blocks, which are unreadable.
6.03.2012 19:09:37 MHDD>SCAN
6.03.2012 19:09:40 Scan started
6.03.2012 19:09:40 MODE: IDE
6.03.2012 19:09:40 ERASE DELAYS: ON, TIME=350 msec
6.03.2012 19:09:40 Device: ST1000LM024 HN-M101MBB
6.03.2012 19:09:40 -------------------------------
6.03.2012 19:09:40 Lap : 1
6.03.2012 19:09:40 LBA scan: 0 to 1000000000
6.03.2012 19:12:07 Erase 255 sectors starting from 32136885
6.03.2012 19:12:11 Erase 255 sectors starting from 32136885
6.03.2012 19:12:14 þ LBA Error: 32137140
...
6.03.2012 19:16:29 Erase 255 sectors starting from 32152185
6.03.2012 19:16:29 þ LBA Timeout: 32152185
6.03.2012 19:16:29 Last scanned LBA: 32152439
6.03.2012 19:16:29 Blocks erased with EraseWaits: 72
6.03.2012 19:16:29 Blocks < 3ms = 125868
6.03.2012 19:16:29 Blocks < 10ms = 161
6.03.2012 19:16:29 Blocks < 50ms = 5
6.03.2012 19:16:29 Blocks < 150ms = 2
6.03.2012 19:16:29 Blocks < 500ms = 5
6.03.2012 19:16:29 Blocks > 500ms = 0
6.03.2012 19:16:29 Errors: 11, Warnings: 5
6.03.2012 19:16:29 Done
6.03.2012 19:16:32 ST1000LM024 HN-M101MBB LBA: 1,953,525,168
In a while I continued the scan and the result looked like this.
6.03.2012 19:16:54 MHDD>SCAN
6.03.2012 19:17:06 Scan started
6.03.2012 19:17:06 MODE: IDE
6.03.2012 19:17:06 ERASE DELAYS: ON, TIME=350 msec
6.03.2012 19:17:06 Device: ST1000LM024 HN-M101MBB
6.03.2012 19:17:06 -------------------------------
6.03.2012 19:17:06 Lap : 1
6.03.2012 19:17:06 LBA scan: 32146065 to 1000000000
6.03.2012 19:17:08 Erase 255 sectors starting from 32146065
6.03.2012 19:17:20 Erase 255 sectors starting from 32146065
6.03.2012 19:17:20 þ LBA Timeout: 32146065
6.03.2012 19:17:20 þ LBA Warning: 32146320
6.03.2012 19:17:21 þ LBA Error: 32147850
6.03.2012 19:17:21 þ LBA Warning: 32148105
6.03.2012 19:17:21 Erase 255 sectors starting from 32148360
6.03.2012 19:17:25 Erase 255 sectors starting from 32148360
6.03.2012 19:17:25 þ LBA Warning: 32148615
6.03.2012 19:17:26 Erase 255 sectors starting from 32152440
6.03.2012 19:17:34 Erase 255 sectors starting from 32152440
6.03.2012 19:17:36 þ LBA Warning: 32152695
6.03.2012 19:17:37 þ LBA Warning: 32154735
6.03.2012 19:25:59 Erase 255 sectors starting from 140557530
6.03.2012 19:26:04 Erase 255 sectors starting from 140557530
6.03.2012 19:26:07 þ LBA Error: 140557785
6.03.2012 19:26:07 Erase 255 sectors starting from 140558040
6.03.2012 19:26:11 Erase 255 sectors starting from 140558040
...
6.03.2012 19:31:32 Erase 255 sectors starting from 182151345
6.03.2012 19:31:33 þ LBA Warning: 182152875
6.03.2012 19:31:34 Erase 255 sectors starting from 182153640
6.03.2012 19:31:37 Erase 255 sectors starting from 182153640
6.03.2012 20:41:10 Blocks erased with EraseWaits: 44
6.03.2012 20:41:12 Time spent: 01:24:03
6.03.2012 20:41:12 Blocks < 3ms = 3566480
6.03.2012 20:41:12 Blocks < 10ms = 228878
6.03.2012 20:41:12 Blocks < 50ms = 103
6.03.2012 20:41:12 Blocks < 150ms = 10
6.03.2012 20:41:12 Blocks < 500ms = 8
6.03.2012 20:41:12 Blocks > 500ms = 0
6.03.2012 20:41:12 Errors: 5, Warnings: 8
6.03.2012 20:41:12 Done
I have noticed a weird thing, sector addresses of bad blocks were absolutely different from bad blocks at the first scan. For example there haven’t been any problems with reading blocks after 100,000,000. Overall amount of errors was smaller than during first scan. Since I knew the position of badly readable blocks, I decided to try remapping them.
6.03.2012 20:52:38 MHDD>SCAN
6.03.2012 20:53:14 Scan started
6.03.2012 20:53:14 MODE: IDE
6.03.2012 20:53:14 REMAP: ON
6.03.2012 20:53:14 Device: ST1000LM024 HN-M101MBB
6.03.2012 20:53:14 -------------------------------
6.03.2012 20:53:14 Lap : 1
6.03.2012 20:53:14 LBA scan: 0 to 182146760
6.03.2012 20:55:40 þ LBA Error: 32137140
6.03.2012 20:59:01 À> Remap try...
6.03.2012 20:59:01 þ LBA Error: 32137141
6.03.2012 21:02:22 À> Remap try...
6.03.2012 21:02:22 þ LBA Error: 32137142
6.03.2012 21:05:42 À> Remap try...
...
6.03.2012 22:16:09 À> Remap try...
6.03.2012 22:16:09 þ LBA Error: 32137164
6.03.2012 22:19:29 À> Remap try...
