Why are my hard drives failing?

Solution 1:

Is your power supply old too? Perhaps its under/overpowering the drive which is causing the failure. If you have a multimeter, I would try measuring the voltage that is running in your hard drives and watch it over a period of time. Another culprit may be 'dirty' electricity, so a UPS may be in order so that it will 'clean' the power going into the PSU.

Solution 2:

I agree with others: power.

However, with a twist.

ALL the components need to have a COMMON ground - the chassis is typical, but in your case, who knows! A "drifting ground" would cause this, I'm sure.

You want all the components tied to a single ground AND that ground tied to the grounding from your facility's "power grid" ground. This is IMPORTANT.

BTW, it is possible that all your old hardware is actually still OK! I have found that equipment that was served with a flaky power supply sometimes survives it OK when a proper supply is provided.

I hope this helps.

RT

Solution 3:

This is an old post and the original question may no longer be relevant to the person asking the question. However, for future reference to people building a budget PC, Power is not an all encompassing issue with disk drives. It is, in my professional opinion as an EMC certified implementation engineer, a misleading answer to blame a power supply as the sole responsible party given that the computer is inside of a card board box.

Hard disks vibrate, and though there is no particular position, vertical, or horizontal, that increases or decreases the longevity of a disk, there is, however, a vibration factor that a hard drive with spindles creates. The drives displayed here are just laying in a card board box. This is an example of budget engineering, and the vibrating drives are sitting on its side, further increasing the resonation on the platter. Though this isn't an answer in itself, improperly mounted hard disks MAY lead to a disk fault because of a vibrating platter disrupting the read and write heads from touching the platter correctly.

Power, cheap power supplies are always bad for computers in general, however, it is unlikely this PSU killed the hard drives and not other more sensitive components on the board. This system is in a cardboard box, so the engineering and power could have led to a more catastrophic failure, but not necessarily his disk fault. It's possible, but not proven in this case.

Heat: heat can destroy a disk, however, if it wasn't hot to the touch at the time of failure, heat is not the culprit. A card board box is not a good feat of engineering for a PC or server. You are better off bolting your parts onto a computer desk or work bench, at least they would be grounded.

Soft RAID and cheap drives. Given the card board box and old parts viewed in the photo, you appear to be using standard desktop drives and a Soft RAID. Desktop drives can be placed onto a RAID controller, however, with the increased I/O on the disk, the chance of a disk fault increases. The disks imaged in this case are not on a hardware RAID controller, but are being grouped together with a software component on the motherboard. This is not ideal for hard drives. This increases the workload on your CPU, and soft RAID's have been known to have errors and kill hard drives prematurely. It is likely that the soft RAID killed these drives above all else.

Prevention for future builds: If you are reading this and seeing this old user scenario via google question or what not:

-ensure that your disks are properly mounted in a stable hard drive chassis. Bolt in your disks with at least 4 hard drive screws, or use a special disk sled that goes with your chassis.

-Ensure that you have adequate air flow in your case, hard disks in a RAID tend to have more I/O on the disk, and will be much hotter than if the physical volume is mounted individually.

-Do not use a cheap power supply. Dirty power is a killer of expensive computer parts. Also ensure that your power supply provides enough wattage to handle the desired work load.

-Use a RAID controller card! Never use the soft RAID on your motherboard. Soft RAID's reduce disk performance and increase the chance of disk failures more so than that of a RAID controller card.

-RAID in general increases chance of disk failure because of the increased I/O across all of your volumes. The larger the pool of disks being joined, the higher the chance of failed drives. If you RAID your drives, always make use of parity drives and hot spares. You may lose your data if you RAID 0 2-3 disks. If you have 3 disks, use RAID 5! 6 disks on RAID 5 (4+1) with a hot spare is ideal if your drives are covered under a warranty. If you can't afford more disks or your disks are out of warranty, don't use RAID.

-Desktop drives are not Enterprise drives. Desktop drives are similar to Enterprise drives, but are not designed to handle huge workloads brought about with RAID controllers. If you buy desktop drives from newegg and RAID them our on your motherboard, you are likely to see at least one drive failure in your first year. The longer you operate your machine on a RAID, the more I/O is being written to disk and the higher the likelihood your volume will have failures. Combine cheap drives with cheap motherboard soft RAID and you will be hurting.

It is likely that this user experienced all of these factors in his shoe box server. Cheap power, bad air flow, old cheap drives not properly mounted in a chassis, and a motherboard soft RAID...this all increases the chances of a disk fault.

Solution 4:

I can't imagine how you have good ventilation and cooling in a shoe box? You really should shell out the 50 or 60 bucks for a real computer case?

Power strips only guard against power surges; common problems for electronic equipment is under voltage (brown out) and over voltage (spiking). Also common is EMI noise - we had an unstable computer a while back which turned out to be caused by having a treadmill on the same circuit (I personally verified this beyond doubt). It would kick the modem offline, and cause the system to just freeze up from time to time.

Also, continual exposure to noise and fluctuations in the power supply with eventually damage the PSU, over time, decreasing the quality of power delivered to the electronics.

EDIT: Electric power fluctuations can be isolated to specific circuits. More importantly, high-draw appliances such as microwaves, refrigerators, treadmills, stove and similar can have a significant impact on the power quality on that circuit. And things like fridges also have a continual on/off cycle of operation which by turn browns and spikes power on the line when the motor kicks in and out.

Also, if you are being served by the same power company, they may be having ongoing trouble supplying voltage across the board. Constantly fluctuating between 105V and 125V will have a negative effect on electronics (as I understand it).