Is it possible to save data from a damaged HDD that I still have access to?
Whenever I try to copy the data I see HDD-related errors such as "I/O error", "input/output error".
Currently, the faulty HDD is connected to a USB port.
HISTORY
Last April 19th I was surprised when suddenly my server in which I host a game was disconnected.
At first, I thought it was only an intermittent connection issue that soon would be reestablished.
One hour later the hosting company told me that the HDD was damaged and they would need to replace it. Next, they restarted Linux in recovery mode and advised me to backup all my data.
When attempting to copy data I noticed that some files simply wouldn't be copied.
My whole database size is 5GB.
I could copy most of the tables, and repair the corrupt ones following the manual. (basically: myisamchk -r table_names)
But, there are a few files I can't extract off the faulty hard drive, thus can not repair those tables. The files that I can not extract are .MYD ones (the ones which store the data).
The hosting company offers only paid backup services. I do have a backup, but it's 6 months old. It's MySQL 5.0 because it's legacy software.
I have read https://stackoverflow.com/questions/879176/how-to-recover-mysql-database-from-myd-myi-frm-files already. The tables that I could extract from the hard drive have been repaired, the problem are those that I couldn't.
ATTEMPTS
While the HDD has been damaged, I can still browse the files on it. It's currently connected to a USB port and I'm trying to copy the files I need.
The commands I have tried so far:
hexdump
hexdump table.MYD > dump-table.hex - it could only dump half of the file.
cp
cp source destination copy half of the file and fails:
cp: error reading '/media/usb/usr/local/mysql-5.0.96-linux-x86_64-glibc23/data/DB/Helmet.MYD': Input/output error
dd
dd if=/input of=/output conv=noerror bs=128k - could copy the two halves of the file, but the final file would be corrupted, and thus missing some of the data. Even though I could repair the file with myisamchk -q table, it would lose thousands of records.
One interesting thing about this problem is that only .MYD files seem to have been affected, and that they are damaged in a way that only the first half of the file can be copied, no matter the file size. This means that if a file has 1.8 mb, I can copy 900 kb before the input/output error shows off. I have checked other files, and whenever I found a damaged one, if I attempted to copy it, only half of it would come out. It seems a strange pattern of damage, isn't it?
Two Samples Using DD
- SHOULDERS.MYD (input/output error)
root:/mnt/disk1/bugged-tables# dd if=/media/usb/usr/local/mysql-5.0.96-linux-x86_64-glibc23/data/DB/Shoulders.MYD of=/mnt/disk1/bugged-tables/Shoulders conv=noerror bs=128k
dd: error reading '/media/usb/usr/local/mysql-5.0.96-linux-x86_64-glibc23/data/DB/Shoulders.MYD': Input/output error
4+1 records in
4+1 records out
630784 bytes (631 kB, 616 KiB) copied, 8.58983 s, 73.4 kB/s
13+2 records in
13+2 records out
1836628 bytes (1.8 MB, 1.8 MiB) copied, 8.7681 s, 209 kB/s
original size: 1,861,204 bytes
dd file size : 1,836,628 bytes
- HELMET.MYD (input/output error)
root:/mnt/disk1/bugged-tables# dd if=/media/usb/usr/local/mysql-5.0.96-linux-x86_64-glibc23/data/DB/Helmet.MYD of=/mnt/disk1/bugged-tables/Helmet conv=noerror bs=128k
dd: error reading '/media/usb/usr/local/mysql-5.0.96-linux-x86_64-glibc23/data/DB/Helmet.MYD': Input/output error
9+1 records in
9+1 records out
1232896 bytes (1.2 MB, 1.2 MiB) copied, 8.54181 s, 144 kB/s
16+2 records in
16+2 records out
original size: 2,316,880 bytes
dd file size : 2,239,056 bytes
2239056 bytes (2.2 MB, 2.1 MiB) copied, 8.62538 s, 260 kB/s
SYSTEM INFORMATION
OS: Ubuntu 20.04 (no-gui)
MySQL: 5.0
HDD: 2x 2TB (no-RAID)
FSCK
root:~# fsck /dev/sda
fsck from util-linux 2.20.1
e2fsck 1.42 (29-Nov-2011)
fsck.ext2: Superblock invalid, trying backup blocks...
