1

I need some advice on how to proceed with an issue that I realized I had with data corruption on a 4TB RAID 1/0 array using the RocketRAID 3522 Controller.

I'm noticing that the SHA1 file hash does not match locally vs when copied to the RAID array. When the video file is played from the RAID array you see artifacts/pixelation which is a sign that it is corrupt which the SHA1 clearly shows it is.

Looking at the admin panel of the RAID controller it shows the status of the RAID array is Normal and the device status shows it is OK. The copy either with Windows 10 Pro copy and paste or via Robocopy always completes with no issues.

Can anyone offer some advice on how to troubleshoot ?

Update 1: After some more testing when the file is > 1.1 GB the corruption occurs

Any insights from anyone?

Exocomp
  • 163
  • 10

1 Answers1

1

Since the physical disks seem to be OK (RAID array is Normal) it’s most probably the issue with the controller itself.

1) I would recommend checking if there is an updated firmware for the controller that might fix those issues.

2) For further troubleshooting please consider disabling RAID caching (turn off Always Read-Ahead and switch to Write-Through). If there are any other caching mechanisms present in RAID array settings, consider switching those off too. These measures will not impact your data (sometimes can be even done on the fly) and might fix the issue. This is a temporary fix and if it helps you will probably need to replace the controller.

Net Runner
  • 6,169
  • 12
  • 34
  • 1
    The firmware for the controller is using the latest they offer. In doing some more testing, I reduced the memory load on the box to around 10% and at that level it is able to write a 3GB file correctly, I did the test several times to confirm. However, now when I try to copy the 6GB the issue starts up again. What do you think, does it sound like I have a bad memory stick on the box itself ? – Exocomp Nov 11 '16 at 17:13
  • 1
    I think I found the bad RAM stick, removed it and ran several 12GB file copy tests and seeing no issues. However, what's weird is that I ran memtest86 on the "bad" memory stick and it reported no errors (1 pass). I'll have to run some more tests on that "bad" stick to see if I can get it to error. – Exocomp Nov 12 '16 at 03:12