Raptor Failure

[dh]

Has anyone had their 36gb WD Raptor fail? One of my 14 month old
Raptors has just died. Click Zing, Click Zing six times and then spin
down, even with the data cable disconnected. With a quoted
"Unsurpassed Reliability: 5-year warranty and 1.2 million hours MTBF",
I am a little surprised it went after ~ 5000 hours (425 days * 12
Hours). My box is well ventilated and the drive stack runs cool to the
touch and this is the first drive failure I have had since the IBM
DeskStar days.
I did a Google search for other Raptor failures but did not find any.
WD gave me a hard time about warranty replacement, said it was an OEM
drive and to go back to the vender. Newegg.com clearly states warranty
by manufacture and WD finally relented and sent me a replacement (
still waiting to arrive, shipped in two days but still waiting for UPS
ground for a week). Just wondering if I should consider RAID 1 even
though I ghost every other day.

 

[ted msn]

Has anyone had their 36gb WD Raptor fail? One of my 14 month old
Raptors has just died. Click Zing, Click Zing six times and then spin
down, even with the data cable disconnected. With a quoted
"Unsurpassed Reliability: 5-year warranty and 1.2 million hours MTBF",
I am a little surprised it went after ~ 5000 hours (425 days * 12
Hours). My box is well ventilated and the drive stack runs cool to the
touch and this is the first drive failure I have had since the IBM
DeskStar days.
I did a Google search for other Raptor failures but did not find any.
WD gave me a hard time about warranty replacement, said it was an OEM
drive and to go back to the vender. Newegg.com clearly states warranty
by manufacture and WD finally relented and sent me a replacement (
still waiting to arrive, shipped in two days but still waiting for UPS
ground for a week). Just wondering if I should consider RAID 1 even
though I ghost every other day.

I dont see the problem! It has a 5 yr warranty and you got a new drive. Did
you expect the drive to die at 1.2milliion hours + 1 second. The MT of MTBF
is Mean Time between failure.
I think that drives suffer from the "bath tub" failure curve ie lots at the
start (we hope the manufactures catch these, and lots at the end (old age)
in between there will be a small number that will die every so often. I have
said before that there are only two types of disk those that have died and
those that will! the trick is to pass on the disk / system to someone else
before it dies!!

Other people (in mags) have said that any RAID with less than 3 disks is not
worth having (in RAID1 how do you know which disk has the good data and
which the bad if things start to go wrong? ie not a total loss of a disk and
if it starts to rebuild the array with the duff data!!).

Me I go with RAID 5 on 4 disks. So I now have four dsiks that can fail
:-( ( and backup to other media) it all depends on how much you "need" the
data and when! Can you do without it for hours, days, weeks. If the PC gets
stolen can you still recover from this? Only you know how much the cost of
time and effort as well as money is required to make "you" happy.
regards
ted

 

[Ron Reaugh]

Other people (in mags) have said that any RAID with less than 3 disks is not
worth having

That's nonsense. Two drive RAID 1 is very effective.

(in RAID1 how do you know which disk has the good data and
which the bad if things start to go wrong? ie not a total loss of a disk and
if it starts to rebuild the array with the duff data!!).

Huh, one drive will be reporting read and/or SMART failures and the other
wont. The non-failing drive becomes the dominant.

Me I go with RAID 5 on 4 disks. So I now have four dsiks that can fail

Yep, if two drives fail then you've lost it all.

 

[Peter Hucker]

Has anyone had their 36gb WD Raptor fail? One of my 14 month old
Raptors has just died. Click Zing, Click Zing six times and then spin
down, even with the data cable disconnected. With a quoted
"Unsurpassed Reliability: 5-year warranty and 1.2 million hours MTBF",
I am a little surprised it went after ~ 5000 hours (425 days * 12
Hours). My box is well ventilated and the drive stack runs cool to the
touch and this is the first drive failure I have had since the IBM
DeskStar days.

