Strange case of vanishing disk

Kaya Saman kayasaman at
Mon Jun 4 13:52:16 UTC 2012

On 06/04/2012 08:34 AM, Gary Aitken wrote:
> On 06/03/12 20:59, Kaya Saman wrote:
>> this is a very strange issue but I guess will either be related to 2
>> things, PSU not being powerful enough or disk controller simply being crap.
>>> Here's what's going on. I have a little Chenbro 4 disk mini-ITX NAS
>>> server with 2x 2TB disks and 2x4TB disks as storage - all spread out
>>> over 2 ZFS storage pools. Additionally I am running the root file system
>>> on a 40GB SSD.
>> [...]
>>> _______
>>> One thing I can think of is to disconnect the questionable disk from the RAID controller card and connect it directly to the motherboard.
>>> Then you'd know whether the fault is with the hard drive or the RAID controller.
>>> PSU = power supply unit? 180 watts seems very little, I didn't know any modern system could run on so little. I thought the minimum would be around 400 watts, and this would not allow for a powerful gaming graphics card.
>>> Maybe you need to replace the power supply with something having more watts, but make sure it will physically fit.
>>> Tom
>> Thanks for the response!
>> Here's some more info that I managed to dig up:
>> Jun 4 02:39:19 Zeta-Ray root: ZFS: vdev I/O failure, zpool=ZFS_POOL_2 path=/dev/ad4 offset=270336 size=8192 error=6
>> Jun 4 02:39:19 Zeta-Ray kernel: ata2: port is not ready (timeout 15000ms) tfd = 000000ff
>> Jun 4 02:39:19 Zeta-Ray kernel: ata2: hardware reset timeout
>> Jun 4 02:39:19 Zeta-Ray kernel: unknown: TIMEOUT - WRITE_DMA48 retrying (1 retry left) LBA=269091394
>> Yeah, 180 Watts is what comes with the chassis as it's an external power supply. Additionally the system is a Mini-ITX so that would account for less power usage however, in this case I think it might be the PSU that's simply not providing enough power.....
>> I will definitely try sticking the "downed" disk into the motherboard controller directly as that will tell me if the disk is the issue or not.
> If the problem is actually insufficient power, this won't tell you a thing.
> You'll have to isolate the power supply as not being a problem before anything else will be relevant.
> If you swap the two new disks, and the one now on the card fails, it's probably not a disk problem.  But you still can't tell if its the card or insufficient power.
> If you can sideline the two original disks and run, it's probably power.  But I'd guess you're oversubscribed in that department.  It should be relatively easy to estimate as mfg specs for cpu + mobo + disks is readily available.
> Gary

Yeah, this is really odd!

I just "offlined" the system and took a look at the BIOS. The hard disk 
controller was set to "ATA NATIVE", I attempted changing to AHCI and the 
system failed to boot thereafter.

Booting into my rescue CD of FreeSBIE, with AHCI enabled only showed 
"ad4" (the troubled disk) to be registering.

Switching the setting back I managed to see all disk including ad4 from 
SBIE. I fixed the /boot/loader.conf file by commenting out 
"ahci_load="YES"" and vuala done! The system booted and the ZPOOL jumped 
back into life:

# zpool status
   pool: ZFS_POOL_1
  state: ONLINE
  scrub: none requested

     ZFS_POOL_1  ONLINE       0     0     0
       ad13      ONLINE       0     0     0
       ad15      ONLINE       0     0     0

errors: No known data errors

   pool: ZFS_POOL_2
  state: ONLINE
  scrub: scrub stopped after 307445734561825857h27m with 0 errors on Mon 
Jun  4 16:24:10 2012

     ZFS_POOL_2  ONLINE       0     0     0
       ad14      ONLINE       0     0     0
       ad4       ONLINE       0     0     0

errors: No known data errors

Here's the weird part now; I had to stop the SCRUB as the READ_DMA 
retrying errors came up again??

I have just swapped disks infact and ad4 seems to be playing up 
again.... :-(

ad4 of course being assigned any disk that's on the controller.

I am not sure of what to make of this?



More information about the freebsd-questions mailing list