fsck.ext2: Bad magic number in super-block while trying to open /dev/sda
The superblock could not be read or does not describe a correct ext2
filesystem. If the device is valid and it really contains an ext2
filesystem (and not swap or ufs or something else), then the superblock
is corrupt, and you might try running e2fsck with an alternate superblock:
e2fsck -b 8193 <device>
DMESG (when attempting to copy data out of the HDD)
[36729.425176] ata1.00: exception Emask 0x0 SAct 0x600000 SErr 0x0 action 0x0
[36729.425196] ata1.00: irq_stat 0x40000008
[36729.425211] ata1.00: failed command: READ FPDMA QUEUED
[36729.425230] ata1.00: cmd 60/08:a8:60:84:91/00:00:32:00:00/40 tag 21 ncq 4096 in
[36729.425230] res 41/40:08:60:84:91/00:00:32:00:00/00 Emask 0x409 (media error) <F>
[36729.425258] ata1.00: status: { DRDY ERR }
[36729.425268] ata1.00: error: { UNC }
[36729.618379] ata1.00: configured for UDMA/133
[36729.618404] sd 0:0:0:0: [sda]
[36729.618409] Result: hostbyte=DID_OK driverbyte=DRIVER_SENSE
[36729.618414] sd 0:0:0:0: [sda]
[36729.618417] Sense Key : Medium Error [current] [descriptor]
[36729.618424] Descriptor sense data with sense descriptors (in hex):
[36729.618427] 72 03 11 04 00 00 00 0c 00 0a 80 00 00 00 00 00
[36729.618446] 32 91 84 60
[36729.618455] sd 0:0:0:0: [sda]
[36729.618461] Add. Sense: Unrecovered read error - auto reallocate failed
[36729.618466] sd 0:0:0:0: [sda] CDB:
[36729.618469] Read(10): 28 00 32 91 84 60 00 00 08 00
[36729.618485] end_request: I/O error, dev sda, sector 848397408
[36729.618546] ata1: EH complete
[36732.425409] ata1.00: exception Emask 0x0 SAct 0x1000000 SErr 0x0 action 0x0
[36732.425430] ata1.00: irq_stat 0x40000008
[36732.425445] ata1.00: failed command: READ FPDMA QUEUED
[36732.425463] ata1.00: cmd 60/08:c0:60:84:91/00:00:32:00:00/40 tag 24 ncq 4096 in
[36732.425463] res 41/40:08:60:84:91/00:00:32:00:00/00 Emask 0x409 (media error) <F>
[36732.425491] ata1.00: status: { DRDY ERR }
[36732.425502] ata1.00: error: { UNC }
[36732.604135] ata1.00: configured for UDMA/133
[36732.604161] sd 0:0:0:0: [sda]
[36732.604166] Result: hostbyte=DID_OK driverbyte=DRIVER_SENSE
[36732.604171] sd 0:0:0:0: [sda]
[36732.604174] Sense Key : Medium Error [current] [descriptor]
[36732.604180] Descriptor sense data with sense descriptors (in hex):
[36732.604184] 72 03 11 04 00 00 00 0c 00 0a 80 00 00 00 00 00
[36732.604203] 32 91 84 60
[36732.604211] sd 0:0:0:0: [sda]
[36732.604217] Add. Sense: Unrecovered read error - auto reallocate failed
[36732.604223] sd 0:0:0:0: [sda] CDB:
[36732.604226] Read(10): 28 00 32 91 84 60 00 00 08 00
[36732.604242] end_request: I/O error, dev sda, sector 848397408
[36732.604286] ata1: EH complete