I've had it with hard disks failing - hardly any make past the end of the warranty! Tried all the makes, they all get really hot and wear out. Got myself four of em now - a stripe/mirror set, each with cooler fins attached.

I did a Google search for other Raptor failures but did not find any.
WD gave me a hard time about warranty replacement, said it was an OEM
drive and to go back to the vender. Newegg.com clearly states warranty
by manufacture and WD finally relented and sent me a replacement (
still waiting to arrive, shipped in two days but still waiting for UPS
ground for a week).

You say "425 days" - that's over a year - never heard of a retailer doing over 1 year! 1 year - end of warranty it's up to the manufacturer surely?

Just wondering if I should consider RAID 1 even

though I ghost every other day.

--
FOURTEEN - CHECK OUT THE BABY! parrots and rising http://www.petersparrots.com
93 silly video clips http://www.insanevideoclips.com
1259 digital photos http://www.petersphotos.com
Served from a pentawatercooled dual 2.8GHz silent Athlon with half TB RAID.

I got the strangest recording when I called the phone company the other day.
It said, "You have been connected to the correct department on the first try. This is against company policy. Please hang up and redial."

 

[Peter Hucker]

I dont see the problem! It has a 5 yr warranty and you got a new drive. Did
you expect the drive to die at 1.2milliion hours + 1 second. The MT of MTBF
is Mean Time between failure.
I think that drives suffer from the "bath tub" failure curve ie lots at the
start (we hope the manufactures catch these, and lots at the end (old age)
in between there will be a small number that will die every so often. I have
said before that there are only two types of disk those that have died and
those that will! the trick is to pass on the disk / system to someone else
before it dies!!

Bath tub indeed. I could draw the curve I've experienced, and it would not be a curve! They fail at 1 month, 2 months, ........, all the way to a few years!

Other people (in mags) have said that any RAID with less than 3 disks is not
worth having (in RAID1 how do you know which disk has the good data and
which the bad if things start to go wrong? ie not a total loss of a disk and
if it starts to rebuild the array with the duff data!!).

Doesn't the OS know when the data is wrong? CRC, parity, whatever it does? With one disk, you get a report of a disk error. With 2, you'd get a report of a filure on drive 1, so it would use the data on drive 2, ansd you'd change drive 1.

--
FOURTEEN - CHECK OUT THE BABY! parrots and rising http://www.petersparrots.com
93 silly video clips http://www.insanevideoclips.com
1259 digital photos http://www.petersphotos.com
Served from a pentawatercooled dual 2.8GHz silent Athlon with half TB RAID.

"I wonder who discovered we could get milk from cows and what the **** did he think he was doing?!" -- Billy Connolly

 

[Peter Hucker]

That's nonsense. Two drive RAID 1 is very effective.


Huh, one drive will be reporting read and/or SMART failures and the other
wont. The non-failing drive becomes the dominant.

[smacks self in face]

I just wrote that. Again.



--
FOURTEEN - CHECK OUT THE BABY! parrots and rising http://www.petersparrots.com
93 silly video clips http://www.insanevideoclips.com
1259 digital photos http://www.petersphotos.com
Served from a pentawatercooled dual 2.8GHz silent Athlon with half TB RAID.

Why isn't 11 pronounced onety one?

 

[Andy Lee]

Other people (in mags) have said that any RAID with less than 3 disks is not
worth having (in RAID1 how do you know which disk has the good data and
which the bad if things start to go wrong? ie not a total loss of a disk and
if it starts to rebuild the array with the duff data!!).

Any raid controller worty of the name will tell you which disk is
failing. Even the Promise controllers on motherboards tell you that
much. Leastways the ones in the two machines I built for the office
have done so when a disk when tits up in both of them recently.

Me I go with RAID 5 on 4 disks. So I now have four dsiks that can fail
:-( ( and backup to other media) it all depends on how much you "need" the
data and when! Can you do without it for hours, days, weeks. If the PC gets
stolen can you still recover from this? Only you know how much the cost of
time and effort as well as money is required to make "you" happy.
regards
ted

Very good point ted These are the sort of calls people should make
when deciding Backup/redundancy stratagies It's fat too easy to spend
serious bucks on this sort of thing when in reality the value of the
data held is in the cold light of day not really that "vital" after
all.

