Regular hard drive clunking noise mystery

[Alan Thorne]

I have bought a Buffalo TeraStation Home Server HS-2.0TGL/R5. It is a
network attached device with 4 internal hard disks. I have 2 earlier
1Tb units which run virtually silently but the new 2Tb model produces a
clunking sound regularly every 3-4 seconds, the volume of which is
several times that of disk read/write activity. I have had the unit
changed twice, and all three produce the same problem sound.

The noise is like the disk heads hitting the end-stops, i.e. resetting,
and I am concerned that the mechanical shock will cause the drive to
fail.

Buffalo technical support maintain that there is no fault, since no
error messages are produced, and that I am being over-sensitive to a
little noise.

Please can someone help: is there any good reason why a HDD would clunk
every few seconds?

The mystery deepens when I investigated further:

- the clunking only starts after the unit has booted up (which takes
about 30s)

- the clunking continues, even when data is being written

- the clunking is produced at the same 3-4 second interval no matter
what combination of HDD drives are in the Buffalo unit, i.e. one HDD
clunks at the same periodic interval as all 4, only quieter, and if one
HDD is removed the remaining 3 also clunk at the same 3-4s rate AFTER
the device has booted up

Any ideas or help would be greatly appreciated

Thanks

 

[David Flory]

I have bought a Buffalo TeraStation Home Server HS-2.0TGL/R5. It is a
network attached device with 4 internal hard disks. I have 2 earlier
1Tb units which run virtually silently but the new 2Tb model produces a
clunking sound regularly every 3-4 seconds, the volume of which is
several times that of disk read/write activity. I have had the unit
changed twice, and all three produce the same problem sound.

The noise is like the disk heads hitting the end-stops, i.e. resetting,
and I am concerned that the mechanical shock will cause the drive to
fail.

Buffalo technical support maintain that there is no fault, since no
error messages are produced, and that I am being over-sensitive to a
little noise.

Please can someone help: is there any good reason why a HDD would clunk
every few seconds?

The mystery deepens when I investigated further:

- the clunking only starts after the unit has booted up (which takes
about 30s)

- the clunking continues, even when data is being written

- the clunking is produced at the same 3-4 second interval no matter
what combination of HDD drives are in the Buffalo unit, i.e. one HDD
clunks at the same periodic interval as all 4, only quieter, and if one
HDD is removed the remaining 3 also clunk at the same 3-4s rate AFTER
the device has booted up

Any ideas or help would be greatly appreciated

Thanks

Hi,

I'd try to get a refund, and get a different brand. Clunking in hard
drives is very bad. It usually is the sound of the heads parking
suddenly, which can have different causes, but none of them are very good.

In your case, it may be a probably a firmware bug in the new high
capacity drives. A firmware update might solve the problem, if one is
available.

If it's too late for a refund, maybe you can get Buffalo to exchange for
the same capacity in 1TB units. I think you may be getting an
early-adopter bug because the 500GB drives are such new technology.

Best of luck,

Dave

 

[Arno Wagner]

I'd try to get a refund, and get a different brand. Clunking in hard
drives is very bad. It usually is the sound of the heads parking
suddenly, which can have different causes, but none of them are very good.

I agree. Or at least very suspicuous.

Arno

In your case, it may be a probably a firmware bug in the new high
capacity drives. A firmware update might solve the problem, if one is
available.

If it's too late for a refund, maybe you can get Buffalo to exchange for
the same capacity in 1TB units. I think you may be getting an
early-adopter bug because the 500GB drives are such new technology.

 

[Folkert Rienstra]

which can have different causes,

Some are by design. Silly, but normal.

but none of them are very good.

I agree.

But then you were always clueless. Some drives are just designed like that.

Or at least very suspicuous.

If it was one drive doing it, yes. Guess what.

Arno

They are not.

 

[Alan Thorne]

Some are by design. Silly, but normal.



But then you were always clueless. Some drives are just designed like that.


If it was one drive doing it, yes. Guess what.


