Slow write speeds on a WD 500GB drive

[Yousuf Khan]

That does support the claim that the drive really is very slow
to write when its not just a benchmark that claims that.

I'd try the drive in another system myself in case its actually
a problem with the machine or the controller, not the drive.

Doing exactly that right now. I exchanged his old 500GB for a different
one. I took the old one and am now testing it on my own system in an
external eSATA case. I'm going to run a full SMART test on it, but I
don't know if it'll find anything, I think those SMART tests are just
read tests, aren't they?

Yousuf Khan

 

[Rod Speed]

Perhaps bad sectors are like cockroaches. When you see one, there may
be a lot more that you don't see.

Nope, not with the SMART data.

 

[Rod Speed]

Rod Speed wrote

Doing exactly that right now. I exchanged his old 500GB for
a different one. I took the old one and am now testing it on
my own system in an external eSATA case. I'm going to run
a full SMART test on it, but I don't know if it'll find anything,
I think those SMART tests are just read tests, aren't they?

Nar, they are proper tests, but don't test the write speed.

 

[Arno]

Perhaps bad sectors are like cockroaches. When you see one, there may
be a lot more that you don't see.

Depends on the cause, but yes, that can happen.

In any case, AFAICS, bad sectors, or heads with weak read elements,
would show up during reads more so than writes.

Not necessarily. The heads need to read in order to position and
positioning for writes needs to be done more carefully.

Arno

 

[Yousuf Khan]

Perhaps bad sectors are like cockroaches. When you see one, there may
be a lot more that you don't see.

In any case, AFAICS, bad sectors, or heads with weak read elements,
would show up during reads more so than writes.

- Franc Zabkar

Well, we'll see now, I've initiated HD Sentinel's surface write test. It
intially estimated that it would take 80 hours to complete. Now it's
finished exactly 1%, and it is now re-estimating 50 hours left. Can't
say I'm overwhelmed at the sudden burst of speed.

Windows performance counters are estimating it is writing at 3500 kB/s,
which is only 3X the speed of a DVD disk. One advantage would be that
unlike other disks which drop off in performance as they reach the end
of the disk, this one may maintain a steady speed from beginning to end,
so the disk time estimate may not change much from here on in.

Yousuf Khan

 

[Yousuf Khan]

A friend of mine had an WD My Book external case, which contained a
500GB WDC WD5000AAVS-00ZTB0 hard drive inside it. At some point along
the way the My Book's interface failed, it's warranty was expired, and
so we took its hard drive out and installed it directly inside his PC
case. This was back in mid- to late-2011, that it was transferred from
external to internal. It seemed to be fine for the most part in
performance, but recently we did some benchmarking and we were shocked
to find that although its reading speeds seem normal (around 50-60 MB/
s), its write speeds were shockingly low (around 10 MB/s, if even
that). What could cause such a weird asymmetry in read vs. write
speeds? It didn't seem that slow while it was inside the My Book case.

This is one of those WD Caviar Green drives, so I'm wondering if there
is some kind of power savings setting affecting it?

Yousuf Khan

Okay, some fascinating results have occurred on this drive, after I did
a full overwrite test on it, using HD Sentinel's surface tester utility.
The surface test found no additional bad sectors on it, but its
performance over the various sections of the drive were fascinating and
may have revealed their own truth. This is a graph of the performance
over various regions of the disk:

http://img403.imageshack.us/img403/8501/20120426wwdcwd5000aavs0.jpg

As you can see from the graph above, there were three regions on this
disk with completely different write characteristics. I'll call them
Slow, Medium, and Fast. It's probably more accurate to call the Medium
and Fast regions as "fast" and "super-fast" instead, because they are
both much faster than the slow region. Once the test entered each of the
regions, it looked like as if it had selected a higher gear. It took the
test 20 hours to write to the first half of the disk, and it only took 4
hours to complete the last half!

In the Slow region, the write speeds were very steady, but slow. The
average was between 3.5-4.5 MB/s (let's say 4.0 MB/s). This region
covers almost exactly the first half of the disk. This is the region
that took the first 20 hours of the write test to complete.

