[Wylug-help] Inactive RAID 10 Array

[Dave Fisher]

On Wed, Apr 15, 2009 at 08:36:16AM +0100, John Hodrien wrote:
> On Wed, 15 Apr 2009, Dave Fisher wrote:
> 
> > As I implied in the original post, most of the data is unimportant, but
> > a tiny part of it could get me in me in very hot water with HMRC  ...
> > and I am not too proud to admit that the prospect scares me.

Just to clarify. It's not the data itself that might get me into trouble
with HMRC ... it's the absense of that data, e.g. copies of receipts,
invoices.  

In truth, I can probably reassamble most of the proofs I need from paper
records (scattered all over), but only at the cost of weeks of
mind-numbing paperwork.

> If it matters that much to you, make sure you're protecting against fire,
> theft, hardware failure and human error.

Point taken. In my case the error was not to ignore backups, but put off
the job until a 'more convenient time' ... yes, I can see the irony ;-)

Some of the data loss could just as easily been caused by insufficiently
*frequent* backups.

> > If that wasn't scary enough, adding new disks has completely changed all
> > the /dev/sd* names ... although I should be able to find their UUIDs ...
> > if only I could remember the necessary incantation.
> 
> The easiest way to prove you've got a good backup of those disks, is to stop
> using the originals.  Can you drop the new PCI-X controller + disks in another
> machine, and try to rebuild the array on the backup?

Good idea, but I'd have to buy a new processor and motherboard to run
the amd64 version of the OS on my working array.

I was thinking about simply disconnecting the SATA data cables from the
original drives ... remembering to map them, first.

Is there any reason why I shouldn't do it that way?

> If you don't, you'll surely find that you've now got duplicate UUIDs (since
> the backups will match won't they?).

Good point.

Dave