-5

I am running a 4 drive RAID 5 via Intel Z97 (EDIT: not Z87) mobo under Windows 2012R2 with the latest RST driver.

The drives are identical model ST1000DM003-1ER162 1TB Seagate.

I seem to be fighting with a cabling issue. Whenever I try to close my computer case the cabling (SATA and power) gets pushed or moved. I've tried buying a new case but still have the issue. I'm not sure why the cabling is so sensitive.

This results in silent data loss and corruption. Lots of files get truncated, the system will freeze/crash after a few days, and chkdsk reports errors.

My worry is that after I do my best to reseat/fix the cabling I am still vulnerable to silent data corruption. Unfortunately I am stuck on Windows and can't run something like ZFS/BTRFS that has better file integrity features.

My questions are:

  1. Is this the expected failure mode? I would have expected a more proactive notification that the driver is failing to write data to the disks.

  2. Is there a different hardware/software setup that can ensure I at least know when my data is at risk so I can take immediate action and minimize the damage?

  3. Would a SAS drive setup tell me when something was going wrong?

EDIT: for the trolls who think this is solved by a "server" you may want to educate yourself.

Recent large academic studies have identified the surprising frequency of silent read failures that are not identified or resolved in enterprise hard disk arrays despite the typical integrity functions. Such errors result in corrupt data being provided by the disk array to the application without any notification or warning, potentially leading to erroneous operations and results.

https://www.necam.com/docs/?id=54157ff5-5de8-4966-a99d-341cf2cb27d3

CoderBrien
  • 179
  • 1
  • 4
  • 1
    R5 with 1TB disks - nope, Z87 mobo-based 'server' - nope, case not up for that task - nope. Come back to us when you have an actual server not a gaming PC with a grown-up OS. – Chopper3 Mar 06 '17 at 15:42
  • Aye. Better on another website. Though the OP should clarify why (s)he thinks it is a cable issue. (and not e.g. heat building up with the case closed). – Hennes Mar 06 '17 at 15:47
  • @hennes i know it's cabling because I monitor temp closely and the drives are running cool. it's take me 6-12 months to figure this out because it happens so infrequently but if i really push on the cabling i can always trigger it. – CoderBrien Mar 06 '17 at 15:55
  • oh dear, that's no way to be talking to people. Flaged as rude/offensive. – user9517 Mar 06 '17 at 16:02
  • 1
    "really push the cabling" sound like me 'when I really push a lot of data over the cable' (in which case you also use a lot of other things more intensively). And that is assuming you are not maxing out the power, though 4 active drives should not use all that much more. Probably in the 15-ish W era. Other than that, for data on FS security see ZFS. Still, add that to the post, and check out SuperUser and similar sites because SF's FAQ means this is probably off-topic on this specific site. (You can flag it with a request to move). – Hennes Mar 06 '17 at 16:05
  • @hennes as i said in the post i'm stuck with NTFS and can't use ZFS. but "thanks". – CoderBrien Mar 06 '17 at 16:12
  • You've proven my point, if this were a proper server instead of something you've hacked together yourself this wouldn't have taken you 6-12 months to figure out. Read our help pages - this site is for professional sysadmins working in a professional environment in a professional manner. – Chopper3 Mar 06 '17 at 16:12
  • How about ReFS? – Chopper3 Mar 06 '17 at 16:13
  • No you seriously should read our help pages and abide by what this site is and isn't for. – Chopper3 Mar 06 '17 at 16:13
  • `Recent large academic studies have identified the surprising frequency of silent read failures that are not identified or resolved in enterprise hard disk arrays despite the typical integrity functions. Such errors result in corrupt data being provided by the disk array to the application without any notification or warning, potentially leading to erroneous operations and results.` - The key point in that blurb being **enterprise hard disk arrays** - which this isn't. – joeqwerty Mar 06 '17 at 16:15
  • @joeqwerty the key point is that my issue IS NOT SOLVED by enterprise drives unless one knows what to look for. – CoderBrien Mar 06 '17 at 16:16

0 Answers0