In the Medium region, I was surprised to find that the speeds just
climbed up like a rocket. However, you can't see it very well in the
overall graph here, but I was watching the instantaneous performance
graphs during this time, and I noticed a very distinctive and regular
saw-tooth pattern to the writes, it would go upto full speed (of 50-60
MB/s) for a few seconds, and then immediately drop back down to the same
4.0 MB/s of the Slow region. It would then rise back upto full speed
again, and it would repeat this pattern every 30 seconds or so. The full
speed sections would last for about 3 of the 30 seconds, while the slow
speed sections would last for the rest. So it was more like a
shark-tooth pattern really, rather than a saw-tooth pattern. This region
occupied the disk from the 50% mark to approximately the 70% mark of the
disk.

In the Fast region, watching the instantaneous performance graph, you
would see that it switched from the saw-tooth to a very steady
high-speed. Between the Medium and Fast regions, the remainder of the
write test only took 4 hours!

So, then I set up three partitions to coincide with these three regions,
and I ran the ATTO disk benchmarks in each of the partitions.

Slow: http://imageshack.us/photo/my-images/840/captureattowd500gbaavsp.jpg/

Medium:
http://imageshack.us/photo/my-images/850/captureattowd500gbaavsp.jpg/

Fast: http://imageshack.us/photo/my-images/851/captureattowd500gbaavsp.jpg/

They basically confirm what I saw during the write testing, but you
don't get a sense of the saw-tooth pattern in the medium test.

So what do you think this is a symptom of? I'm thinking that perhaps
there is a problem with the write-head of one of the platters, but not
on the others?

Yousuf Khan

 

[Rod Speed]

bbbl67 wrote [which was just you too

A friend of mine had an WD My Book external case, which contained a
500GB WDC WD5000AAVS-00ZTB0 hard drive inside it. At some point along the
way the My Book's interface failed, it's warranty was expired, and so we
took its hard drive out and installed it directly inside his PC
case. This was back in mid- to late-2011, that it was transferred from
external to internal. It seemed to be fine for the most part in
performance, but recently we did some benchmarking and we were shocked to
find that although its reading speeds seem normal (around 50-60 MB/s),
its write speeds were shockingly low (around 10 MB/s, if even that). What
could cause such a weird asymmetry in read vs. write
speeds? It didn't seem that slow while it was inside the My Book case.
This is one of those WD Caviar Green drives, so I'm wondering if there is
some kind of power savings setting affecting it?

Okay, some fascinating results have occurred on this drive, after I did a
full overwrite test on it, using HD Sentinel's surface tester utility. The
surface test found no additional bad sectors on it, but its performance
over the various sections of the drive were fascinating and may have
revealed their own truth. This is a graph of the performance over various
regions of the disk:

http://img403.imageshack.us/img403/8501/20120426wwdcwd5000aavs0.jpg

As you can see from the graph above, there were three regions on this disk
with completely different write characteristics. I'll call them Slow,
Medium, and Fast. It's probably more accurate to call the Medium and Fast
regions as "fast" and "super-fast" instead, because they are both much
faster than the slow region. Once the test entered each of the regions, it
looked like as if it had selected a higher gear. It took the test 20 hours
to write to the first half of the disk, and it only took 4 hours to
complete the last half!

In the Slow region, the write speeds were very steady, but slow. The
average was between 3.5-4.5 MB/s (let's say 4.0 MB/s). This region covers
almost exactly the first half of the disk. This is the region that took
the first 20 hours of the write test to complete.

In the Medium region, I was surprised to find that the speeds just climbed
up like a rocket. However, you can't see it very well in the overall graph
here, but I was watching the instantaneous performance graphs during this
time, and I noticed a very distinctive and regular saw-tooth pattern to
the writes, it would go upto full speed (of 50-60 MB/s) for a few seconds,
and then immediately drop back down to the same 4.0 MB/s of the Slow
region. It would then rise back upto full speed again, and it would repeat
this pattern every 30 seconds or so. The full speed sections would last
for about 3 of the 30 seconds, while the slow speed sections would last
for the rest. So it was more like a shark-tooth pattern really, rather
than a saw-tooth pattern. This region occupied the disk from the 50% mark
to approximately the 70% mark of the disk.

