[Discuss] weird disk issues

[Stephen Adler]

On Thu, 2015-10-15 at 11:29 -0400, Dan Ritter wrote:
> On Thu, Oct 15, 2015 at 11:25:32AM -0400, Stephen Adler wrote:
> > 
> > Recently I have had a rash of disk failures on my home server. One
> > of
> > them was an SSD 120G drive, a Samsung EVO 840. Then a seagate
> > drive.
> > The this which is puzzling me is that I have all my drives tied
> > into
> > raid arrays of various sorts in order to protect myself from disk
> > failures. But the odd thing is that when I have my disk failures,
> > it
> > affect more than one drive in the array knocking out the whole raid
> > array.
> > 
> > what I've had to do is boot up using a live fedora image and then
> > issue
> > mdadm commands to restore the array. In one case, it was only one
> > disk
> > which was bad in the array, even though 2 disk's were kicked out. I
> > had
> > to reassemble the array to get it back up in degraded mode, after
> > which
> > replacing the one drive got it back up. In another case, two SSD
> > drives
> > went out, killing my raid 1 array (mirror mode). Again, mdadm'ing
> > getting it back, and only replacing one SSD drive got the system
> > going.
> > Then again last night, my SSD raid 1 array kicked both drives out,
> > but
> > I was able to reassemble it without replacing drives.
> > 
> > All this has occurred within one week! I've owned the system for
> > like 7
> > years and never a glitch, until now.
> > 
> > So, is it really the drives? could it be my motherboard or SATA
> > controller on the motherboard is going bad? Bad power supply?
> > Should I
> > consider replacing the motherboard? Or am I just having a bad raid
> > week?
> 
> 7 year old system? I would consider the power supply and the
> SATA controller prime for replacement, and also replace the SATA
> cables - they're cheap.
> 
> -dsr-
> 

The SATA controller's on the motherboard. I've been looking for a
replacement motherboard but it's going to cost me $200+some. uggg...

My problem with motherboards is that I've gotten into this, "I have to
have ECC memory", mode and I won't put together any system which does
not support ECC.

I'll do a cable replacement tonight tho...

Cheers. Steve.