Recently changed hardware over to a Supermicro H8DME-2 motherboard with dual opteron cpus and two SAT2-MV8 sata cards.  Tested the hardware with three disks and a test unraid setup (5.05) for two weeks and no problems with reads, writes, cache, parity checks, etc.  Of course as soon as I moved to my old parity, cache and data drives problems started (SAT2-mv8 cards are also from my old setup which was PCI rather than PCI-X).  I recently submitted a post where parity and a data drive both went orange.  Recovered from that with help from member megalodon.  Finally, finished a parity check with 0 errors.  Started adding data to the array and a new 2Tb drive red-balled while the mover was adding data from the cache drive.  I've had multiple red balls on new drives in different slots in my drive cages. and occasionally a drive goes "missing" only to be recognized upon reboot.  This is getting crazy.

 

I've included a syslog for a case when a new drive 13 apparently developed errors.  It was so new and had just passed pre-clearing that I went through the trust disk procedure and started a data rebuild.  It failed at exactly the same spot with the same number of errors in the same place (this relates to attached syslog).  Maybe the drive is bad, but problems are so frequent I'm wondering if I have bad I/O hardware.

 

I'm hoping that the attached syslog sheds some light on this.  It's frustrating as my old setup was rock solid for years, but it was so old it had a 2TB drive limit.

syslog-20140905-152153.zip

I think something is wrong with that card. I have the same issue with the same card as well. I have a different mobo. I am going to order the USAS2-L8i version of that card and see how it goes.

The interesting thing is the same cards worked for a long time on a PCI interface (slow but solid).  Now on PCI-X they are faulty?  Could be I guess.

Cannot say. I have it in a PCI-E x8 slot.  Hopefully the l8i version of the card will work.

 

I think i found the issue.

 

"This is an HBA, not a raid controller. This is meant for software raid solutions such as ZFS." <- Quote by Kevin W. from NewEgg Reviews

So now comes in the issue on how to fix.

That quote is interesting, however, there are many posts in the unraid forum where this card has been working well even with 4TB drives.

Cannot say. I have it in a PCI-E x8 slot.  Hopefully the l8i version of the card will work.

 

I think i found the issue.

 

"This is an HBA, not a raid controller. This is meant for software raid solutions such as ZFS." <- Quote by Kevin W. from NewEgg Reviews

 

This is not the issue. You want a HBA. A RAID controller is more difficult to configure if it's compatible at all.

 

The SAT2-MV8 is a PCI-X card. It is backwards compatible with PCI. It should not fit in a PCIe slot.

Cannot say. I have it in a PCI-E x8 slot.  Hopefully the l8i version of the card will work.

 

I think i found the issue.

 

"This is an HBA, not a raid controller. This is meant for software raid solutions such as ZFS." <- Quote by Kevin W. from NewEgg Reviews

 

This is not the issue. You want a HBA. A RAID controller is more difficult to configure if it's compatible at all.

 

The SAT2-MV8 is a PCI-X card. It is backwards compatible with PCI. It should not fit in a PCIe slot.

 

I am aparently talking about a different card and it is a pcie-x8 card.

This is the one I have. http://www.supermicro.com/products/accessories/addon/AOC-SAS2LP-MV8.cfm

 

But yes the card that the OP is talking about is a PCI-X card.

http://www.supermicro.com/products/accessories/addon/AOC-SAT2-MV8.cfm

 

The question is is it something in the MV8 cards in general?

I thought I'd report back on this in case it helps others as I have solved the problem.

 

I tried replacing the power supply thinking that the H8DME-2 motherboard was underpowered - no change.

 

I replaced the CPUs (I had bought 2 AMD Opteron 2419's off ebay for $10) - no change.

 

Then I moved one of my two AOC-SAT2-M8 cards so that one was on the PCI-x 133/100 slot and the other on the PCI-x 100/66 slot (I originally had my two cards in the two 133/100 slots).  After moving the card Unraid turned back into the bullet-proof system that I have known for the past decade.  Everything works fine and all of my random sporatic red balls and failures vanished.

 

I have one issue remaining with WOL that I will address in a new post