10.1 RC4 r273903 - zpool scrub on ssd mirror - ahci command timeout

Steven Hartland killing at multiplay.co.uk
Mon Dec 22 14:26:53 UTC 2014


On 22/12/2014 14:09, Peter HEINER wrote:
>> So I now think the whole problem centers around the Samsung 850 PRO
>> 512GB SSDs. Too bad I do not have the budget to just buy two Intel (or
>> other) SSDs of similar size and see if the timeouts disappear..
>>
>> I wonder if this is a firmware issue with the drive or just some
>> misguided fancy energy saving feature of this particular drive
>> model causing the whole trouble.
> I don't think it's the SSDs, as I'm seeing the exact same behaviour
> using 4 2TB WD Reds in a raidz1 with 10.1-RELEASE on a HP N40L as well.
> Whenever I do heavy reads like a scrub or send, there are CAM timeouts
> and  eventually the drives detach. I had to downgrade to 10.0-RELEASE,
> which seems to work fine so far. The box ran Ubuntu until recently,
> without issues.
> I am happy to troubleshoot further, should anyone want to pick this up.
> I'm relatively new to FreeBSD, so any guidance on next steps to take is
> welcome.
Are all the disks your seeing errors with running at 6Gbps?

If so have you tried downgrading the connection to SATA 2 speeds to see 
if that makes
a difference?

An example of this which goes in /boot/loader.conf is:
hint.ahcich.0.sata_rev="2"

We had a similar problem with a Dell chassis where the connection via 
the Hotswap
backplane wasn't up to scratch for SATA 3 speeds, and this fixed it.

     Regards
     Steve


More information about the freebsd-stable mailing list