[SOLVED] Is there hope? Trying to rescue some files from damaged NTFS harddrive that doesn't mount.

AndreasZ · 09-01-2019, 05:47 AM

Hi everyone!

When trying to mount the HDD, dmesg gives me repeating messages like:

Code:

[165000.248103] INFO: task kworker/u8:2:8424 blocked for more than 120 seconds.
[165000.248110]       Not tainted 4.7.0-1-686-pae #1
[165000.248112] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
[165000.248115] kworker/u8:2    D 00000020     0  8424      2 0x00000000
[165000.248127] Workqueue: writeback wb_workfn (flush-8:32)
[165000.248132]  f6166180 f79a9ec0 00000001 00000020 f6c46ac0 f79bd140 069f511d 000095f2
[165000.248139]  00000000 00000082 cab1e000 00000000 7fffffff cab1da70 d058147d 00000000
[165000.248146]  cab1dad4 d05840a5 d02b49c1 00000000 cab1daa4 d02b4a86 00000000 00000001
[165000.248152] Call Trace:
[165000.248162]  [<d058147d>] ? schedule+0x2d/0x80
[165000.248166]  [<d05840a5>] ? schedule_timeout+0x185/0x210
[165000.248171]  [<d02b49c1>] ? blk_run_queue_async+0x41/0x50
[165000.248175]  [<d02b4a86>] ? queue_unplugged+0x96/0xa0
[165000.248180]  [<d00e0feb>] ? ktime_get+0x4b/0x110
[165000.248184]  [<d0580c4e>] ? io_schedule_timeout+0x8e/0xf0
[165000.248187]  [<d02b7362>] ? get_request+0x3a2/0x700
[165000.248193]  [<d00ac430>] ? wake_atomic_t_function+0x70/0x70
[165000.248196]  [<d02b986b>] ? blk_queue_bio+0xfb/0x3a0
[165000.248199]  [<d02b7d80>] ? generic_make_request+0xd0/0x1b0
[165000.248202]  [<d02b7ec7>] ? submit_bio+0x67/0x170
[165000.248206]  [<d02afdd5>] ? bio_alloc_bioset+0x1b5/0x250
[165000.248210]  [<d0206ed5>] ? submit_bh_wbc+0x115/0x170
[165000.248214]  [<d0208f13>] ? __block_write_full_page.constprop.38+0x1e3/0x3c0
[165000.248218]  [<d0209360>] ? I_BDEV+0x10/0x10
[165000.248222]  [<d0209183>] ? block_write_full_page+0x93/0x100
[165000.248225]  [<d0209360>] ? I_BDEV+0x10/0x10
[165000.248229]  [<d0170890>] ? __writepage+0x10/0x40
[165000.248232]  [<d0170880>] ? wb_update_dirty_ratelimit+0x200/0x200
[165000.248235]  [<d0171555>] ? write_cache_pages+0x185/0x470
[165000.248238]  [<d0170880>] ? wb_update_dirty_ratelimit+0x200/0x200
[165000.248242]  [<d017187e>] ? generic_writepages+0x3e/0x70
[165000.248246]  [<d009c21c>] ? check_preempt_wakeup+0xec/0x220
[165000.248249]  [<d0173b5a>] ? do_writepages+0x1a/0x30
[165000.248253]  [<d01fdf1a>] ? __writeback_single_inode+0x3a/0x2f0
[165000.248257]  [<d01fe62e>] ? writeback_sb_inodes+0x1be/0x3b0
[165000.248262]  [<d01fe89c>] ? __writeback_inodes_wb+0x7c/0xb0
[165000.248265]  [<d01feada>] ? wb_writeback+0x20a/0x2c0
[165000.248269]  [<d01ff1dc>] ? wb_workfn+0x1fc/0x400
[165000.248273]  [<d00a65e9>] ? pick_next_task_fair+0x459/0x520
[165000.248278]  [<d0083c4f>] ? process_one_work+0x11f/0x380
[165000.248282]  [<d0083ee9>] ? worker_thread+0x39/0x460
[165000.248286]  [<d0083eb0>] ? process_one_work+0x380/0x380
[165000.248289]  [<d0089226>] ? kthread+0xa6/0xc0
[165000.248294]  [<d0584ee2>] ? ret_from_kernel_thread+0xe/0x24
[165000.248297]  [<d0089180>] ? kthread_create_on_node+0x130/0x130

and this:

Code:

[166630.289845] sd 7:0:0:0: [sdc] tag#0 FAILED Result: hostbyte=DID_ABORT driverbyte=DRIVER_OK
[166630.289854] sd 7:0:0:0: [sdc] tag#0 CDB: Write(10) 2a 00 00 5b e7 00 00 00 f0 00
[166630.289858] blk_update_request: I/O error, dev sdc, sector 6022912
[166630.289861] buffer_io_error: 230 callbacks suppressed
[166630.289864] Buffer I/O error on dev sdc1, logical block 6020864, lost async page write
[166630.289869] Buffer I/O error on dev sdc1, logical block 6020865, lost async page write
[166630.289872] Buffer I/O error on dev sdc1, logical block 6020866, lost async page write
[166630.289874] Buffer I/O error on dev sdc1, logical block 6020867, lost async page write
[166630.289877] Buffer I/O error on dev sdc1, logical block 6020868, lost async page write
[166630.289880] Buffer I/O error on dev sdc1, logical block 6020869, lost async page write
[166630.289883] Buffer I/O error on dev sdc1, logical block 6020870, lost async page write
[166630.289885] Buffer I/O error on dev sdc1, logical block 6020871, lost async page write
[166630.289894] Buffer I/O error on dev sdc1, logical block 6020872, lost async page write
[166630.289897] Buffer I/O error on dev sdc1, logical block 6020873, lost async page write
[166661.204065] usb 3-1: reset high-speed USB device number 4 using ehci-pci

"smartctl -a /dev/sdc" and
"fdisk -l /dev/sdc" both non-responding.

