Advice on possibly clicking hard drive

[Rakk]

Hi guys

So I could do with a small bit of advice.
At the weekend I was busy copying over some movie files onto my Caviar Black drive, and I heard a clicking noise (followed by some face pulling cos clicking from a hard drive is never good and some hasty copying of everything useful onto an external drive).
The clicking stopped when the files finished copying over - so is this the sign of impending doom or if it was about to die would it still be clicking?

Notes: it's not my OS drive and the main game I play is not on this drive, so its not often under intense use - it's got all my docs and music and a few smaller apps that i use sometimes but nothing too intensive.
It was pretty full, in fact it wouldn't let me copy everything due to lack of space - I since cleared a bit of space but its still pretty full (as in about 50-100GB free space out of 1TB now which I realise isn't that large an amount).

I ran WD's software to check disk drives, but all it really checks for is bad sectors and errors I think, and I wouldn't have thought that clicking was caused by bad sectors/errors anyway, so it passed those tests

So yeah, it is still within it's warranty period (only just) - yay for Western Digitals 5 year warranty
But the question is, since the click has vanished does that mean I was just imagining it (which is quiet possible ) or maybe that it was something else, and does clicking on a hard drive that's about to fail come and go ?

 

[ubuysa]

I think the clue here is "It was pretty full...". The seeks will be long and if the MFT is close to the start of the volume (and it probably is) you'll be getting a lot of long seeks (especially if you have fragmented files - which you probably do) and that's probably why you can hear it. The fact that the click has gone now you've freed up space confirms that I think.

 

[Rakk]

I think the clue here is "It was pretty full...". The seeks will be long and if the MFT is close to the start of the volume (and it probably is) you'll be getting a lot of long seeks (especially if you have fragmented files - which you probably do) and that's probably why you can hear it. The fact that the click has gone now you've freed up space confirms that I think.

Actually it does a defragment every week I think and claimed it was 0% fragmented when I checked after googling after hearing the click.
So you reckon it was just a tad too full and that's what was causing the click?

Not sure what an MFT is
The drive actually has two partitions, one small one (610MB) that was once a Windows Recovery partition (this drive was my OS drive in my previous machine) and the rest is the main partition that I use.

 

[ubuysa]

Actually it does a defragment every week I think and claimed it was 0% fragmented when I checked after googling after hearing the click.
So you reckon it was just a tad too full and that's what was causing the click?

Not sure what an MFT is
The drive actually has two partitions, one small one (610MB) that was once a Windows Recovery partition (this drive was my OS drive in my previous machine) and the rest is the main partition that I use.

The MFT is the Master File Table, it's effectively the directory of where files (and their fragments) are located on the disk. If the disk is very full the heads have to keep seeking all the way across the disk and then back to the MFT to find the next fragment or file. I believe it's the inertia of the heads as they stop after these long seeks that causes the click.

A very full disk also greatly reduces the performance of the disk, seek time is the major component of I/O response time and a full disk causes much longer seeks and thus slower I/O response. It also tends to increase the probability of file fragmentation because there are fewer large contiguous areas of free space and it slows down, and could even prevent, a defragmenter from working because free space is needed during a defrag to temporarily store fragments whilst a file is defragged. When I was responsible for disk storage management in large mainframes we'd never allow any disk to become more than about 60% full for exactly these reasons.

 

[Rakk]

The MFT is the Master File Table, it's effectively the directory of where files (and their fragments) are located on the disk. If the disk is very full the heads have to keep seeking all the way across the disk and then back to the MFT to find the next fragment or file. I believe it's the inertia of the heads as they stop after these long seeks that causes the click.

A very full disk also greatly reduces the performance of the disk, seek time is the major component of I/O response time and a full disk causes much longer seeks and thus slower I/O response. It also tends to increase the probability of file fragmentation because there are fewer large contiguous areas of free space and it slows down, and could even prevent, a defragmenter from working because free space is needed during a defrag to temporarily store fragments whilst a file is defragged. When I was responsible for disk storage management in large mainframes we'd never allow any disk to become more than about 60% full for exactly these reasons.