 

[ted msn]

Any raid controller worty of the name will tell you which disk is
failing. Even the Promise controllers on motherboards tell you that
much. Leastways the ones in the two machines I built for the office
have done so when a disk when tits up in both of them recently.

I agree that the theory says you "should know" but if I can quote in part
from an item in PC Pro (UK) nov 2003 by Steve Cassidy, in talking about
RAID failures...
"I have never, ever, encountered a mirrored RAID setup that worked as
advertised"
"The gotcha isn't in the theoretical design, but rather in the failure mode"

He goes onto say that while the systems might say they are working OK behind
in the background they can be writing all sorts of duff data to the mirror
before they fail.

He says that in effect you get what you pay for, big expesive RAID cards are
"good" if you need data 24/7 and simple on motherboard systems/cards are
good if you wish to feel good at being with the "big boys" but dont think
you will get the same level of service from the system if things go AWOL.

As a caution, in the company I worked for a few years ago (no names!) "we"
had a system with an expensive RAID setup but the email notification of
errors was not set up no-one had any idea that the RAID 5 had lost one of 4
disks until the second died, ho hum! (backups are great when they work!!)
regards
ted

 

[dg]

As a caution, in the company I worked for a few years ago (no names!) "we"
had a system with an expensive RAID setup but the email notification of
errors was not set up no-one had any idea that the RAID 5 had lost one of 4
disks until the second died, ho hum! (backups are great when they work!!)

My dad worked for a company that had major data loss. They didn't verify
the tape backup and it turns out the tape drive hadn't made a valid backup
in months or years-didn't throw an error message either. As far as raid
working or not, I try not to consider raid a backup solution, just a more
robust storage method.

--Dan

 

[Ron Reaugh]

is disk
I agree that the theory says you "should know" but if I can quote in part
from an item in PC Pro (UK) nov 2003 by Steve Cassidy, in talking about
RAID failures...
"I have never, ever, encountered a mirrored RAID setup that worked as
advertised"
"The gotcha isn't in the theoretical design, but rather in the failure mode"

He goes onto say that while the systems might say they are working OK behind
in the background they can be writing all sorts of duff data to the mirror
before they fail.

That's pure crap by some wacko you found somewhere.

He says that in effect you get what you pay for, big expesive RAID cards are
"good" if you need data 24/7 and simple on motherboard systems/cards are
good if you wish to feel good at being with the "big boys" but dont think
you will get the same level of service from the system if things go AWOL.

BULL!

 

[ted msn]

That's pure crap by some wacko you found somewhere.
AWOL.

BULL!

facts are always good, a simple rant against a "journo" is very cheap, he
earns his living by what he writes what do you do? note he does say failure
mode, how many RAID systems have you seen fail, what type, names, types and
numbers?
just asking, as I wish to know if there is an issue with what Steve writes
as he has always seem to give good advice in the past.
regards
ted

 

[Ron Reaugh]

facts are always good, a simple rant against a "journo" is very cheap, he
earns his living by what he writes what do you do? note he does say failure
mode, how many RAID systems have you seen fail, what type, names, types and
numbers?
just asking, as I wish to know if there is an issue with what Steve writes
as he has always seem to give good advice in the past.

Simply trash.

 

[Arno Wagner]

news:[email protected]...

[...]
facts are always good, a simple rant against a "journo" is very cheap, he
earns his living by what he writes what do you do? note he does say failure
mode, how many RAID systems have you seen fail, what type, names, types and
numbers?
just asking, as I wish to know if there is an issue with what Steve writes
as he has always seem to give good advice in the past.

