[clug] raid question

[Eyal Lebedinsky]

Thanks Steve,

While not extremely expensive, this disk is not a cheap one either. I run raid6 so that
I can futz with one disk and still have redundancy, so I do not need a disk on the shelf
(as I did when I used raid5).

One reallocated sector is unlikely to be acceptable for RMA (the disk is early into its
5y warranty). I had such disks that continued to work for many years until replaced
still in working order.

Reallocation is OK and it does not scare me at all, but I do want to understand why the
software raid does not trigger an i/o error when directly accessing a member of the
array does. I can see how simply reading the fs over the array can miss the bad sector
(checksums do not need to be read, maybe there is an option to request this?) but not
how a specific 'check' request misses it.

cheers
	Eyal

On 02/20/14 16:33, steve jenkin wrote:
> Disks are cheap and you've got a spare sitting on the shelf for when a disk in your array breaks, don't you?
>
> Pop the suspect drive and replace it, then let the system rebuild under the mounted filesystem. If you're not using hot-plug, this is a power-down...
> You're running RAID-6 exactly because it will cope with _two_ failed drives.
>
> During the rebuild you can sustain another failed drive. You're not putting your data at increased risk.
>
> You can then mount the suspect drive somewhere else and stress test it...
>
> If it's not 12 months old, it's covered by a statutory warranty & the drive vendor is obliged to replace it for you.
>
>
>
> On 20/02/2014, at 1:08 PM, Eyal Lebedinsky wrote:
>
>> In short: smartctl lists one pending sector. A dd provokes an i/o error as expected.
>> An mdadm 'check' does not find a problem and does not trigger an i/o error. Why?
>
> --
> Steve Jenkin, IT Systems and Design
> 0412 786 915 (+61 412 786 915)
> PO Box 48, Kippax ACT 2615, AUSTRALIA
>
> mailto:sjenkin at canb.auug.org.au http://members.tip.net.au/~sjenkin

-- 
Eyal Lebedinsky (eyal at eyal.emu.id.au)