Windows data integrity issues

Forums › Microsoft Windows › Windows data integrity issues - Win 7 / 10

timmmay

20429 posts

Uber Geek

Trusted

Lifetime subscriber

#165922 24-Feb-2015 18:36

NOTE: please read from this post 9 March further down the thread, this post is now out of date.

I have the latest W10 build, fully up to date. It's running on an i7 2600K, 16GB RAM (which passes memtest x86 tests), good brand motherboard (I forget which one I got but I could check), two SSS (Samsung 840 pro, OWC 120 pro) and four hard disks. Two of the disks are 4TB HGST formatted ReFS in a mirror pair.

I've just run a Macrium Reflect image, putting the image onto my OWC SSD. The image has been successfully verified twice. I copied the image file (16GB) from the SSD to the ReFS SS volume, which completed without error. When I ran the Reflect verify on the file it failed. I copied the image again and it passed verification.

Looking into the event logs I can see events that are ReFS failures - around one a day, different files. It fixes the errors each time.

Text: The file system detected a checksum error and was able to correct it.
Source: ReFSv1
EventID: 132

Unfortunately it doesn't seem to say which disk the error was with so I don't know if it's just one faulty disk. It seems that Windows copy is the thing that's failing - you'd have thought that was an absolute basic that was licked years ago. There doesn't seem to be a good program to copy then verify - I tried one but it made little difference.

I'm looking for general advice or suggestions. Memtest x86 has been run for > 24 hours, no issues there.

NOTE: please read from this post 9 March further down the thread, this post is now out of date.

1 | 2

1185 posts

Uber Geek

Trusted

#1245814 24-Feb-2015 19:02

Just reading http://en.wikipedia.org/wiki/ReFS and from the look of things the problem may actually be with ReFS itself due to immaturity of the system.

Good advice might be to not use it, especially on a pre-release operating system...

AliExpress

Shop now on AliExpress (affiliate link).

timmmay

20429 posts

Uber Geek

Trusted

Lifetime subscriber

#1245847 24-Feb-2015 19:53

I'm starting to think you may be right. My main other option seems to be ZFS which is much more expensive. Software RAID of some type could be an option I guess.

freitasm

BDFL - Memuneh
79056 posts

Uber Geek

Administrator

ID Verified

Trusted

Geekzone

Lifetime subscriber

#1245879 24-Feb-2015 20:34

I am running Windows Server 2012 R2 with 500GB (boot) and two 4TB HDD ReFS mirror (OS, not Spaces). Not seeing errors on that.

Have you tried Teracopy instead?

timmmay

20429 posts

Uber Geek

Trusted

Lifetime subscriber

#1245900 24-Feb-2015 20:40

Thanks M, I'll give it a shot. Could be anything causing the problem - a single bad memory cell, a cable, motherboard, or bad luck.

timmmay

20429 posts

Uber Geek

Trusted

Lifetime subscriber

#1245928 24-Feb-2015 21:18

Ok so I copied two Reflect images using Teracopy from the ReFS SS and it said the CRC matched. They files are 30GB, bigger than RAM, so it's not reading from a disk cache. Macrium Reflect is saying the images are corrupt. The source images are still fine.

I copied a few more files using Teracopy images around. A couple copied fine, more than half Teracopy reported the CRC didn't match. Of the ones that said they copied fine sometimes Reflect reports they're ok, sometime it reports failures. So I can have Teracopy saying "copy failed" but Macrium saying "it copied fine". Argh!

Is there any kind of Windows program that can test disks without destroying the data? Say I want to have it fill 1TB of disk with a specific pattern then read it back, or something like that? As I said memtestx86 reports no problem and hardware load testers like prime95 also report no problems.

At a bit of a loss here.

timmmay

20429 posts

Uber Geek

Trusted

Lifetime subscriber

#1246632 25-Feb-2015 18:56

Ok I have new results. When copying from any disk to any disk with Trucopy I can get CRC failures reported occasionally - around 30-50% of the time. No single disk is worse than another as far as I can see. The ReFS SS doesn't seem to fail any more often than the others, if anything it seems slightly more reliable. When I copy images to other disks using Trucopy (being away from the PC I assume there were no CRC errors) then validate using Reflect it sometimes fails.

I'm thinking there's something fundamental wrong with this PC... it passes every theoretical test I give it, but fails practical tests.

timmmay

20429 posts

Uber Geek

Trusted

Lifetime subscriber

#1253658 9-Mar-2015 07:57

I've been doing more testing. I've installed Windows 7 on another disk, bought it fully up to date, and done some testing. Copying 100GB of data between two NTFS disks using Teracopy with verify turned on I get 3-15 CRC fail errors. This happens whether the drives are plugged into the motherboard or my add on SATA card. This happens the same on W7 and W10, so I can rule out the OS as a factor. It could be driver related, but Windows isn't bad at updating drivers itself these days.