They are not.

No only are all the drives doing it, but they are doing it in sync,
even after the NAS unit has been running for over an hour. Seems
improbable that 4 drives would be identical to such a degree without
quartz timing...

The other indicator that it is the NAS system rather than the drives
themselves is the delay before the clunking starts. HDDs spin up in a
few seconds? Yet it is over half a minute before the synchronised
clunking starts.

What do you think?

 

[Arno Wagner]

No only are all the drives doing it, but they are doing it in sync,
even after the NAS unit has been running for over an hour. Seems
improbable that 4 drives would be identical to such a degree without
quartz timing...

The drives do have quarz timing usually. You can expect around 50ppm
accuracy for an unadjusted quarz. That would give you a deviation of
up to 3.6 sec per hour. So if they are perfectly (<.5) sec
in sync after an hour, I would say this is caused by some external
effect.

The other indicator that it is the NAS system rather than the drives
themselves is the delay before the clunking starts. HDDs spin up in a
few seconds? Yet it is over half a minute before the synchronised
clunking starts.

What do you think?

You can make a cross-check: Disconnect the data line of one or
all disks. If they still clunk, it is the disks. If not, then it
should be the NAS. Peronslly I would expect they did somethign
supid in the NAS, like sending a device reset regularly,
because their software is unreliable...


Arno

 

[Alan Thorne]

The drives do have quarz timing usually. You can expect around 50ppm
accuracy for an unadjusted quarz. That would give you a deviation of
up to 3.6 sec per hour. So if they are perfectly (<.5) sec
in sync after an hour, I would say this is caused by some external
effect.



You can make a cross-check: Disconnect the data line of one or
all disks. If they still clunk, it is the disks. If not, then it
should be the NAS. Peronslly I would expect they did somethign
supid in the NAS, like sending a device reset regularly,
because their software is unreliable...

Great idea Arno. With all data lines disconnected there is no clunking!
Hard drives spin up to speed in 15s when disk activity starts, when
data lines are attached. First clunk occurs after 62s amongst irregular
disk activity, part of the NAS boot-up process presumably. Regular
clunking begins after 1 minute 59 seconds. Period between regular
clunks thereafter is exactly 5 seconds.

Thanks again for the idea. I will feed back the findings to Buffalo
tech support.

 

[Arno Wagner]

Arno Wagner wrote:

Great idea Arno. With all data lines disconnected there is no clunking!
Hard drives spin up to speed in 15s when disk activity starts, when
data lines are attached. First clunk occurs after 62s amongst irregular
disk activity, part of the NAS boot-up process presumably. Regular
clunking begins after 1 minute 59 seconds. Period between regular
clunks thereafter is exactly 5 seconds.

Thanks again for the idea. I will feed back the findings to Buffalo
tech support.

Glad I could help!

Arno

 

[Folkert Rienstra]

Not only are all the drives doing it, but they are doing it in sync,
even after the NAS unit has been running for over an hour. Seems
improbable that 4 drives would be identical to such a degree without
quartz timing...

The other indicator that it is the NAS system rather than the drives
themselves is the delay before the clunking starts. HDDs spin up in a
few seconds? Yet it is over half a minute before the synchronised
clunking starts.

What do you think?

That your hunch is probably right. Should be easy to check, no?

 

[Folkert Rienstra]

Great idea Arno. With all data lines disconnected there is no clunking!

Hard drives spin up to speed in 15s when disk activity starts, when
data lines are attached.

That sentence does not make (immediate) sense, if at all.
And obviously there will be no clunking when the drives aren't spinning.

First clunk occurs after 62s amongst irregular disk activity, part of
the NAS boot-up process presumably.

Regular clunking begins after 1 minute 59 seconds.

Would that be after a period of inactivity?
That would make some sense if the software idles the drives, parking
their heads.

Period between regular clunks thereafter is exactly 5 seconds.

That doesn't. It sounds as if someone ****ed up in the programming.

