We have a FreeBSD storage server, listing the specs:

Server1:

  • 3Ware 9500S-12 RAID Card
  • 12 x WD20EARS (2TB WD Caviar Green) SATA Disks connected to it via JBOD export (Not as single units due to the TLER problem with these disks
  • 2 x raidz2 zpools, 1 using 5 disks, the other using 7.
  • 3Ghz i7
  • 16 GB RAM
  • FreeBSD RELEASE-8.2

Currently we are having performance and reliability problems with the RAID card, and since we have been exporting the disks as JBODs we thought it would be better to just buy a bunch of SATA controllers to plug all the disks into SATA connectors directly rather than aggregate via the 3ware card. (modern 3ware cards like the 9650SE-12ML are about $800.. when 3 sata controllers would be about $140)

We have a number of sata ports on the motherboard already, so yesterday I tried destroying the 5 disk zpool, disconnecting the disks from the 3ware card and plugging them into the motherboard sata connectors.

When I booted up the box, I could see that the disks were seen however the console was dumping huge amounts of lines like this:

ad10: FAILURE - READ_DMA status=51<READY,DSC,ERROR> error=4<ABORTED> LBA=0
ad10: FAILURE - READ_DMA status=51<READY,DSC,ERROR> error=4<ABORTED> LBA=512
ad10: FAILURE - READ_DMA status=51<READY,DSC,ERROR> error=4<ABORTED> LBA=64
ad10: FAILURE - READ_DMA status=51<READY,DSC,ERROR> error=4<ABORTED> LBA=0
ad10: FAILURE - READ_DMA status=51<READY,DSC,ERROR> error=4<ABORTED> LBA=2
ad10: FAILURE - READ_DMA status=51<READY,DSC,ERROR> error=4<ABORTED> LBA=16
ad10: FAILURE - READ_DMA status=51<READY,DSC,ERROR> error=4<ABORTED> LBA=128
ad10: FAILURE - READ_DMA status=51<READY,DSC,ERROR> error=4<ABORTED> LBA=0
ad10: FAILURE - READ_DMA status=51<READY,DSC,ERROR> error=4<ABORTED> LBA=1
ad10: FAILURE - READ_DMA status=51<READY,DSC,ERROR> error=4<ABORTED> LBA=0
ad10: FAILURE - READ_DMA status=51<READY,DSC,ERROR> error=4<ABORTED> LBA=0

(For all the disks attached).

Checking their smart status everything was fine, when running a diskinfo on them it would recognise the disk correctly but then get read/write errors when trying to perform tests.

We tried everything we could think of yesterday, but couldn't resolve it. In the end we repluged the disks back into the 3ware card and they came up fine again.

Does anyone know why this might be happening? I would assume since we are just using straight passthrough (jbod) via the 3ware card the disk should just show up in tact? Is there anything we need to do to the disks to get them to show up?

Any thoughts/comments/help to point us in the right direction would be much appreciated.

link|improve this question

59% accept rate
2  
The READ_DMA errors almost always indicate a hardware failure of some sort. Since the only change was the new controller its probably safe to say there is an issue there. Have you tried the disks via the on-board SATA ports just to see if they work there (proving the controllers are the issue)? Also, its possible but unlikely that there is a geometry issue where you'd need to repartition the drive to use it on another controller (but that doesn't yield READ_DMA errors usually). – polynomial Oct 11 '11 at 1:56
feedback

Know someone who can answer? Share a link to this question via email, Google+, Twitter, or Facebook.

Your Answer

 
or
required, but never shown

Browse other questions tagged or ask your own question.