Intermittent System File Corruption - Please HELP!

[Baasha]

For the past month or so, my X58 system has been freezing intermittently. This first happened in Win 7 x64 SP1 and then started happening on Win 8 Pro x64.

Win 7 is installed on a 2x SSD RAID-0 array and Win 8 is installed on a "regular" 7200RPM HDD.

At first, the computer was freezing even in Safe Mode! Then, I logged into Win 8 and then ran SFC /SCANNOW on the C-drive (Win 7) and it was able to complete without freezing.

It says it found corrupt files but is unable to fix them. I ran it several times and it keeps giving the same result.

It seemed to be fixed and then a week later, same thing. This has happened four times now - freezing, run SFC /SCANNOW several times, works okay for a few days, then freezing again.

I've run MalwareBytes, virus scans, ComboFix etc. to no avail - this keeps happening every few days.

I don't know what's causing the system files to keep getting corrupted! Could the SSDs be failing? If that's the case, why does it happen on Win 8 (regular HDD) as well?

I got these SLC SSDs in May 2009 - are they in their last throes?

I have an "SFCDETAILS.TXT" file but it's almost 60MB in size! I can't attach it here but some of the lines in there are like this:

Microsoft-Windows-Migration-ReplacementManifests-ds, Version = 6.2.9200.16384, pA = PROCESSOR_ARCHITECTURE_AMD64 (9), Culture neutral, VersionScope = 1 nonSxS, PublicKeyToken = {l:8 b:31bf3856ad364e35}, Type neutral, TypeName neutral, PublicKey neutral in the store, file is missing

2013-07-18 10:51:23, Info CSI 00011b77 [SR] Cannot repair member file [l:20{10}]"wclWdi.dll" of Microsoft-Windows-Wcl, Version = 6.2.9200.16384, pA = PROCESSOR_ARCHITECTURE_AMD64 (9), Culture neutral, VersionScope = 1 nonSxS, PublicKeyToken = {l:8 b:31bf3856ad364e35}, Type neutral, TypeName neutral, PublicKey neutral in the store, file is missing
2013-07-18 10:51:23, Info CSI 00011b78 [SR] Cannot repair member file [l:56{28}]"register-cimprovider.exe.mui" of Microsoft-Windows-WMIV2-Register-CimProvider-exe.Resources, Version = 6.2.9200.16384, pA = PROCESSOR_ARCHITECTURE_AMD64 (9), Culture = [l:10{5}]"en-US", VersionScope = 1 nonSxS, PublicKeyToken = {l:8 b:31bf3856ad364e35}, Type neutral, TypeName neutral, PublicKey neutral in the store, file is missing

App-Host.Resources, Version = 6.2.9200.16384, pA = PROCESSOR_ARCHITECTURE_AMD64 (9), Culture = [l:10{5}]"sv-se", VersionScope = 1 nonSxS, PublicKeyToken = {l:8 b:31bf3856ad364e35}, Type neutral, TypeName neutral, PublicKey neutral in the store, file is missing

 

[denis280]

If SFC could not fix something, then run the command again to see if it may be able to the next time. Sometimes it may take running the sfc /scannow command 3 or more times to completely fix everything that it's able to.

 

[Cerb]

Can you post SMART data for all the drives (screenies of CrystalDiskInfo, HDTach, from in Intel RST, or another such utility would be the easiest to read)?

 

[Bubbaleone]

If SMART doesn't report anything physically wrong with your disk/s (as Cerb noted), boot the Windows 7 installation disk and when you get to the Install Windows screen, select the repair option in order to use the advanced recovery tools. On the recovery tools screen open the Command prompt and run these consecutive commands, pressing Enter after each command:

diskpart (this loads the diskpart utility)

list disk (view a numbered list of all the disks)

select disk <#> (replace <#> with the actual number of the operating system disk you wish to work with)

detail disk (view the volume drive letter/s on the selected disk/s and write them all down. Do this for the drives that contain Windows 7 and 8)

exit (exits the diskpart utility)

The reason for first using diskpart to determine the drive letter/s for your operating system drives is because the volume drive letter assignments in the Windows Recovery Environment will differ from the volume drive letters you would see in Disk Management or Explorer.

Now that you know which drive letters correspond to Windows 7 and 8, you'll know which drive letters to use when you run chkdsk and sfc from this command prompt. Now run this chkdsk command:

chkdsk X: /x /v /r /b (replace X: with the actual drive letter, and run this command on every volume drive letter that diskpart listed)

When you've finished running chkdsk on every volume (which will take a while depending on the size of the drive/s), run the sfc /scannow command on the Windows 7 and 8 system volumes.

.

 

[Steltek]

Windows file system corruption can also be an indirect result of a bad memory module. It would probably be a good idea to run a few passes of MEMTEST86+ on each of your memory modules, one at a time, to ensure you don't have a module going south on you.

 

[Baasha]

Can you post SMART data for all the drives (screenies of CrystalDiskInfo, HDTach, from in Intel RST, or another such utility would be the easiest to read)?

