OT: Advice on Western Digital warranty

[Bruno Prior <bruno@xxxxxxx>]

Sorry for the OT post, but I have seen some pretty useful discussions about
consumer rights on here before, so hoped people might be able to help me.

I bought 4 Western Digital 120Gb hard disks from CCL last October. They are
for use in a Software-RAID configuration on my home (linux) server. When I
first installed them, one of them would appear to fail within a day or two,
and a second would go down within a week (crashing the server and
corrupting
the filesystem if I didn't sort it out before the second disk crashed).
There was no strict pattern. Any of them could go down, although one was
worse than the other three. When they went down, the machine would not soft
reboot, as the disk appeared to still be dead when the machine came to
reboot. However, leaving it to stand for a short while (a few seconds would
do) would see the disk spring back to life again when I turned the machine
back on.

My first thought was that it was from overheating - they were in removable
caddies, which, although ventilated, might have led to increased
temperatures from the lack of circulation. So I first removed them from the
caddies, which didn't help, and then invested in a Lian-Li case with two
fans blowing air over the open drive cage. In this case, they never get
more
than lukewarm to the touch, yet the problem continued.

At this point, I decided it wasn't to do with temperature. I contacted CCL,
who suggested I swap the worst disk. However, given the performance
problems, I was unwilling to send them a disk to test, and for them to send
me a replacement after testing, as that would leave the RAID running in
degraded mode, which was too risky given the frequency of failure. So I
bought another disk to swap out the worst disk (CCL agreed to refund me if
the disk proved to be faulty when I returned it). (As an aside, WD had
changed their disk specs in the meantime, which meant the new disk was
marginally smaller than the old one, which is a nightmare if you are
running
RAID-5, as I am). This reduced the failure frequency to 1-2 weeks, but that
is still much too often for a server.

As all 5 disks (including the new model) presented the same problem, I
continued to believe that the problem must be in my configuration, rather
than with the disks themselves. My next attempt to solve the problem was to
replace the motherboard, as discussions with various linux gurus indicated
that this was symptomatic of DMA problems, and the Athlon/Via combination
was known to have DMA issues. The new motherboard (Gigabyte GA-8PE667
Ultra2) is pretty high-spec - an Intel 845PE-based board with a P4/2.4.

Around the same time, I was pointed to an article on WD's website that
noted
that exactly this problem was known to occur with their 200Gb and 180Gb
drives. The person who pointed me to the link believed that the 120Gb
drives
had the same problem, and my experience appeared to bear that out. I
downloaded their DOS utility and upgraded the firmware on the drive
controllers.

These two actions appeared, at first, to have solved the problem. The
machine went for over a month without a disk dropping out of the array!
Then
it started again, and now the disks are dropping out just as frequently as
before, and getting worse.

I have no idea why this is happening, but the point is, I think I have
tried
more than most to solve the problem. I am all out of ideas now, and my only
solution seems to be to get a different set of drives. I am still within
warranty (3-year - first with CCL, 2nd and 3rd with WD), so I could
theoretically send them back one by one. But given the intermittency, there
is every chance that CCL will not see the problem when they test it. And if
they do, I will presumably get another WD disk, and I have every reason to
expect that it would present the same problems.

CCL Customer Support have not been bad, but they are not able to simply
refund me so that I can get a different set of disks. Apparently they could
have done this if I had sent the disks back claiming incompatiblity within
the first 30 days. So I shot myself in the foot trying to figure this out.
The best they have been able to come up with is an offer to accept all four
disks back for testing, so they can try to recreate my configuration as
closely as possible for testing. But I am not willing to do that, as (a)
that means my home server, and therefore all the other machines (which NFS
mount /home from the server) will be unuseable until they have finished
testing, (b) I have no confidence that they can recreate anything remotely
like my configuration without sending the whole box (which they won't
accept
as it may contain components not bought from them - although most were),
and
(c) even if they recreate the conditions, the fault is intermittent, so
they
may not see it unless they test for a longer period than would be
acceptable
to either them or me.

It seems my only course of action is to send the disks back one by one (I
have one in hand as I never sent back the bad one that I replaced - so I
can
swap them out one by one without having to run the array degraded). But if
their tests don't show the fault, I will get the same disk back and a bill
for the handling. And if their tests do find the fault, I will just get
another one of these disks, which I'd rather not have. I can see myself in
an endless loop of sending back disks until the warranty expires.

Is there anything I can do about this? I don't think CCL's position is
unreasonable. Is my only option to "bin" these disks and live
with £470 down
the drain, or do I have some recourse against Western Digital? It is
apparent to me that these disks are not fit for purpose, but proving it is
another matter.

Cheers,

Bruno