Ah right - that certainly makes sense, I may do some testing over the holiday weekend and see if it does it again if I nearly fill it again and then I'll do a mass tidy up - I've been meaning to reorganise for a while, I think I need another HDD
It was probably at about 750GB out of 931GB when I started copying over 150GB of videos and that copy was when it was clicking (it was the next lot that refused to copy due to lack of space).

 

[ubuysa]

Ah right - that certainly makes sense, I may do some testing over the holiday weekend and see if it does it again if I nearly fill it again and then I'll do a mass tidy up - I've been meaning to reorganise for a while, I think I need another HDD
It was probably at about 750GB out of 931GB when I started copying over 150GB of videos and that copy was when it was clicking (it was the next lot that refused to copy due to lack of space).

You can't have too much disk space.

If you look at the specs for most HDDs they quote something called "average seek time", this is the time to seek 1/3 of the way across the disk. Seagate list average seek time (see http://www.seagate.com/staticfiles/...sc/desktop-hdd-data-sheet-ds1770-1-1212us.pdf) but as far as I can see Western Digital do not.

You'll typically see something under about 10ms for average seek. Some vendors also quote "maximum (or full stroke) seek time" which is the time to seek all the way across the disk, this can be as much as 45ms. You might also see "track to track (or adjacent) seek time" quoted and this is the shortest seek possible to the next track, this is often less than 2ms.

All disk I/Os also suffer latency delay (due to the rotational speed of the disk) and this is on average the time for the disk to make 1/2 a rotation. On a 5400rpm average disk latency is 5.6ms, on a 7200rpm disk average latency is 4.2ms.

Thus an I/O with average seek and average latency will take a shade under 15ms on a 7200rpm disk. The same I/O with near full stroke seek will take nearly 50ms, and that's assuming the file is a sequential read and is not fragmented. If it's in two fragments for example, you'll suffer the average (or full throw) seek plus latency twice, once for each fragment, making your I/O access time around 30ms with average seek and a whopping 100ms for a long throw seek (for half the file anyway). That's why fragmentation is such an issue and it's also why a faster disk rpm is better.

That's also why you want your disk to be "short stroking", which means it's always performing the average seek or less, the only way to achieve that is to not overfill the disk more than about 60%, ensure all files and folders are placed around the MFT and with the MFT in the centre of the data. Windows does not have any tools that let you specify file, folder, and MFT placement, but my favourite (and pretty cheap) hard disk tool Ultimate Defrag (from http://disktrix.com/) does allow you to specify file, folder, and MFT placement and it allows you to get the very best performance from your hard disk.

 

[Frank100]

Hi,

Other than the excellent information and advice offered already I wanted to just say that a clicking noise from the disk is usually caused by the actuator arms that hold the read/write heads. A consistent clicking, say every 1 sec of the same volume and general sound is likely to be an arm that is sticking or cannot read the disk and so it keeps seeking back and forth across the platters looking for something it can understand. Normally it is a physical problem with the actuator arm, although less commonly it can be a controller chip sending erroneous commands to the actuator arm.

If the clicking is slightly louder than it is with normal use but is more random in volume and length then it is more likely to just be excessive moment of the arms and reads/writes taking place. The process of reading/writing requires the heads to identify data at the beginning of the area it is writing to which enables it to synchronise its speed with the speed of the platters and following a read/write to conduct a CRC check for the write, if that was what it did. This synchronising aspect is what causes some noise when the heads read/write, as opposed to just sit there hanging over the platters waiting for something to do.

Maybe just keep an eye on the disk for a few weeks to be sure. If it didn't sound like the first description I gave then it probably wasn't a sticking actuator arm. Even if it was this is fixable but not cheap as it should be done in a particle free cabinet.

Frank100