Got 3 new drives. 2 clear fine but one says:

 

== Disk /dev/sdc has NOT been precleared successfully
== skip=294400 count=200 bs=1000448 returned 00004 instead of 00000

 

 

I've run preclear 3 or 4 times now. I'm doing an extended SMART check now. Not sure what to make of this.

Preclear does three things -

 

1. It reads the disk (read phase)

2. It writes binary zeros to the disk (write phase)

3. It reads the disk again, making sure it is full of zeroes (verify phase)

 

Your preclear verify phase failed. This is extremely rare. If fact, I think I wrote that error message and I haven't seen it since I tested it.

 

What to do is entirely a different matter. A disk is always supposed to accurately read and write. There are those that debate that at the extremes of the reliability stats, it is possible to have a bit flip. The difference between a zero (0000) adds a four (0100) is just one bit. Maybe the magic number hit (like winning the lottery over and over). I doubt it. Disks are not perfect, but they employ complicated redundancy schemes for detecting and correcting errors, and like a steel trap, are never supposed to let an error get out.

 

A memory error would be my first suspect. Although memory is more reliable than a disk surface, there is no redundancy built in (unless you use ECC memory). I would suggest running a 24 hour memory test to see if you have a failing memory module.

 

There are other possibilities. A cable run too close to another cable, or power cable, can create magnetic fields and disrupt signal flow. A disk controller can have a memory error.  The cache in the drive. The firmware in the drive. But with these kinds of issues, you might expect more frequent errors.

 

I'd try the memory test to start with, and try to make sure your cables are run perpendicular, and not twisted together causing magnetic properties to be encouraged.

 

This could be a challenging one to narrow down. But important to make sure it is not happening once you trust your data on the server.

 

There has been more than one case of memory causing this, including me.

I precleared 3 disks at the same time, the single disk failed. Then I ran it again on just that disk and it failed again. Mem test on the unraid machine or the one that's running the terminal command?

 

I can try running it again from a different slot...

30 minutes ago, Pandemic said:

Mem test on the unraid machine

On the boot menu

Could be something about that drive Did error happen in exact same place?

 

If so, I would be highly suspect of the drive.

 

I once had a disk act up in one of my drive slots. I forget the exact symptoms, but attempts to fix replace cables, clean contacts on the backplane, etc. were unsuccessful. To determine if it was the slot or the drive, I swapped it with a neighboring slot. The new drive had no issues, there, and the problematic drive had no problems in its new home. Was not the most satisfying solution - but a solution nonetheless and neither complained for their lifespans.

 

Good luck!

I'm running preclear again. If it fails, I will move it to a slot on a different controller to see what happens and report back.

Moved the HDD to a different slot and now preclear works fine. I'll have to take a look into wtf is going on with the original bay.

i'm going to put another new drive in that bay and clear it again to see what happens.

Put another new drive in the slot that was having issues. Everything's fine now. No idea what the deal was but now all 3 drives have been precleared twice.

23 hours ago, Pandemic said:

Put another new drive in the slot that was having issues. Everything's fine now. No idea what the deal was but now all 3 drives have been precleared twice.

 

Similar to my experience I documented above. Sometimes you never get to a satisfying root cause. But continue to monitor for signs of trouble.

Yeah. I've run preclear like 8x on these drives. Probably good to go but... odd.