I've started mirroring the device using "ddrescue", which will take days. Is there any chance the copy will mount or a least give access to some files? (Sending the drive to a data rescue service isn't an option)

Thanks in advance
Andreas

wpeckham · 09-01-2019, 07:33 AM

Make certain that you have ONE OR MORE good backups first. I would try ntfsfix (it is linked to fsck.ntfs on some systems). Then, if there are still issues, you might try something like photorec to recover files.

If you have a Windows machine where it can be checked using the Microsoft tools, that is also a valid option. I assume you do else why have an NTFS drive.

ondoho · 09-01-2019, 07:34 AM

NTFS?
Stop right there, and use Windows tools to rescue your data.
Because NTFS is Windows' baby, and Linux support will always be sub-optimal.

jsbjsb001 · 09-01-2019, 07:47 AM

From your output above, it looks like at best the drive has bad sectors, at worst, it's failing altogether.

So trying to "fix" the filesystem might actually be making things worse, and lesson the chances of recovering the data on it.

Your first step should be to verify whether the drive itself is healthy or not, before trying to "fix" the filesystem on it. In any case, and as said above, use Windows itself to fix NTFS, not Linux.

Try checking it's SMART status before doing anything else;

Code:

smartctl -a /dev/sdX

Replace "sdX" with the correct device node for that drive. If you get a "command not found" or similar error message from the shell when trying to run the above command, install the smartmontools package. Also, you'll need to run the above command either as the root user, or with "sudo" in front of the above command, depending on your distribution.

Firerat · 09-01-2019, 08:05 AM

Quote:

Originally Posted by ondoho

NTFS?
Stop right there, and use Windows tools to rescue your data.
Because NTFS is Windows' baby, and Linux support will always be sub-optimal.

meh,

I've recovered data from many dead ntfs drives using linux
( but this one looks very pooched )

you could try to ddrescue partitions instead of the whole drive
But sometimes drives are just dead

it could be the PCB , which you could replace with a matching PCB, but this is complicated by the fact that you may have to replace fw chip on the new pcb with one from the old.

If the data is valuable consider a professional service.

if the expense can not be justified then wait for ddrescue, then clone the image and work with the clone

no magic can fix the HW

AndreasZ · 09-01-2019, 08:06 AM

Thanks so far!
* I've tried "smartctl -a /dev/sdX", but this leads to a hanging, unkillable process, same as "fdisk -l".
* My main question is: Is it worth the time to run ddrescue for 2 weeks, or will the resulting filesystem most likely be broken anyway?
* Unfortunately I had stopped the ddrescue-process today to make more attempts to mount the drive, but this didn't work and now ddrescue only reads errors. Could it be that "mount" damaged the drive more than 3 days of "ddrescue"?
* I'm surprise to hear that windows-tools are so much better at dealing with this. Thanks for the advice

Firerat · 09-01-2019, 08:15 AM

Quote:

Originally Posted by AndreasZ

Thanks so far!
* I've tried "smartctl -a /dev/sdX", but this leads to a hanging, unkillable process, same as "fdisk -l".
* My main question is: Is it worth the time to run ddrescue for 2 weeks, or will the resulting filesystem most likely be broken anyway?
* Unfortunately I had stopped the ddrescue-process today to make more attempts to mount the drive, but this didn't work and now ddrescue only reads errors. Could it be that "mount" damaged the drive more than 3 days of "ddrescue"?
* I'm surprise to hear that windows-tools are so much better at dealing with this. Thanks for the advice

ddrescue skips past the good that it has already recovered
so you may be just seeing the bad, if that is lots my guess is that drive is very broken
not much you can do, sorry

AndreasZ · 09-01-2019, 08:18 AM

Quote:

you could try to ddrescue partitions instead of the whole drive

There's only one partition. It'd be great to rescue a few files - but I suppose this isn't happening as the drive can't even be mounted, right?

Quote:

it could be the PCB , which you could replace with a matching PCB, but this is complicated by the fact that you may have to replace fw chip on the new pcb with one from the old.

The drive has been running smoothly during the ddrescue-attempt, wouldn't a hardware problem seem unlikely then?

Quote:

if the expense can not be justified then wait for ddrescue, then clone the image and work with the clone

yeah.. too late I guess

jsbjsb001 · 09-01-2019, 08:28 AM

If smartctl is hanging, then that's a bad sign. IMO it sounds like the drive is on it's way out, and is indeed failing altogether. smartctl gets the drive's SMART status from the drive itself, as SMART is built into the drive itself.

AndreasZ · 09-01-2019, 09:02 AM

Sure, I would have backed it up if it was my harddrive.

Marking it as "SOLVED" since the question "Is there hope?" has been answered...