I just realized I don't have Intel RST installed in Win 7!?! I tried to download it but the link kept saying, "Page Not Found." I will try again today. I don't have the SMART info for the drives (yet).

 

[Baasha]

If SMART doesn't report anything physically wrong with your disk/s (as Cerb noted), boot the Windows 7 installation disk and when you get to the Install Windows screen, select the repair option in order to use the advanced recovery tools. On the recovery tools screen open the Command prompt and run these consecutive commands, pressing Enter after each command:

diskpart (this loads the diskpart utility)

list disk (view a numbered list of all the disks)

select disk <#> (replace <#> with the actual number of the operating system disk you wish to work with)

detail disk (view the volume drive letter/s on the selected disk/s and write them all down. Do this for the drives that contain Windows 7 and 8)

exit (exits the diskpart utility)

The reason for first using diskpart to determine the drive letter/s for your operating system drives is because the volume drive letter assignments in the Windows Recovery Environment will differ from the volume drive letters you would see in Disk Management or Explorer.

Now that you know which drive letters correspond to Windows 7 and 8, you'll know which drive letters to use when you run chkdsk and sfc from this command prompt. Now run this chkdsk command:

chkdsk X: /x /v /r /b (replace X: with the actual drive letter, and run this command on every volume drive letter that diskpart listed)

When you've finished running chkdsk on every volume (which will take a while depending on the size of the drive/s), run the sfc /scannow command on the Windows 7 and 8 system volumes.

.

Thanks for the detailed response.

Yesterday, I ran CHKDSK from elevated command prompt at boot for BOTH drives (Win 7 --> C: and Win 8 --> F🙂 and they completed fine. Both of them said some bad sectors were replaced/fixed etc. I then ran SFC /SCANNOW on the Win 7 drive (SSD array) and it said there were no integrity violations. It didn't let me run the scan for the Win 8 drive until a reboot. I rebooted into Win 7 and then ran SFC /SCANNOW on the F: drive (Win 8). However, it froze at 89%! 😡

Then the computer itself froze forcing me to hard-reset! This is extremely aggravating.

I'm going to put everything to stock settings (remove the CPU OC) and test it for a few days. I'm quite sure that it's not the OC since it has been running perfectly at these settings from day 1 (4.4GHz @ 1.30V).

*EDIT: I just ran SFC /SCANNOW on both drives (Win 7 & Win 8) in Win 7 (elevated Command Prompt) and they completed fine - No Integrity Violations were found!

I will keep using the computer and see if the freezing occurs again. If it does, what should I do?

 

[Cerb]

What about SMART? SMART is very good about positive prediction of failure. If any important parameters are out of line, it's a sure sign something is wrong, and your BIOS SMART check may not look at the right ones. Do this whether it does anything else in the short-term or not.

If you run the chipset at stock speeds, you should never see bad sectors from Chkdsk, FI. A single bad sector is some kind of bad news, unless it only occurs on a chkdsk forced after a nasty shutdown (there's always an exception, right? 🙂). Now, running your CPU at stock is a good idea, as the fundamental problem could be CPU, RAM, or PSU, not the drives, with a bad OC, but do check the drives themselves.

It wouldn't surprise me at all, FI, if your SSDs are at their last. Wear-leveling has gotten better over the years, to the point that older SLC might not last as long as new MLC, due to drastic reductions in average write amplification.

 

[Baasha]

What about SMART? SMART is very good about positive prediction of failure. If any important parameters are out of line, it's a sure sign something is wrong, and your BIOS SMART check may not look at the right ones. Do this whether it does anything else in the short-term or not.

If you run the chipset at stock speeds, you should never see bad sectors from Chkdsk, FI. A single bad sector is some kind of bad news, unless it only occurs on a chkdsk forced after a nasty shutdown (there's always an exception, right? 🙂). Now, running your CPU at stock is a good idea, as the fundamental problem could be CPU, RAM, or PSU, not the drives, with a bad OC, but do check the drives themselves.

It wouldn't surprise me at all, FI, if your SSDs are at their last. Wear-leveling has gotten better over the years, to the point that older SLC might not last as long as new MLC, due to drastic reductions in average write amplification.

I installed CrystalDiskInfo and here are the screenshots:

It looks the drives are fine(?).

What should I do next? I haven't reset the OC yet.

 

[Cerb]

The WD looks like it has a bad cable, or a chipset overclock. The UDMA CRC Error Count should be 0 or near it. Otherwise, it looks fine.That's errors occurring somewhere between the DMA request to system RAM and the drive's controller, and is usually a cable, but occasionally RAM or PSU.

The SSDs, I'm not sure about. You've got 48 uncorrectable errors on one, and 62 on the other. Unless that's somehow the result of a bad OC (shouldn't be, since that's inside the drive, and you should have been leaving the chipset and PCI-e speeds alone, so no way to OC SATA), that's kind of worrisome, given relatively low wear (for what it should be capable of, anyway). I wonder if it might be, "OK," in some sense, but not be retaining data long enough? I can't find any docs for it one way or the other (FI, if it was rated to 3 months, it might be considered normal). Just noticed, though, both have used some of their reserved blocks. I can't say for certain, given that those are moderately vendor-specific in how they're reported, but that doesn't look good, either.