In the Fast region, watching the instantaneous performance graph, you
would see that it switched from the saw-tooth to a very steady high-speed.
Between the Medium and Fast regions, the remainder of the write test only
took 4 hours!

So, then I set up three partitions to coincide with these three regions,
and I ran the ATTO disk benchmarks in each of the partitions.

Slow:
http://imageshack.us/photo/my-images/840/captureattowd500gbaavsp.jpg/

Medium:
http://imageshack.us/photo/my-images/850/captureattowd500gbaavsp.jpg/

Fast:
http://imageshack.us/photo/my-images/851/captureattowd500gbaavsp.jpg/

They basically confirm what I saw during the write testing, but you don't
get a sense of the saw-tooth pattern in the medium test.

So what do you think this is a symptom of?

A ****ed drive, silly |-(

I'm thinking that perhaps there is a problem with the write-head of one of
the platters, but not on the others?

Dunno, I would have expected all the partitions to involve
all the heads, particularly with a drive of that size which was
before things got exotic with larger than 512 byte sectors used.

And while WD doesn't say how many platters it is, its almost
certainly just one given the size of it, and its likely just using
one surface and one head too given that its the second
smallest drive in the series. Presumably the other one is so
bad that it doesn't get used at all. Although the problem
with that theory is that it would be surprising if they get
enough drives with just one surface usable from the factory.

And it doesn't explain why only a quarter of the drive is viable.

I guess the most likely explanation is that it got dropped and
that's what ****ed it but I spose it might be ****ed by design
in an early attempt at a green design and that's the problem.

Odd that no one else has seen it if its that last tho.

How likely is it that its always had a completely ****ed write performance ?

Is he that unlikely to have noticed ?

 

[Rod Speed]

bbbl67 wrote [which was just you too

A friend of mine had an WD My Book external case, which contained a
500GB WDC WD5000AAVS-00ZTB0 hard drive inside it. At some point along
the way the My Book's interface failed, it's warranty was expired, and
so we took its hard drive out and installed it directly inside his PC
case. This was back in mid- to late-2011, that it was transferred from
external to internal. It seemed to be fine for the most part in
performance, but recently we did some benchmarking and we were shocked
to find that although its reading speeds seem normal (around 50-60
MB/s), its write speeds were shockingly low (around 10 MB/s, if even
that). What could cause such a weird asymmetry in read vs. write
speeds? It didn't seem that slow while it was inside the My Book case.
This is one of those WD Caviar Green drives, so I'm wondering if there
is some kind of power savings setting affecting it?

Okay, some fascinating results have occurred on this drive, after I did a
full overwrite test on it, using HD Sentinel's surface tester utility.
The surface test found no additional bad sectors on it, but its
performance over the various sections of the drive were fascinating and
may have revealed their own truth. This is a graph of the performance
over various regions of the disk:
http://img403.imageshack.us/img403/8501/20120426wwdcwd5000aavs0.jpg
As you can see from the graph above, there were three regions on this
disk with completely different write characteristics. I'll call them
Slow, Medium, and Fast. It's probably more accurate to call the Medium
and Fast regions as "fast" and "super-fast" instead, because they are
both much faster than the slow region. Once the test entered each of the
regions, it looked like as if it had selected a higher gear. It took the
test 20 hours to write to the first half of the disk, and it only took 4
hours to complete the last half!
In the Slow region, the write speeds were very steady, but slow. The
average was between 3.5-4.5 MB/s (let's say 4.0 MB/s). This region covers
almost exactly the first half of the disk. This is the region that took
the first 20 hours of the write test to complete.
In the Medium region, I was surprised to find that the speeds just
climbed up like a rocket. However, you can't see it very well in the
overall graph here, but I was watching the instantaneous performance
graphs during this time, and I noticed a very distinctive and regular
saw-tooth pattern to the writes, it would go upto full speed (of 50-60
MB/s) for a few seconds, and then immediately drop back down to the same
4.0 MB/s of the Slow region. It would then rise back upto full speed
again, and it would repeat this pattern every 30 seconds or so. The full
speed sections would last for about 3 of the 30 seconds, while the slow
speed sections would last for the rest. So it was more like a shark-tooth
pattern really, rather than a saw-tooth pattern. This region occupied the
disk from the 50% mark to approximately the 70% mark of the disk.
In the Fast region, watching the instantaneous performance graph, you
would see that it switched from the saw-tooth to a very steady
high-speed. Between the Medium and Fast regions, the remainder of the
write test only took 4 hours!
So, then I set up three partitions to coincide with these three regions,
and I ran the ATTO disk benchmarks in each of the partitions.
Slow:
http://imageshack.us/photo/my-images/840/captureattowd500gbaavsp.jpg/
Medium:
http://imageshack.us/photo/my-images/850/captureattowd500gbaavsp.jpg/
Fast:
http://imageshack.us/photo/my-images/851/captureattowd500gbaavsp.jpg/
They basically confirm what I saw during the write testing, but you don't
get a sense of the saw-tooth pattern in the medium test.
So what do you think this is a symptom of?

