[rescue] raid array failure fun

[Derrick Daugherty]

It's rumored that around Sun, Aug 26, 2001 at 07:10:48PM -0400
Kurt Huhn <kurt at k-huhn.com> wrote:
> >
> > The NetApp SE, a pal of mine, came out and saved the day by swapping the
> > logic board from a spare drive to one of the failed drives.  Poof,
> > system came up, started rebuilding the RAID.
> >
> 
> There's a trick you only learn about with experience!  I suppose htough,
> that only coveres you in hte event of a logic board failure - a head crash,
> or other mechanical failure, would probably still mean a ruined day.
> 
> I haven't even had a minor burp in my F820's operation <knock knock>.

IIRC from their sales pitches and me always voicing the concern with
RAID4 they have ways around even the two disk failure.  Of course for
the life of me I can't remember what it is right now.  It's part of
their WAFL file system and the parity disk, it's basicaly double xor of
the data.  you have hte parity disk, and then you have a log file of the
fs with parity data as well, or something similar.  So even if you don't
mirror your parity disk, and loose it, you're still ok.  One thing cool,
that info, the os and your logs, is saved on all disks, so it doesn't
matter which one goes.

I've had several F720 failures.  The hardware failures have been power
supplies and faulty drive bays.  The OS has been my biggest nemesis.

I was a happy camper when I handed those over to the net-ops group.

Oh, and I mirror my raid volumes for all important databases.
Stripe+mirror, I would imagine a lot of other people on here do as well
;)

Only raid5 I have is on the backup array where data is transferred
across the san direclty to disk then offloaded to the tape lib.  

Which should all be auctioned off sometime soon :D  There's _no_
physical way to get the emc 8730 out of the basement w/o a blow torch,
mu-hahahah

^Derrick