 

[Baasha]

The WD looks like it has a bad cable, or a chipset overclock. The UDMA CRC Error Count should be 0 or near it. Otherwise, it looks fine.That's errors occurring somewhere between the DMA request to system RAM and the drive's controller, and is usually a cable, but occasionally RAM or PSU.

The SSDs, I'm not sure about. You've got 48 uncorrectable errors on one, and 62 on the other. Unless that's somehow the result of a bad OC (shouldn't be, since that's inside the drive, and you should have been leaving the chipset and PCI-e speeds alone, so no way to OC SATA), that's kind of worrisome, given relatively low wear (for what it should be capable of, anyway). I wonder if it might be, "OK," in some sense, but not be retaining data long enough? I can't find any docs for it one way or the other (FI, if it was rated to 3 months, it might be considered normal). Just noticed, though, both have used some of their reserved blocks. I can't say for certain, given that those are moderately vendor-specific in how they're reported, but that doesn't look good, either.

Okay, I will replace the SATA cable fro the WD drive and see if that helps. I didn't really understand what you said about the SSDs though. So do the unrecoverable errors mean that they are done? Is there a way for me to secure erase the SSDs and set them to 'factory default' to "fix" these issues or is that not possible?

I am going to reinstall the OS on another HDD I have lying around but if these SSDs can be used and used safely, I would prefer to use them.

Thanks for the help btw.

 

[Matt1970]

Every time I have seen intermittent corrupted files it's a sign the hard drive is going to fail.

 

[Cerb]

I didn't really understand what you said about the SSDs though. So do the unrecoverable errors mean that they are done?

Done, no, but possibly causing corruption, maybe.

Is there a way for me to secure erase the SSDs and set them to 'factory default' to "fix" these issues or is that not possible?

No. Just reading every block, such as with a chkdsk /r, will do whatever can be done at the moment (chkdsk isn't actually doing anything but reading; the SSD being forced to read the pages and handle any undo errors is where the work is done).

I am going to reinstall the OS on another HDD I have lying around but if these SSDs can be used and used safely, I would prefer to use them.

I can't say for certain, but my first instinct is that retention may be getting worse, and they may not have been rated like consumer flash to begin with. It's used reserved blocks, for instance, but to what end? Samsung's website pretends like everything earlier than the 830 never happened, so I don't know how much that's telling (if erase or program fail count are 0, as are re-allocation, what needs to force it to use the reserved blocks?).

OTOH, a new ~120GB MLC SSD will be better in every way. We're not in 2009, anymore, when your options were JMicron, Indilnix, and 1st-gen SFs.

 

[Baasha]

Done, no, but possibly causing corruption, maybe.

No. Just reading every block, such as with a chkdsk /r, will do whatever can be done at the moment (chkdsk isn't actually doing anything but reading; the SSD being forced to read the pages and handle any undo errors is where the work is done).

I can't say for certain, but my first instinct is that retention may be getting worse, and they may not have been rated like consumer flash to begin with. It's used reserved blocks, for instance, but to what end? Samsung's website pretends like everything earlier than the 830 never happened, so I don't know how much that's telling (if erase or program fail count are 0, as are re-allocation, what needs to force it to use the reserved blocks?).

OTOH, a new ~120GB MLC SSD will be better in every way. We're not in 2009, anymore, when your options were JMicron, Indilnix, and 1st-gen SFs.

Okay so I disconnected the SSDs and used a different cable for the regular HDD (Win 8) and tried to boot but I got a "Error Loading OS" message. I tried to switch the Storage Configuration from RAID to ACHI but no dice! :thumbsdown:

Any idea why I can't boot to Windows 8? I installed Win 8 *from* Win 7 on that drive after creating a partition on it (one half is "Storage" and the other half is a separate drive "H:" which is Win 8 Pro). Could that be the reason? I was booting dual OS's when the SSDs were working - It would bring up a blue screen that allowed me to choose either OS. Now, it just gives me that error message.

Please help!

 

[Ketchup]

Okay so I disconnected the SSDs and used a different cable for the regular HDD (Win 8) and tried to boot but I got a "Error Loading OS" message. I tried to switch the Storage Configuration from RAID to ACHI but no dice! :thumbsdown:

Any idea why I can't boot to Windows 8? I installed Win 8 *from* Win 7 on that drive after creating a partition on it (one half is "Storage" and the other half is a separate drive "H:" which is Win 8 Pro). Could that be the reason? I was booting dual OS's when the SSDs were working - It would bring up a blue screen that allowed me to choose either OS. Now, it just gives me that error message.

Please help!

The boot menu (or MBR) was probably on the SSD. IIRC, in windows 7 booting off the OS DVD and selecting Repair My Computer could fix this problem. I would think/hope that Windows 8 does the same.