A ****ed drive, silly |-(

Dunno, I would have expected all the partitions to involve
all the heads, particularly with a drive of that size which was
before things got exotic with larger than 512 byte sectors used.

And while WD doesn't say how many platters it is, its almost
certainly just one given the size of it, and its likely just using
one surface and one head too given that its the second
smallest drive in the series. Presumably the other one is so
bad that it doesn't get used at all. Although the problem
with that theory is that it would be surprising if they get
enough drives with just one surface usable from the factory.

And it doesn't explain why only a quarter of the drive is viable.

I guess the most likely explanation is that it got dropped and
that's what ****ed it but I spose it might be ****ed by design
in an early attempt at a green design and that's the problem.

Odd that no one else has seen it if its that last tho.

How likely is it that its always had a completely ****ed write performance
?

Is he that unlikely to have noticed ?

I guess one possibility is that whatever ****ed the interface,
presumably a power supply failure, ****ed the drive electronics
too. But that doesn't explain that weird 1st graph. You wouldn't
expect that ****ed electronics would produce that result, that
has to be a mechanical problem of some kind.

That as the era when a mate of mine had 20 WD drives in a row all die.

Dunno what the death symptom was tho.

 

[Yousuf Khan]

Dunno, I would have expected all the partitions to involve
all the heads, particularly with a drive of that size which was
before things got exotic with larger than 512 byte sectors used.

And while WD doesn't say how many platters it is, its almost
certainly just one given the size of it, and its likely just using
one surface and one head too given that its the second
smallest drive in the series. Presumably the other one is so
bad that it doesn't get used at all. Although the problem
with that theory is that it would be surprising if they get
enough drives with just one surface usable from the factory.

And it doesn't explain why only a quarter of the drive is viable.

Hard to say without WD telling us that information in its specs.

I guess the most likely explanation is that it got dropped and
that's what ****ed it but I spose it might be ****ed by design
in an early attempt at a green design and that's the problem.

If it got dropped, then there would be other drives in the same system
case that would have similar problems, since this thing was an internal
drive for the last little bit.

Odd that no one else has seen it if its that last tho.

How likely is it that its always had a completely ****ed write
performance ?

Is he that unlikely to have noticed ?

When it was in the external enclosure, it was being used as an extension
drive for a PVR. So it wasn't even holding PC data at the time, and it
was formatted to a proprietary file system. It was only put into a PC
for the first time when it was put into this case. Since that time it
was reformatted to NTFS, and it was holding several hundred GB's of
data: so it had to have been filled with data in a reasonable amount of
time. As far as I recall, everything about the drive was normal during
the time it was put into the PC. It was running fine for months, but the
current problem was only noticed when I went to help my friend
reorganize his data, and we were going to make this drive part of a
Windows Disk Management spanned volume. We had moved data off of it,
which had operated normally; but when we tried to move data back onto
it, and we noticed this.

Yousuf Khan

 

[Rod Speed]

Rod Speed wrote

Hard to say without WD telling us that information in its specs.

The read data for the full drive, your first chart, does indicate
that it doesn't do anything unusual as far as heads is concerned,
we do see the usual progression across the drive and no evidence
that it switches heads say half way.

If it got dropped, then there would be other drives in the same system
case that would have similar problems, since this thing was an internal
drive for the last little bit.

