They come in threes...

Do you have a question? Post it now! No Registration Necessary.  Now with pictures!

Threaded View
Bad things, that is.

After a long spell of being ignored my old home-built Windows Home Server  
started to show a variety of odd problems. I used WD's diagnostics program  
to check each of the 6 disks sequentially and finally came to drive #5  
which showed stalling and slow progress in the sequential read test. That  
lead me to digging through a stack of spare drives looking for a  
replacement. That is when the first death occurred -- I accidentally  
dropped a 3tB WD Green drive onto its corner on a hardwood floor; it tested  
OK for a few minutes after the drop (whew!) but died immediately afterward  

Back to the drive stack I dug out a 2tB Green that tested good but by then  
I was unable to programmatically 'remove' the #5 drive from the storage  
array and had to sacrifice all of my backups to swap the drive. So I  
started to do some replacement backups only to find that progress was  
impossible -- backup rate was so slow that it would take days to process  
one machine. That lead me to run diagnostics on the sole untested drive in  
the server -- it was defective too in the extended test.

Back to the drive stack to recover another 2tB drive. None there. Order one  
from Amazon since they had gotten quite cheap lately. And when it arrived I  
swapped it out after once again finding it impossible to 'remove' the bad  
drive from the array. At that point I had a healthy server but one which  
had no backups so I had to start that all over again.

I really can't fault WD on the drive failures. The 3tB I dropped certainly  
wasn't their doing and the two 2tB drives that failed were quite old by  
this time -- they were bought when that model drive had just been released  
and had been running 24X7 ever since.

I think that from now on I'll maybe run a test on the server drives every  
now and then and no ignore it for years on end...

Re: They come in threes...

John McGaw wrote:
Quoted text here. Click to load it

I recommend both "SMART" checking, as well as running a
transfer rate benchmark on each drive. (You might need
to move the drives to another machine, where they're not
considered to be part of a RAID, to do your block-level

I ran into one case, where the slow read benchmark warned
that an excessive number of spared sectors were involved.
Whereas SMART showed "0" for Reallocated.

SMART works best for surveillance, if the spared sectors
are spaced uniformly over the disk. If a disk develops
a "bad spot", SMART may show a clean bill of health.
But a read transfer rate benchmark, might show 10MB
transfer rate for a 60GB patch on the disk, which is a
warning that the patch is "scruffy". And the drive
should be replaced immediately. There may be no CRC errors
in the patch, the data is still intact, but trouble for
you is right around the corner.

And drive failures can be instantaneous - the FDB spindle motor,
if it runs out of lubricant, can "seize" instantly, and no
longer rotate. Attempts to restart the drive, lead to the
"buzzing sound" they use to attempt to free up stuck
spindles. But that's not going to work, if the bearing
is dry. And apparently, the lubrication is only a
couple drops of oil. There is no "reservoir" as such.
No oil pan with 3 liters of oil.

To verify motor designs in the hard drive lab, they
weigh the motors on a milligram balance, to detect
when the two drops of oil have "disappeared". They don't
try to take the motor apart, and instead use "delta_weight"
to detect loss of lubricant. Weigh the motor before
starting a lifetime test, and weigh it again when it
seizes. That's how tiny the amount of lubricant is.


Re: They come in threes...

On 12/4/2015 1:19 PM, Paul wrote:
Quoted text here. Click to load it

The SMART results were totally useless in this case -- all six of the  
drives, a 400gB and five 2tBs, showed that they were fine. At least that is  
what the WD diagnostics showed in the SMART quick-test mode. It wasn't  
until I did a long-form test that the stalling showed up. You know that you  
are having trouble when the displayed estimated duration starts jumping up  
instead of down and settles in the range of 2000 hours.

I'm certain that the motors didn't fail here although I guess it could have  
been a speed problem of sort causing the stalling. Even the dropped 3tB  
drive was certainly spinning up since when I checked it on the eSATA port  
on my Shuttle it showed perfect gyroscopic behavior.

Re: They come in threes... Following up...

On 12/4/2015 12:05 PM, John McGaw wrote:
Quoted text here. Click to load it

On the road to destroying the drives for security I took the time to do  
complete disassembly after I realized that I hadn't actually looked into  
any modern drives.(and I wanted the magnets to play with) It was quite a  
revelation: first off was the realization that one of the 2tB was an older  
generation than I had imagined since it actually had four platters in it  
while the slightly newer ones have only three. The 3tB also had four  
platters and I believe that the full capacity of their newest drives is  
around 1.25tB per platter or maybe more as I write. Second was the amazing  
shrinkage of the heads; they are now down to a size that looks like a bit  
of very fine wire sticking out of the ends of the drive arms. The fact that  
heads so small can actually do anything and do it so quickly and accurately  
is incredible to me.

Oh, and I have eight really really powerful magnets now although I haven't  
figured out how to remove them from their metal mounting plates. Any ideas?

Re: They come in threes... Following up...

John McGaw wrote:
Quoted text here. Click to load it

In the two pictures here, it almost looks like some substance
was placed on the plate, to hold the magnet in a known position.

    "The magnets in the hard drive were glued to the brackets. Although
     the magnets look like they are made of steel, they are actually
     somewhat delicate. We pulled them off using ViceGrip pliers because
     that was the safe-for-people way to get them off. It risked damaging
     the magnet; however trying to slip a screwdriver under the magnet
     to pry it off is the you-are-probably-going-to-slip-and-hurt-yourself
     way, so, um, don?t do that."

AFAIK, the magnets are positioned in an "attracting
each other" orientation.

           N ---------- S

           S ---------- N

So a means is needed to keep them from
slamming together and pinching the voice coil.


Re: They come in threes... Following up...

On 12/11/2015 2:32 PM, Paul wrote:
Quoted text here. Click to load it

I think that I'm going to get out my 1" carpentry (as opposed to  
woodworking) chisel and try to separate a magnet that way. If there is glue  
(and I'm pretty sure that there must be) a bit of inclined-plane action  
might have a chance of getting it off intact. I've got eight to play with  
so there isn't much to lose. These magnets see to be way more powerful than  
those I salvaged in the old days and I was toying with the idea of "the  
world's strongest refrigerator magnet" using two of the magnets in a  
sculpted exotic wood block and capable of holding up a ream of paper. Well,  
maybe not a ream but perhaps a cm?

Following up, chisel works

On 12/11/2015 3:04 PM, John McGaw wrote:
Quoted text here. Click to load it

I verified that a sharp chisel works a treat in removing the magnets from  
their plates but the adhesive that was used is so tenacious that it pulls  
the shiny plating from the back of the magnet and leaves it on the plate.  
This is probably not good for the long-term life of the magnets but I'll  
continue with it since I don't have any clue about a better way. The  
magnets are damned strong and when they stick to the chisel they are  
impossible to pull off and must be slid off the edge.

Site Timeline