There are some people here that positively seem to hate anybody with
true competence or insights. Possibly because they try to impress
people with half-knowledge and those that truly know make it
hard for them. Indicators are e.g. the lack of manners, the
putting down of things said with no or insufficient explanation,
the classification of some people (e.g. me) as "trolls" and the
often amusingly naive technical opinions.

My solution is to mostly ignore them. Works well.

Arno

 

[Eric Gisin]

So Arnie, how many people respect your uninformed opinions? Ever consider
people are called trolls for a reason?

 

[0_Qed]

There are some people here that positively seem to hate anybody with
true competence or insights. Possibly because they try to impress
people with half-knowledge and those that truly know make it
hard for them. Indicators are e.g. the lack of manners, the
putting down of things said with no or insufficient explanation,
the classification of some people (e.g. me) as "trolls" and the
often amusingly naive technical opinions.
My solution is to mostly ignore them. Works well.

Bravo!
Well said.

Qed.

 

[Folkert Rienstra]

"ted msn" <[email protected]> wrote in message news:[email protected]...

[...]
facts are always good, a simple rant against a "journo" is very cheap, he
earns his living by what he writes what do you do? note he does say failure
mode, how many RAID systems have you seen fail, what type, names, types
and numbers?
just asking, as I wish to know if there is an issue with what Steve writes
as he has always seem to give good advice in the past.

There are some people here that positively seem to hate anybody with
true competence or insights.

Blowing your own trumpet again, Arnie, since noone else does it for you?

Possibly because they try to impress people with half-knowledge

The sort of half-knowledge that Ultra DMA mode 2 doesn't have CRC
detection, Arnie? That ATAPI command set is much simpler than SCSI?
Like IDE-flashdisk somehow not being removable flash disk?
That glass platters can be bend, like a pretzel?
That Command queuing is only in the next SATA standard?
That the DOS drive letter system is braindead?
That only IBM can unlock Travelstars?
That the IDE interface specification does not support
an unpowered drive because of short circuiting?
Hot-plugging is a newer trend?

That sort of half-knowledge, Arnie?

and those that truly know make it hard for them.

Sounds like a pretty good description of yourself, Arnie.

Indicators are e.g. the lack of manners, the putting down
of things said with no or insufficient explanation,

Like "Expect it to die soon", Arnie?
"Obviously you don't know how a modern differential junction temperature sensor works", Arnie?

the classification of some people (e.g. me) as "trolls"

Which of course you are when you admit to litter your
messages with typoes every time, " to keep us happy".

and the often amusingly naive technical opinions.

Poor Arnie who hasn't got the sole right to do that himself.

My solution is to mostly ignore them. Works well.

Ofcourse it does.
Why draw more attention to yourself when you had to be
corrected once again, right, Arnie "babblemouth" Wagner?

 

[dh]

Getting back to the original thread, received the replacement drive
and installed. It is a refurbished drive and that is ok, but boy is
this drive LOUD. It is out of balance, shakes the whole cabinet. It is
MUCH louder than my other two Raptors and gets much warmer to the
touch. I had to remove it from its normal mounting place and place it
on a piece of foam rubber in the bottom of the case otherwise it
causes the side panels to buzz and rattle.
Heat and vibration are the two things that can kill a hard disk and
this drive generates both on its own. I was afraid it might hurt my
other drives, so yet another replacement is on its way.

 

[Arno Wagner]

Previously ted msn <[email protected]> wrote:

[...]
The sort of half-knowledge that Ultra DMA mode 2 doesn't have CRC
detection, Arnie? That ATAPI command set is much simpler than SCSI?
Like IDE-flashdisk somehow not being removable flash disk?
That glass platters can be bend, like a pretzel?
That Command queuing is only in the next SATA standard?
That the DOS drive letter system is braindead?
That only IBM can unlock Travelstars?
That the IDE interface specification does not support
an unpowered drive because of short circuiting?
Hot-plugging is a newer trend?

That sort of half-knowledge, Arnie?

This guy has a _list_! ROTFL!
I must be a real threat to him!