I've also run a program that writes masses of data to the drives then reads it back to verify it. It reports no failures in W7 (four runs, all NTFS on a WD Black and an SSD) but in W10 I do get verification errors. I'll be re-running this test in both OS's today to verify these results. I may also run badblocks if I can work out a way to do it easily - maybe a live Linux distribution.

At this stage my only conclusion is it's a weird isolated minor hardware failure somewhere on the motherboard causing the data corruption. I'd love another explanation/solution because right now it looks like I'll have to replace the entire PC (it's an old socket 1155 which is difficult to buy new), which I can afford but I'd prefer not to do.

Athlonite

1828 posts

Uber Geek
Inactive user

#1253759 9-Mar-2015 10:04

sounds like a weird one you have there which could be hard to diagnose
is it the software
is it the mobo
is it the Hdd or SSD
is it the SATA cables

software sounds unlikely to be the problem as you've tried different copy methods
could be a failing SATA controller or dicky driver either update the driver if it still happens then try rolling back to a previous version and see if it still has problems
test the HDD's and SSD's in a different system if no problems then it's unlikely to be that
try new SATA cables (some cables that come with mobo's are just cheap n nasty) I like to use AC Ryan cables they cost a little more but well worth it better shielding and all round better quality cables

timmmay

20429 posts

Uber Geek

Trusted

Lifetime subscriber

#1253782 9-Mar-2015 10:22

I think I've ruled out software as W7 and 10 have the same problems. I think I've ruled out HDD/SSD as it affects every disk I have - 6 at the moment. I've tried swapping SATA cables, but not systematically, more randomly, but that's a possibility. I have a couple of new ones, I could try that. I suspect it's the mobo.

I should try the disks in another system, I do have access to one I could try easily enough at a friends place. I'd be very surprised if it's the disk, as it's affecting two different brands of SSD inc a Samsung 840 pro, a WD Black, a Seagate Barracuda, and a 2x4TB HGST ReFS mirror.

Unfortunately I think mobo is the most likely option :(

Athlonite

1828 posts

Uber Geek
Inactive user

#1253799 9-Mar-2015 10:25

try the drivers first then to rule it out or in same problem with older or newer drivers would certainly suggest controller problems

timmmay

20429 posts

Uber Geek

Trusted

Lifetime subscriber

#1253854 9-Mar-2015 11:10

So I guess I go to the motherboard manufacturer's site and download their latest drivers? The motherboard is 3-4 years old, so their drives are probably way out of date compared with the ones available through windows update?

Athlonite

1828 posts

Uber Geek
Inactive user

#1253887 9-Mar-2015 11:33

No I'd go to Intels website and grab them from there

timmmay

20429 posts

Uber Geek

Trusted

Lifetime subscriber

#1253915 9-Mar-2015 12:07

So I'd look for the latest Intel P67 based drivers? Or a generic Intel storage type driver?

This is my motherboard here - Gigabyte GA-P67A-UD3R-B3. If you could tell me what to download so I don't stuff it up that'd be much appreciated :)

Athlonite

1828 posts

Uber Geek
Inactive user

#1253979 9-Mar-2015 13:13

try any of these https://downloadcenter.intel.com/download/24293 or go with the one listed on your mobos website here http://www.gigabyte.co.nz/products/product-page.aspx?pid=3764#dl

richms

27982 posts

Uber Geek

Trusted

Lifetime subscriber

#1253990 9-Mar-2015 13:25

I would just replace motherboard and CPU/ram if it is suspected of corrupting data.

Drivers dont cause it. I have had machines running bog standard windows drivers on a core 2 quad vintage machine that has probably transfered exabytes between drives as I have moved things around. Thats internal sata, esata, USB2.0 and pcie card esata to a storage enclosure. All on a fresh install of 7 with just windows updates and I think the esata PCIe card driver from the silicon image website.

Think I have had a couple of occasions where I have had files corrupt and it was a read error off a HDD that was quickly picked up when I rechecked the torrent files.

For the number of errors you are having, something is wrong. If there is that much reading or writing improperly I would expect frequent bluescreens etc as well.

Bin the hardware, get an equivilant machine ex lease or replace it with current stuff and then problems should stop. Drivers are a red-herring. They may help with performance but there is no way that anyone would ship hardware that corrupts files on a stock installation.

Richard rich.ms

1 | 2