[ILUG] HOWTO recover a borked 3ware RAID-1 array (long)

[Joseph A o'Loughlin]

Hi Ronan,

I replied off list from my mobile phone which has a toy email client and as
you can see I can't spell.

I also find the quality of raid going downhill, for example on motherboards
and sata/pata to be mostly pants, and agree with your recommendation to use
software raid with MD.

On my home server I'm using a LSI/ATI MegaRaid controller with the <CTRL>M
bios but have seen turkeys from LSI on blade servers.  It's happily worked
its way through a bunch of disks rescued from a skip.  While the controllers
list price was circla £IR 3,000 you can still pay top dollar and get a
controller that is lacking when it comes to recovering from a failure.

It's a pity that it's only from the school of hard knocks that one discovers
why antivirus programs have an option to disable scanning boot sectors.

Had a similar problem with a hotel where you'd get a shock from the
radiators.  Ground loops were a huge problem.  Ended up using specialist
RJ45 cables with a 9th wire running through the clip part connected to each
chassis using tinfoil tape.  The girls in booking and reception by habit use
the plastic door keys etc to manipulate anything that's shocked them before.

While I've had no complaints since, have not been there since...

Joseph


-----Original Message-----
From: Ronan Cunniffe [mailto:ronan at iaa.es] 
Sent: 02 November 2006 19:00
To: Joseph o'Loughlin
Subject: Re: [ILUG] HOWTO recover a borked 3ware RAID-1 array (long)

Joseph o'Loughlin wrote:
> Hi Ronan,
>
> Just a fyi
>
> Sopme antivirus scanners, when trying to determine if their compromised by
an active memory resident virus that remaps disk sectors to conceal itself
ask the hard drive whether it can access a location at anegitave number of
heads, cylendars, sectors, and in so doing corrupt controllers that use disk
coersion.  
>   
Yikes!  You'd think the controllers would, you know, maintain the facade....
> Did you see any dmesg entries reflecting this happening.
>   
Hah.  Um, no.  Two servers in the rack; both dropped dead at the same 
time.  Victim B is getting its second mobo transplant this year.  The 
site is up on a mountain ridge, on granite.  Local ground potential has 
been measured more than 70V away from the middle of the 3-phase.  
Everything is *supposed* to be Faraday caged and isolation 
transformer'd.  Nobody has found what's wrong, but PCs just.... die.  
(And don't give me crap about UPSs, they're *on* an overspec'd UPS.  
Their second in 18 months, IIRC).

Anyway, at some point the Big People are going to figure that paying an 
electrician/exorcist would be cheaper than replacing all the blackened 
bits, but that hasn't happened yet, so ve patch zem up unt send zem back 
to Ze Eastern Front.

Did you mean to post off-list, BTW?  That's a very useful piece of 
knowledge....

Ronan

__________ NOD32 1.1821 (20061021) Information __________

This message was checked by NOD32 antivirus system.
http://www.eset.com