My day is made! =============)

Not that everything on his list is actually a mistake and disregarding
that a zero-mistake requirement is completely unusable to identify
competent people...

Arno

 

[Folkert Rienstra]

Previously ted msn <[email protected]> wrote:

[...]
The sort of half-knowledge that Ultra DMA mode 2 doesn't have CRC
detection, Arnie? That ATAPI command set is much simpler than SCSI?
Like IDE-flashdisk somehow not being removable flash disk?
That glass platters can be bend, like a pretzel?
That Command queuing is only in the next SATA standard?
That the DOS drive letter system is braindead?
That only IBM can unlock Travelstars?
That the IDE interface specification does not support
an unpowered drive because of short circuiting?
Hot-plugging is a newer trend?

That sort of half-knowledge, Arnie?

This guy has a _list_! ROTFL!

Challenge and you might get what you don't wish for.
Actually, you are giving yourself more credit again for something that
sits in my Sent Items folder for free. I only have to run a search on
your name and voila.

I must be a real threat to him!

Wrong again, Arnie, all it means is that you are a threat to the newsgroup.

My day is made! =============)

Good for you. It's an imbecile that actually enjoys being hung out to dry.

Not that everything on his list is actually a mistake

You mean, you really stand by those clueless rantings?

and disregarding that a zero-mistake requirement is completely
unusable to identify competent people...

Actually I stopped at 1/3 that list, Arnie. There was lots more.
You are miles away from zero-mistake and hardly competent.

 

[Andy Lee]

I agree that the theory says you "should know" but if I can quote in part
from an item in PC Pro (UK) nov 2003 by Steve Cassidy, in talking about
RAID failures...
"I have never, ever, encountered a mirrored RAID setup that worked as
advertised"

All I can say to that is his working use and knowledge of Mirrored
Raid setups must be very limited. COMPAQ/HP Raid controllers used with
Insight Manager give very detailed reports on any write/read errors
they encounter and tell which drive incurred the errors.
I read PC Pro every month (subscribed through work) And this sort of
article just confirms my feeling that some the writers on the mag have
very rarely worked in the industry and that the technical knowledge is
patchy in the extreme. The article critisising overclocking last year
prompted a letter response from me which is very rare (Not published
probably due to my language). Now they have released the "Extreme"
section which just strikes me as pandering to a fad. If I wanted crap
on led illuminated fans and lights in my PC I read Max Power or what
ever the PC equivelent is called.

"The gotcha isn't in the theoretical design, but rather in the failure mode"

He goes onto say that while the systems might say they are working OK behind
in the background they can be writing all sorts of duff data to the mirror
before they fail.

Well in my admitted limited use of motherboard devices like the
Promise and Si devices they have performed as expected i.e. I don't
expect the level of info provided by the Compaq/HP devices I use on a
daily basis but the do provide a level of protection comparable to the
price paid. They have recently saved me from having to do a full
system rebuild and restore of 2 SMS servers so I'm not going to
critisize them. As somebody else has already stated they are not an
alternative to backups but a complement to it.

He says that in effect you get what you pay for, big expesive RAID cards are
"good" if you need data 24/7 and simple on motherboard systems/cards are
good if you wish to feel good at being with the "big boys" but dont think
you will get the same level of service from the system if things go AWOL.

I don't under stand what he means by same level of service but anybody
who works with IT equipment for any length of time will subscribe to
the "get what you paid for" philosophy

As a caution, in the company I worked for a few years ago (no names!) "we"
had a system with an expensive RAID setup but the email notification of
errors was not set up no-one had any idea that the RAID 5 had lost one of 4
disks until the second died, ho hum! (backups are great when they work!!)
regards
ted

Bad practice then if nobody can be bothered to check a simple thing
like insight manager or any of the other utilities that ALL the big
vendors provide with their kit. Don't blame the kit because of
somebodies laziness or incompetance. Even Promise provide a Raid
monitoring utility which lives in the system tray

Err who told you this? or have I miss understood you here