 

[Alan Thorne]

That sentence does not make (immediate) sense, if at all.
And obviously there will be no clunking when the drives aren't spinning.



Would that be after a period of inactivity?
That would make some sense if the software idles the drives, parking
their heads.


That doesn't. It sounds as if someone ****ed up in the programming.

Apologies for confusion. From 15 seconds up until the 1 minute 59
seconds point there are irregular noises from the drives indicating
disk activity, including some clunking. Some clunks are only a couple
of seconds apart. Thereafter, the clunking becomes regular, every 5
seconds, after the start-up disk activity has ceased. The 5 second
metronome clunking continues even while writing a lot of data (my
initial data take-on which I aborted).

I go with the programming error theory. Buffalo haven't responded to
any of my emails since last Thursday, including my feedback of the
results of Arno's idea first thing this morning.

 

[Folkert Rienstra]

Apologies for confusion.

Which still remains on the part of the drives not clunking when
not spinning undermining the conclusion in the preceding sentence.
That's what the confusion is about and in need of clearing up.

From 15 seconds up until the 1 minute 59 seconds point there are irregu-
lar noises from the drives indicating disk activity, including some clunking.

Some clunks are only a couple of seconds apart.

A drive may also clunk when it accesses track zero, when it seeks
to track zero and overshoots it, running in to the end stops.
So a few clunks now and then aren't necessarily conspicuous.
Of course this only happens in drives that still park their heads
on the platters. These may also clunk when parking their heads.
Drives with off platter parking obviously behave differently.

Thereafter, the clunking becomes regular, every 5 seconds, after the start-
up disk activity has ceased. The 5 second metronome clunking continues even
while writing a lot of data (my initial data take-on which I aborted).

I go with the programming error theory. Buffalo haven't responded to
any of my emails since last Thursday, including my feedback of

the results of Arno's idea first thing this morning.

Which were obviously inconclusive with the drives not spinning.

 

[Alan Thorne]

Which still remains on the part of the drives not clunking when
not spinning undermining the conclusion in the preceding sentence.
That's what the confusion is about and in need of clearing up.



A drive may also clunk when it accesses track zero, when it seeks
to track zero and overshoots it, running in to the end stops.
So a few clunks now and then aren't necessarily conspicuous.
Of course this only happens in drives that still park their heads
on the platters. These may also clunk when parking their heads.
Drives with off platter parking obviously behave differently.



Which were obviously inconclusive with the drives not spinning.

Your English is confusing me a little! To clarify: with the disks
spinning but not connected to the NAS unit they are silent. When
connected, they make the clunking noises in the sequence I described
above.

Thanks for the mechanical explanation for clunking. I would not be
concerned by a few clunks. I am concerned by the number: every 5
seconds means 24*60*60/5 = 17,280 impacts every day.

This seems excessive and I would be concerned for the safety of my data
since the unit is used for long-term storage of data that I don't have
room for on my main machine. It is the backup, so if it fails I loose
the data permanently.

 

[Folkert Rienstra]

Folkert Rienstra wrote:
Arno Wagner wrote:
[snip]
Hard drives spin up to speed in 15s when disk activity starts, when
data lines are attached.

That sentence does not make (immediate) sense, if at all.
And obviously there will be no clunking when the drives aren't spinning.

First clunk occurs after 62s amongst irregular disk activity, part of
the NAS boot-up process presumably.

Regular clunking begins after 1 minute 59 seconds.

Would that be after a period of inactivity?
That would make some sense if the software idles the drives, parking
their heads.

Period between regular clunks thereafter is exactly 5 seconds.

That doesn't. It sounds as if someone ****ed up in the programming.

Apologies for confusion.

Which still remains on the part of the drives not clunking when
not spinning undermining the conclusion in the preceding sentence.
That's what the confusion is about and in need of clearing up.

From 15 seconds up until the 1 minute 59 seconds point there are irregu-
lar noises from the drives indicating disk activity, including some clunking.

Some clunks are only a couple of seconds apart.

