CAM status: Uncorrectable parity/CRC error

[Robert Smith]

Hi,

Is anybody running into these “CAM status: Uncorrectable parity/CRC error” with on-board SATA ports on the AsRock E3C226D2I motherboard?

I have an Intel Core i3-4350 system with 16 GB ECC memory built on the E-ITX CFI A7879 Mini-ITX chasees, with four 4TB NAS-grade hard drives.

• At first I started getting the error on ada0. I replaced the hard drive on ada0 with a new one, the errors stayed.
• I replaced the SATA cable on ada0 with a new one. The error moved to ada2.
• I replaced the rest of the SATA cables with new ones. Still get the error on the ada2.

The error is not there all the time. The system could run fine for a few months, and then it pops out of the blue. No correlation to UPS events.

Smartctl ether shows nothing of note or it shows increase in UDMA_CRC_Error_Count. Scheduled long/short tests and scrubs are running.

Contemplating what to do next; what would you recommend?

Relevant thread: https://forums.freenas.org/index.php?threads/uncorrectable-parity-crc-error.14404/

 

[Ericloewe]

Does that thing have a backplane? Sounds like the likely culprit.

Does the error follow a drive or a SATA port?

 

[Robert Smith]

Yes, it has a backplane: certainly a suspect…

The error jumped everything: drive, port, cable, backplane slot, this is why I am also suspecting the motherboard.

 

[Ericloewe]

In that case, I'd start by bypassing the backplane and just connecting the drives (temporarily) directly with SATA cables.

 

[Robert Smith]

I am no longer getting UDMA_CRC errors after removing the backplanes and connecting the drives directly.

Not happy about wasting so much time swapping cables and drives around. I thought I was being prudent by purchasing special chassis with SATA backplanes; turned out I bought a headache.

In the event anyone else runs into the same issue, backplane part number is CFI-B23XA (the case has two). Not even sure which revision of SATA these backplanes conform to. These are just dumb backplanes with some resistors, capacitors and diodes. They have no microchips.

The motherboard has six SATA3 6.0 Gb/s ports.

 

[Robert Smith]

In that case, I'd start by bypassing the backplane and just connecting the drives (temporarily) directly with SATA cables.

Looks like your hunch about the backplane was correct. Here we are, three weeks after removing the backplane, and not a single new UDMA_CRC error.

Thank you.