But not initially. Maybe it got dropped in the external
case and he didn't notice the lousy write performance
or the lousy write performance showed up later etc well
after the drop happened but was still due to the drop.

When it was in the external enclosure, it was being used as an extension
drive for a PVR. So it wasn't even holding PC data at the time, and it was
formatted to a proprietary file system. It was only put into a PC for the
first time when it was put into this case. Since that time it was
reformatted to NTFS, and it was holding several hundred GB's of data: so
it had to have been filled with data in a reasonable amount of time.

Guess he might have started that off and then gone to bed etc.

 

[Yousuf Khan]

I guess one possibility is that whatever ****ed the interface,
presumably a power supply failure, ****ed the drive electronics
too. But that doesn't explain that weird 1st graph. You wouldn't
expect that ****ed electronics would produce that result, that
has to be a mechanical problem of some kind.

I don't know what ****ed up the original external enclosure electronics.
It just stopped sending data at some point, and therefore became useless
as a PVR storage device. I doubt that what happened to the internal
drive and the external interface are related as they happened almost a
year apart.

This particular enclosure had all three major interfaces: USB, Firewire,
and eSATA. It was mainly being used through USB on the PVR, and when
that stopped working, we tried testing it on eSATA on the PC, and that
one was also not working (they were all previously working). Didn't
bother with Firewire, since two out of three interfaces were dead,
chances are slim that the 3rd was still working.

I talked to HD Sentinel support about this, and he suggested that the
problem might lie in the total Load/Unload Cycle Count, which is over
600,000. I think this drive is only rated for 300,000. He suggested I
disable spin-down of these drives.

Yousuf Khan

Here's the SMART data:

1,Raw Read Error Rate,51,200,200,OK,0,0,Enabled
3,Spin Up Time,21,201,52,OK,2950,0,Enabled
4,Start/Stop Count,0,96,96,OK (Always passing),4738,0,Enabled
5,Reallocated Sectors Count,140,200,200,OK,0,0,Enabled
7,Seek Error Rate,51,200,200,OK,0,0,Enabled
9,Power On Time Count,0,62,62,OK (Always passing),28244,0,Enabled
10,Spin Retry Count,51,100,100,OK,0,0,Enabled
11,Drive Calibration Retry Count,51,100,100,OK,0,0,Enabled
12,Drive Power Cycle Count,0,97,97,OK (Always passing),3702,0,Enabled
192,Power off Retract Cycle Count,0,196,196,OK (Always
passing),3590,0,Enabled
193,Load/Unload Cycle Count,0,1,1,OK (Always passing),620508,0,Enabled
194,Disk Temperature,0,123,79,OK (Always passing),24,0,Enabled
196,Reallocation Event Count,0,200,200,OK (Always passing),0,0,Enabled
197,Current Pending Sector Count,0,200,200,OK (Always passing),0,0,Enabled
198,Off-Line Uncorrectable Sector Count,0,200,200,OK (Always
passing),1,0,Enabled
199,Ultra ATA CRC Error Count,0,200,200,OK (Always passing),1,0,Enabled
200,Write Error Rate,51,200,199,OK,0,0,Enabled

 

[Yousuf Khan]

Guess he might have started that off and then gone to bed etc.

Well, as I said, it took 20 hours just to write to the first half of
this disk during the write test! No matter if you set it to start
copying and then go off to bed, you'll definitely notice this amount of
slow performance. So it is likely this problem didn't exist, when he
originally copied data into this drive.

Also since we see that the problem doesn't exist on the last third of
the disk, if this drive was more than 70% full, then it's likely he
won't have even noticed a problem at all, nor when it originally
started. Everything would appear normal on that section of the disk.
Also since read performance was completely normal, so earlier data that
doesn't get changed very much would also appear safe.

Yousuf Khan

 

[Arno]

Okay, some fascinating results have occurred on this drive, after I did
a full overwrite test on it, using HD Sentinel's surface tester utility.
The surface test found no additional bad sectors on it, but its
performance over the various sections of the drive were fascinating and
may have revealed their own truth. This is a graph of the performance
over various regions of the disk:

http://img403.imageshack.us/img403/8501/20120426wwdcwd5000aavs0.jpg

Very nice. If you take into acount that disks seek from outer
edhe to center, and that virbration/positioning is more problematic
on the outside, I think my suspicion of seek-errors for
the more sensitive write-seeks is right on.