A drive may also clunk when it accesses track zero, when it seeks
to track zero and overshoots it, running in to the end stops.
So a few clunks now and then aren't necessarily conspicuous.

Oops, meant suspicious of course.

Your English is confusing me a little!

No really? *My* english huh, but
" Hard drives spin up to speed in 15s when disk
activity starts, when data lines are attached"
makes perfect sense to you.

Please explain to us how drives are spinning up when they are already spinning.

To clarify: with the disks spinning but not connected to the NAS unit
they are silent.

When connected, they make the clunking noises in the sequence I described
above.

Right, so how difficult was it to say that in the first place when I first raised
the question.

Thanks for the mechanical explanation for clunking. I would not be
concerned by a few clunks. I am concerned by the number: every 5
seconds means 24*60*60/5 = 17,280 impacts every day.

This seems excessive and I would be concerned for the safety of my data
since the unit is used for long-term storage of data that I don't have
room for on my main machine.

It is the backup,

No, it isn't.

so if it fails I loose the data permanently.

Then it's not a backup, now is it, if there isn't an original copy of it
somewhere. And what backup is online all the time. Switch it off.

 

[Arno Wagner]

Folkert Rienstra wrote:

Folkert Rienstra wrote: [...]
the results of Arno's idea first thing this morning.

Which were obviously inconclusive with the drives not spinning.

Your English is confusing me a little! To clarify: with the disks
spinning but not connected to the NAS unit they are silent.

I think Folkert just does not know that current (S)ATA drives
do auto-spin as soon as power is applied ....

When
connected, they make the clunking noises in the sequence I described
above.

Thanks for the mechanical explanation for clunking. I would not be
concerned by a few clunks. I am concerned by the number: every 5
seconds means 24*60*60/5 = 17,280 impacts every day.

This seems excessive and I would be concerned for the safety of my data
since the unit is used for long-term storage of data that I don't have
room for on my main machine. It is the backup, so if it fails I loose
the data permanently.

While I do not have any numbers, I would be concerned too. The
recalibration is supposed to be an operation not done too often.
I would nto be surprised if after, say, 1'000'000 of these
the drive mechanics start to break down...

Arno

 

[Folkert Rienstra]

Folkert Rienstra wrote: [...]
the results of Arno's idea first thing this morning.

Which were obviously inconclusive with the drives not spinning.

Your English is confusing me a little! To clarify: with the disks
spinning but not connected to the NAS unit they are silent.

I think Folkert just does not know that current (S)ATA drives
do auto-spin as soon as power is applied ....

You know what I think, babblebot?
That terminally stupid babblebots like yourself do not know that
current drives can be set to not spin-up "as soon as power is applied".

Take a nap, babblebot. Re-grow a few braincells.

While I do not have any numbers, I would be concerned too.

The recalibration is supposed to be an operation not done too often.

Stupid babblebot, as always. You can't force recalibrations.

 

[Alan Thorne]

Folkert Rienstra wrote:
Arno Wagner wrote:
[snip]
Hard drives spin up to speed in 15s when disk activity starts, when
data lines are attached.

That sentence does not make (immediate) sense, if at all.
And obviously there will be no clunking when the drives aren't spinning.

First clunk occurs after 62s amongst irregular disk activity, part of
the NAS boot-up process presumably.

Regular clunking begins after 1 minute 59 seconds.

Would that be after a period of inactivity?
That would make some sense if the software idles the drives, parking
their heads.

Period between regular clunks thereafter is exactly 5 seconds.

That doesn't. It sounds as if someone ****ed up in the programming.

Apologies for confusion.

Which still remains on the part of the drives not clunking when
not spinning undermining the conclusion in the preceding sentence.
That's what the confusion is about and in need of clearing up.

From 15 seconds up until the 1 minute 59 seconds point there are irregu-
lar noises from the drives indicating disk activity, including some clunking.

Some clunks are only a couple of seconds apart.