6.03.2012 22:19:30 þ LBA Timeout: 32137165
6.03.2012 22:19:30 Last scanned LBA: 32137419
6.03.2012 22:19:30 Blocks < 3ms = 125867
6.03.2012 22:19:30 Blocks < 10ms = 147
6.03.2012 22:19:30 Blocks < 50ms = 13
6.03.2012 22:19:30 Blocks < 150ms = 1
6.03.2012 22:19:30 Blocks < 500ms = 0
6.03.2012 22:19:30 Blocks > 500ms = 0
6.03.2012 22:19:30 Errors: 25, Warnings: 0
6.03.2012 22:19:30 Done
Seems that HDD had a huge problem with remapping bad blocks, which is a scary sign. However, HDD was doing well with writing data to there blocks, since ERASE command didn’t cause any new SMART errors.
SMART showed following parameters:
6.03.2012 23:26:35 MHDD>SMART ATT
6.03.2012 23:26:35 Getting SMART attributes...
6.03.2012 23:26:35 SMART READ ATTRIBUTES
6.03.2012 23:26:35 HDD: ST1000LM024 HN-M101MBB
6.03.2012 23:26:35 --------------------------------------------------------
6.03.2012 23:26:35 SMART attributes:
6.03.2012 23:26:35 Name Val Worst Raw
6.03.2012 23:26:35 Att # 1 : Read error rate : 100 100 5270
6.03.2012 23:26:35 Att # 5 : Reallocated sectors count : 252 252 0
6.03.2012 23:26:35 Att # 9 : Power-on time : 100 100 10615
6.03.2012 23:26:35 Att # 194 : HDA Temperature : 56 45 44
6.03.2012 23:26:35 Att # 196 : Reallocate event count : 252 252 0
6.03.2012 23:26:35 Att # 197 : Current pending sectors : 100 99 24
6.03.2012 23:27:15 Att # 200 : Write error rate : 100 100 25719
The amount of pending sectors (ones, that could not be read) has decreased, but still there weren’t any bad ones.
Hard disk was turned off, and in a while SCAN option was launched again. Result was just unpredictable, 0 read errors.
6.03.2012 20:08:20 MHDD>SCAN
6.03.2012 20:08:41 Scan started
6.03.2012 20:08:41 MODE: IDE
6.03.2012 20:08:41 ERASE DELAYS: ON, TIME=350 msec
6.03.2012 20:08:41 Device: ST1000LM024 HN-M101MBB
6.03.2012 20:08:41 -------------------------------
6.03.2012 20:08:41 Lap : 1
6.03.2012 20:08:41 LBA scan: 0 to 1000000000
6.03.2012 21:32:18 Blocks erased with EraseWaits: 0
6.03.2012 21:32:19 Time spent: 01:23:36
6.03.2012 21:32:19 Blocks < 3ms = 3692574
6.03.2012 21:32:19 Blocks < 10ms = 228916
6.03.2012 21:32:19 Blocks < 50ms = 74
6.03.2012 21:32:19 Blocks < 150ms = 5
6.03.2012 21:32:19 Blocks < 500ms = 0
6.03.2012 21:32:19 Blocks > 500ms = 0
6.03.2012 21:32:19 No warnings, no errors
6.03.2012 21:32:19 Done
And SMART still showed zero reallocated rectors.
6.03.2012 21:38:19 MHDD>SMART ATT
6.03.2012 21:38:19 Getting SMART attributes...
6.03.2012 21:38:19 SMART READ ATTRIBUTES
6.03.2012 21:38:19 HDD: ST1000LM024 HN-M101MBB
6.03.2012 21:38:19 --------------------------------------------------------
6.03.2012 21:38:19 SMART attributes:
6.03.2012 21:38:19 Name Val Worst Raw
6.03.2012 21:38:19 Att # 1 : Read error rate : 100 100 5345
6.03.2012 21:38:19 Att # 5 : Reallocated sectors count : 252 252 0
6.03.2012 21:38:19 Att # 9 : Power-on time : 100 100 10625
6.03.2012 21:38:19 Att # 194 : HDA Temperature : 43 43 57
6.03.2012 21:38:19 Att # 196 : Reallocate event count : 252 252 0
6.03.2012 21:38:19 Att # 197 : Current pending sectors : 252 99 0
6.03.2012 21:38:28 Att # 200 : Write error rate : 100 100 25719
As a matter of experiment, I installed clean Debian-based system on this HDD, and in a while I could see errors during OS boot and an increased number of errors in SMART.
6.03.2012 19:11:15 MHDD>SMART ATT
6.03.2012 19:11:15 Getting SMART attributes...
6.03.2012 19:11:15 SMART READ ATTRIBUTES
6.03.2012 19:11:15 HDD: ST1000LM024 HN-M101MBB
6.03.2012 19:11:15 --------------------------------------------------------
6.03.2012 19:11:15 SMART attributes:
6.03.2012 19:11:15 Name Val Worst Raw
6.03.2012 19:11:15 Att # 1 : Read error rate : 100 100 5349
6.03.2012 19:11:15 Att # 9 : Power-on time : 100 100 10635
6.03.2012 19:11:19 Att # 200 : Write error rate : 100 100 25756
Conclusion of the research
- This HDD seem to have a problem with reading blocks at different sectors, which looks like problem with read/write heads.
- Hard disks can get data from read-error sectors after device reboot or during next scans, which means that full disk image can be created via
dd_rescuevia multiple disk reads even without a costly replacement of read-write head from another HDD of that exact model. - NEVER use broken HDD to store data due to potential data loss. This HDD is unstable, causes system files errors on clean OS, and its future usage could cause complete failure and data loss.
Always make backups and check the state of your HDD, and the risk of data loss will be drastically decreased!