Arno

 

[Arno]

I don't know what ****ed up the original external enclosure electronics.
It just stopped sending data at some point, and therefore became useless
as a PVR storage device. I doubt that what happened to the internal
drive and the external interface are related as they happened almost a
year apart.

This particular enclosure had all three major interfaces: USB, Firewire,
and eSATA. It was mainly being used through USB on the PVR, and when
that stopped working, we tried testing it on eSATA on the PC, and that
one was also not working (they were all previously working). Didn't
bother with Firewire, since two out of three interfaces were dead,
chances are slim that the 3rd was still working.

I talked to HD Sentinel support about this, and he suggested that the
problem might lie in the total Load/Unload Cycle Count, which is over
600,000. I think this drive is only rated for 300,000. He suggested I
disable spin-down of these drives.

I agree there. Might even be as low as 50'000 for a 3.5" drive.
I had some disks trying to do the same to themselves and
caught them in time. One was up yo 800'000 load cycles,
but these are 2.5" drives that can (roughly) withstand
10 times as many load cycles.

Well, seems this is another suicide via power-management.
Searching for "wdidle3" should bring up the last few
threads dealing with this, although I think you posted in
some of them.

As to the drive, while it does not seem dead, it seems
basically unuasable.

Arno

 

[Yousuf Khan]

I agree there. Might even be as low as 50'000 for a 3.5" drive.
I had some disks trying to do the same to themselves and
caught them in time. One was up yo 800'000 load cycles,
but these are 2.5" drives that can (roughly) withstand
10 times as many load cycles.

After learning about this, I looked at one my own drives and found that
it was already upto 200,000 cycles out of the 300,000 recommendation!
I've now enabled the anti-power management option to stop the
deterioration right there in its tracks.

I have no idea how this drive got upto to so many unload cycles in such
a short amount of time, except that this drive used to be inside an
external USB enclosure too before I bought a bigger system case and was
able to bring it inside. I'm thinking that the USB API might enable some
power management by default, without even asking you.

Some drive manufacturers don't even monitor the load/unload cycles in
SMART. Either they think their drives are immune to it, or there is a
false sense of security.

Well, seems this is another suicide via power-management.
Searching for "wdidle3" should bring up the last few
threads dealing with this, although I think you posted in
some of them.

If I did participate in them, I don't remember them anymore.

As to the drive, while it does not seem dead, it seems
basically unuasable.

Well, I think this is one of those very rare types of failures of
hardware that we all dream about happening to us but rarely ever
happens: the hardware failure where the drive is telling you that it's
about to die, and still gives you the chance to recover your old data
before it goes. Overall, I'd love to have had a warning like this before
I drive goes.

Yousuf Khan

 

[Rod Speed]

Rod Speed wrote

I don't know what ****ed up the original external enclosure electronics.
It just stopped sending data at some point, and therefore became useless
as a PVR storage device.

But you don't see too many relatively simple devices like
that suffer two completely different very serious faults.

I doubt that what happened to the internal drive and the external
interface are related as they happened almost a year apart.

You can get delayed results like that, particularly when the power
supply seriously over voltages the system for a short time.

Harder to explain the weird symptom that the write speed
obviously varys with where the head is on the platter that
way tho and that only affecting writes, not reads.

If we ever do work out what the fault is due to, bet we
will kick ourselves for not thinking of that earlier tho.

This particular enclosure had all three major interfaces: USB, Firewire,
and eSATA. It was mainly being used through USB on the PVR, and when that
stopped working, we tried testing it on eSATA on the PC, and that one was
also not working (they were all previously working). Didn't bother with
Firewire, since two out of three interfaces were dead, chances are slim
that the 3rd was still working.

I talked to HD Sentinel support about this, and he suggested that the
problem might lie in the total Load/Unload Cycle Count, which is over
600,000. I think this drive is only rated for 300,000.

Cant see how that would produce that very unusual symptom
of the write speed so dramatically affected by the location of
the head on the platter, and not the read speed.

He suggested I disable spin-down of these drives.

They certainly are notorious for unloading much too frequently.