A drive may also clunk when it accesses track zero, when it seeks
to track zero and overshoots it, running in to the end stops.
So a few clunks now and then aren't necessarily conspicuous.

Oops, meant suspicious of course.

Your English is confusing me a little!

No really? *My* english huh, but
" Hard drives spin up to speed in 15s when disk
activity starts, when data lines are attached"
makes perfect sense to you.

Please explain to us how drives are spinning up when they are already spinning.

To clarify: with the disks spinning but not connected to the NAS unit
they are silent.

When connected, they make the clunking noises in the sequence I described
above.

Right, so how difficult was it to say that in the first place when I first raised
the question.

Thanks for the mechanical explanation for clunking. I would not be
concerned by a few clunks. I am concerned by the number: every 5
seconds means 24*60*60/5 = 17,280 impacts every day.

This seems excessive and I would be concerned for the safety of my data
since the unit is used for long-term storage of data that I don't have
room for on my main machine.

It is the backup,

No, it isn't.

so if it fails I loose the data permanently.

Then it's not a backup, now is it, if there isn't an original copy of it
somewhere. And what backup is online all the time. Switch it off.

Oops, did not mean to cause any offence. Postings can get a little too
succinct sometimes. To clarify my poor English

The sequence after switching on the Buffalo unit is as follows:
1) From zero to 15 seconds, the sound is of the HDDs spinning up
2a) Thereafter, if the IDE connectors ("data lines" to use a previous
contributer's phrasing) are not connected then there is no variation in
noise as the HDDs spin at full speed with no read/write activity
2b) Therefter, if the IDE connectors are connected then there are
irregular noises from the drives indicating disk activity, including
some clunking, some of which are only 2 seconds apart. This continues
up to 1 minute 59 seconds after initial switch-on
3b) From 1 minute 59 seconds onwards with the IDE connectors connected,
there is a clunk every 5 seconds. This corresponds to 24*60*60/5 =
17,280 impacts every day, which I am concerned about.

Next sloppy English admission: yes, data which is not stored anywhere
else is not a "backup", it is an archive of non-current or rarely
accessed data.

My remaining question: are HDDs really designed to have their heads
reset / parked / overshot into the end stops (or whatever else is
producing the clunking and associated mechanical vibration) 17,280 time
a day?

 

[David Flory]

My remaining question: are HDDs really designed to have their heads
reset / parked / overshot into the end stops (or whatever else is
producing the clunking and associated mechanical vibration) 17,280 time
a day?

Definitely not!

Something IS seriously wrong. My guess would be a drive firmware glitch
that doesn't play well with your NAS setup. Whatever the cause, the
units should be fixed/replaced. It's sure to cause the drives to fail
long before their rated lifespan.

 

[Folkert Rienstra]

Folkert Rienstra wrote:
Folkert Rienstra wrote:
Arno Wagner wrote:
[snip]
Hard drives spin up to speed in 15s when disk activity starts, when
data lines are attached.

That sentence does not make (immediate) sense, if at all.
And obviously there will be no clunking when the drives aren't spinning.

First clunk occurs after 62s amongst irregular disk activity, part of
the NAS boot-up process presumably.

Regular clunking begins after 1 minute 59 seconds.

Would that be after a period of inactivity?
That would make some sense if the software idles the drives, parking
their heads.

Period between regular clunks thereafter is exactly 5 seconds.

That doesn't. It sounds as if someone ****ed up in the programming.

Apologies for confusion.

Which still remains on the part of the drives not clunking when
not spinning undermining the conclusion in the preceding sentence.
That's what the confusion is about and in need of clearing up.

From 15 seconds up until the 1 minute 59 seconds point there are irregu-
lar noises from the drives indicating disk activity, including some clunking.

Some clunks are only a couple of seconds apart.

A drive may also clunk when it accesses track zero, when it seeks
to track zero and overshoots it, running in to the end stops.

So a few clunks now and then aren't necessarily conspicuous.

Oops, meant suspicious of course.

