WD Caviar Green / GP - clicking - normal?

[mgarl10024]

Hi all,

Before anyone says it, yes, I do know that harddisks click.

I've recently bought 2xWD10EADS - Western Digital Green Power 1TB.
viewtopic.php?p=456447

I've just emailed WD with the below query.

Hi,

I bought two of these drives (1TB Green Power) and have been very pleased so far. I'm assuming that both drives are doing the same as described below.

When the machine is idle, about every 10-20 seconds I hear a click from the drive and the power usage drops by a few watts. Then a few seconds later, I hear another few clicks (like a file writing), and the power usage jumps back up to where it was. This cycle repeats.

At first, I was concerned that this was a well documented problem where the drive is unloading the heads, however the SMART parameter (193 Load_Cycle_Count) is at only 36 for 117 hours of use. So it can't be that.

I just wondered
- what these drives could be doing every 20 seconds or so
- whether this was 'normal' or a sign of failing
- whether this frequent clicking could prematurely wear out the drive
- whether you had some way of confirming my SMART readings

Any help appreciated,

MG

WD's response was to just send me a Windows app (even though I told them I'm running Linux). *sigh*

I have seen some owners of these drives on this forum. I just wondered if you could comment on your experiences - do your drives do exactly as mine do?

Thanks in advance,

MG

[Inglix]

The 32mb cache green power drives now hide the head unloading count from the user. The load cycle count mirrors the spin up & down stat.

With WD's Idle firmware tool (wdidle3_1_00.zip) you can increase the time before the heads parks up to 25.5 seconds, or disable the head parking entirely. I'm going to set my new build's drives to 25.5 seconds tonight because its os drive is doing the cycle down every 8s and wake up 2s later thing.

[mgarl10024]

Hi Inglix,

Thanks for your reply.

The 32mb cache green power drives now hide the head unloading count from the user. The load cycle count mirrors the spin up & down stat.

What you are saying certainly seems to correlate with my SMART data. Can I ask how you know this - do you have any urls that state this or is it just a theory you have from owning these drives?

If you're right, I'm actually quite annoyed about this. What's the point of having SMART data if Western Digital choose not to update values they see as inconvenient? How do I know that my drives aren't fitted with cheap heads that misread all the time, but that WD have disabled the SMART value from updating because if it did I might send the drive back?

[mgarl10024]

Hi Inglix,

I did some further searching and found that you were very much right - in some of the later drives the figure isn't updated which I think it very misleading by WD and quite disappointing.

I did a lot of researching about how to disable the unloading before posting and had already found the wdidle3 util, however I also found someone else who had posted a response from WD in which they distanced themselves from it as much as possible, stated it was for old drives and described it as warranty voiding.

With a lack of alternatives, I took the plunge (don't tell WD!) and it worked. No more parking. A word of advice though - after setting the new value in the drive, you'll need to restart the machine (I shut it down and restarted). Don't just change the value and expect it to stop immediately like I did.

These were the first WD drives that I've ever bought, and while I like them I can't help feeling let down by WD.
If people are experiencing issues (even if it is with a Linux O/S), I don't think it is good enough for a company to simply shrug and say "test the drive, if it passes all is well", although I do accept they have to draw the line somewhere and don't want to be dragged into systems administration.
I also don't appreciate their silence on the issue, or their lack of alternatives - I don't feel that I should have to use a largely undocumented and warranty-voiding tool to prevent undesired behaviour in a new product, with no other options from the manufacturer.

Anyhow - enough moaning - *touch wood* all seems to be working ok.
Thanks once again for your advice,

MG

[penguinbelly]

Edit: nvm

[karl]

Sounds like these drives wouldn't be so good to use for Linux Software RAID due to their odd power and speed management. It looks like you're using them for that, and the unsupported tool got things working OK... risky.

[mgarl10024]

Hi Karl,

I'm not so sure that it's the RAID that's the issue - more the usage pattern.

If this drive is used as a storage drive which is accessed infrequently (imagine a log is written to it every 24hrs), then this drive is perfect. After 8 seconds, the aggresive power management would park the heads and for the next 23 and a bit hours, you'd be saving near 4 watts.
Imagine a datawarehouse containing 1000 of these drives all with their heads parked - electricity saving alone would be 4kw (4wx1000drives) which could be £3000-£4000 annually here in the uk. Not to mention lower heat and so cooling charges. I can honestly see what they are trying to do.

However, this isn't a server drive, and if you had this as your main system drive (especially under Linux), or a media drive where a player was reading the next bit of an mp3 file every 10 seconds), this drive isn't for you. The aggressive power management techniques wont save you a lot, the click/clunk will annoy you, and the drive could be running the risk of wearing out quicker as you will soon be exceeding the number of unload cycles this drive is designed to survive 'as a minimum'.

All I wanted as a user is the ability to disable this feature when I feel that it's not suitable for my requirements. I don't think that's a lot to ask, and WD's lack of options is disappointing, meaning I had to resort to the unsupported tool.
It's the same reason why I don't go around in my job as a developer hard-coding system properties - because the customer may know what they want better than the system provider, and will always demand flexibility...

MG

[wlan-11g]

I'm considering purchasing these drives as well

Just reading through this thread, the clicking noise happens every 20 seconds? Is this the only WD drive that does this? Is it very audible such as a mis-read?

Is this HDD suitable for running 24/7 in your opinion?

[mgarl10024]

Hi wlan-11g,

the clicking noise happens every 20 seconds?

20 seconds was a bit of an overestimation - although with your ears alone it's trickly to get right!
The timer on the drive is set to 8 seconds as default, but your frequency of access determines what happens.

So,
-file accessed
-8 seconds passes with no access
-heads unload/park
-some time passes. Could be just 2 seconds if your OS/app is accessing every 10 seconds
- heads unload
- cycle repeats.

Is this the only WD drive that does this?

I think all of their green range do (anyone disagree?).
Though, the sneaky bit is that early drives correctly log the number of times this is happening through SMART. When people started getting concerned, WD disabled the logging to the SMART parameter, so my drives don't accurately tell me how often this is happening.

Is it very audible such as a mis-read?

Difficult to say really. It's a "different sound" so while it certainly isn't loud, you notice it. Sounds like something small unhinging - not that my description helps at all.

Is this HDD suitable for running 24/7 in your opinion?

Take a look at my response to Karl in this thread. Really depends on usage
- for occasional access it's great and will probably save you a lot of power
- for frequent (but with accesses more than 8 seconds apart by default) accessing like a Linux OS/App writing every 10 seconds, you wont save a lot of power and the drive will be far more mechanically active.
It's worth me adding that whilst I am frustrated with WD about their silence on this issue, the drives have been faultless (so far!).

Hope that helps,

MG