Havent see anyone report that weird write speed result with them tho.

Here's the SMART data:

Whats that from, HD Sentinel presumably ?

1,Raw Read Error Rate,51,200,200,OK,0,0,Enabled
3,Spin Up Time,21,201,52,OK,2950,0,Enabled
4,Start/Stop Count,0,96,96,OK (Always passing),4738,0,Enabled
5,Reallocated Sectors Count,140,200,200,OK,0,0,Enabled
7,Seek Error Rate,51,200,200,OK,0,0,Enabled
9,Power On Time Count,0,62,62,OK (Always passing),28244,0,Enabled
10,Spin Retry Count,51,100,100,OK,0,0,Enabled
11,Drive Calibration Retry Count,51,100,100,OK,0,0,Enabled
12,Drive Power Cycle Count,0,97,97,OK (Always passing),3702,0,Enabled
192,Power off Retract Cycle Count,0,196,196,OK (Always
passing),3590,0,Enabled
193,Load/Unload Cycle Count,0,1,1,OK (Always passing),620508,0,Enabled
194,Disk Temperature,0,123,79,OK (Always passing),24,0,Enabled
196,Reallocation Event Count,0,200,200,OK (Always passing),0,0,Enabled
197,Current Pending Sector Count,0,200,200,OK (Always passing),0,0,Enabled
198,Off-Line Uncorrectable Sector Count,0,200,200,OK (Always
passing),1,0,Enabled
199,Ultra ATA CRC Error Count,0,200,200,OK (Always passing),1,0,Enabled
200,Write Error Rate,51,200,199,OK,0,0,Enabled

Looks fine except for the load/unload count.

 

[Rod Speed]

Very nice. If you take into acount that disks seek from outer
edhe to center, and that virbration/positioning is more
problematic on the outside, I think my suspicion of seek-
errors for the more sensitive write-seeks is right on.

Doesn’t explain the sudden transition from ultra slow to normal tho.

And why it doesn’t affect the reads either.

 

[Rod Speed]

Arno wrote

After learning about this, I looked at one my own drives and found that it
was already upto 200,000 cycles out of the 300,000 recommendation! I've
now enabled the anti-power management option to stop the deterioration
right there in its tracks.

I have no idea how this drive got upto to so many unload cycles in such a
short amount of time,

Those WD drives are utterly notorious for an insane unload timeout.

except that this drive used to be inside an external USB enclosure too
before I bought a bigger system case and was able to bring it inside. I'm
thinking that the USB API might enable some power management by default,
without even asking you.

Or it might be another WD drive.

Some drive manufacturers don't even monitor the load/unload cycles in
SMART. Either they think their drives are immune to it,

More likely they just don't distinguish all the similar
events and lump some of them together more.

or there is a false sense of security.
Unlikely.

If I did participate in them, I don't remember them anymore.

Don't recall you did, but they did get quite a bit of traffic.

Well, I think this is one of those very rare types of failures of hardware
that we all dream about happening to us but rarely ever happens: the
hardware failure where the drive is telling you that it's about to die,
and still gives you the chance to recover your old data before it goes.
Overall, I'd love to have had a warning like this before I drive goes.

Plenty do. Just had one Toshiba in a Toshiba laptop that did that.

And did it with a very naïve user so that user was sure there was a problem
too.

If even advised him to backup, and he did that, and it
had just 2 weeks to go till the warranty ran out too.

 

[Rod Speed]

http://www.google.com.au/search?q=wd+very+slow+writes
produces some very interesting hits.

Looks like WD might well have a problem.

 

[Yousuf Khan]

More likely they just don't distinguish all the similar
events and lump some of them together more.



Don't recall you did, but they did get quite a bit of traffic.

Well, I found some of the old threads here (Google Groups):

http://is.gd/CpVL7N
http://is.gd/Qre36O
http://is.gd/a0DUvU

I do recall seeing some of these old threads with passing interest, but
at that time it did not interest me as I wasn't aware that they affected
me too. I can now see that they may have been the very same issues I was
dealing with myself. HD Sentinel actually implemented a resident program
just to keep the hard disks from being put to rest, so it must be a
well-known problem.

Yousuf Khan