Of course this only happens in drives that still park their heads
on the platters. These may also clunk when parking their heads.
Drives with off platter parking obviously behave differently.

Thereafter, the clunking becomes regular, every 5 seconds, after the start-
up disk activity has ceased. The 5 second metronome clunking continues even
while writing a lot of data (my initial data take-on which I aborted).

I go with the programming error theory. Buffalo haven't responded to
any of my emails since last Thursday, including my feedback of

the results of Arno's idea first thing this morning.

Which were obviously inconclusive with the drives not spinning.

Your English is confusing me a little!

No really? *My* english huh, but
" Hard drives spin up to speed in 15s when disk
activity starts, when data lines are attached"
makes perfect sense to you.

Please explain to us how drives are spinning up when they are already spinning.

To clarify: with the disks spinning but not connected to the NAS unit
they are silent.

When connected, they make the clunking noises in the sequence I described
above.

Right, so how difficult was it to say that in the first place when I first raised
the question.

Thanks for the mechanical explanation for clunking. I would not be
concerned by a few clunks. I am concerned by the number: every 5
seconds means 24*60*60/5 = 17,280 impacts every day.

This seems excessive and I would be concerned for the safety of my data
since the unit is used for long-term storage of data that I don't have
room for on my main machine.

It is the backup,

No, it isn't.

so if it fails I loose the data permanently.

Then it's not a backup, now is it, if there isn't an original copy of it
somewhere. And what backup is online all the time. Switch it off.

Oops, did not mean to cause any offence. Postings can get a little too
succinct sometimes.

Even though you bottom post you are basically a top poster and they
usually are not very good listeners, (and that's an understatement).

To clarify my poor English

The sequence after switching on the Buffalo unit is as follows:
1) From zero to 15 seconds, the sound is of the HDDs spinning up

See, you don't listen.
You have explained this already in the previous post, no need to explain
it yet a third time.

2a) Thereafter, if the IDE connectors ("data lines" to use a previous
contributer's phrasing) are not connected then there is no variation in
noise as the HDDs spin at full speed with no read/write activity
2b) Therefter, if the IDE connectors are connected then there are
irregular noises from the drives indicating disk activity, including
some clunking, some of which are only 2 seconds apart. This continues
up to 1 minute 59 seconds after initial switch-on
3b) From 1 minute 59 seconds onwards with the IDE connectors connected,
there is a clunk every 5 seconds. This corresponds to 24*60*60/5 =
17,280 impacts every day, which I am concerned about.

Next sloppy English admission: yes, data which is not stored anywhere
else is not a "backup", it is an archive of non-current or rarely
accessed data.

My remaining question: are HDDs really designed to have their heads
reset / parked / overshot into the end stops (or whatever else is
producing the clunking and associated mechanical vibration) 17,280 time
a day?

Your guess is as good as mine. Me, I'm already surprised that they survive
one such incident, given the delicate heads and even more delicate surfaces.
I've had a drive that made a racket like a pinball machine when it didn't
want to start to spin, making me hit the powerswitch in panic. Yet it still
worked perfectly afterwards when it eventually managed to spin up on the
next try. Still baffles me to this day. Apparently they are a lot sturdier
than we tend to think.

Btw, if you go back a decade or so those thumps/clunks were just normal
for a drive merely seeking.

 

[Alan Thorne]

Your guess is as good as mine. Me, I'm already surprised that they survive
one such incident, given the delicate heads and even more delicate surfaces.
I've had a drive that made a racket like a pinball machine when it didn't
want to start to spin, making me hit the powerswitch in panic. Yet it still
worked perfectly afterwards when it eventually managed to spin up on the
next try. Still baffles me to this day. Apparently they are a lot sturdier
than we tend to think.

Btw, if you go back a decade or so those thumps/clunks were just normal
for a drive merely seeking.

However, a decade ago, drives were a fraction of today's capacities
with much lower mechanical tolerences. So, what was a reasonable level
of noise and vibration a decade ago probably is no longer true for
modern, maximum-available-capacity drives?