HD randomly cuts off, restarts with loud click

Topic

New Reply

IBM SCSI 18GB HD MAY BE NEAR DEATH

DEVICE Model— IBM DNES 18Y CLAR18 SCSI HD– 18 GB SCSI

PROBLEM— HD clicks loudly, and immediately afterward I hear an accelerating drive motor, after which all is “normal” again. Problem occurs maybe once daily, although it occurred three times, today. Only once in maybe 50 times has the cut-off and noise been accompanied by a Windows “Drive Caching Failure” error message.

BACKGROUND— This 18gb SCSI was born in 2000-2001, and is connected to an Adaptec 39160 controller. The drive was bought new, along with three other such drives, and has very few hours on it.

DIAGNOSIS— At first, I thought this might be related in some way to drive caching, which is a performance enhancing feature very UNdesirable for keeping data safe. As an option, I made sure it was off on all my other drives, but on this drive, caching was ON, for some reason. I switched caching OFF, but I am still awaiting results. Frankly, I do not believe caching is the culprit.

YOUR THOUGHTS— I need them. When I Googled, “drive caching failure”, it appeared some believe this is a sign of impending drive failure. This astonishes me, since I treat drives well, and this is directly out of storage, technically still “new”. (I realize storage is not the kindest thing I could do to one of these drives.)

At this time, I am running SpinRite 6 on all my SCSI HDs, to make sure the data is relocated if required from head skip damage. Meanwhile, I have imaged the drive volume (using Acronis 2010) already, but after SR6, I plan to re-image.

Reply | Quote

Replies

Hi Alphaa,

I assume the drive noise you are hearing is much louder than it use to be.

Do you have a software setting to power down the drive after a specified period of idleness?

You mentioned the drive has very little usage on it and you have recently taken it out of storage. How long has it been in storage? Hard drives are precision devices and use high precision bearings which rely on self contained lubrication. If a drive sits in storage, even a new drive, for extended periods (years) without being spun up periodically (a few minutes every other month or so), the lubricant can harden somewhat causing more than normal heat generation. This can shorten a drive’s life.

It will be interesting to see the results from your SpinRite testing.

I am not familiar with these older drives, but it is good you have imaged the drive just in case. I have an 8GB IDE drive in a Pentium II machine dating from 1998 that I have maintained for playing vintage flight simulators and games that do not play well an any OS more recent than Windows 98. I rarely play any of the old games anymore, but I hook the computer up and run it for an hour or so every other month. This is the oldest functioning hard drive I possess. I would not be surprised if it refused to spin up the next time I power up the computer. Then again, I have seen drives go bad anywhere from the date of purchase to five years of use.

Reply | Quote

According to Scott Moulton from my hard drive died, when you hear a click followed by the sound of a fresh spinup, it means the data track info was lost so the heads don’t know where they are, so they park and the drive spins up again to try and lock back onto the data tracks. The “click of death” is when this repeats over and over 3 or 4 times, each time the drive failing to lock back onto the data tracks, after which the drive just sort of gives up until its restarted again and goes through the process again.

Sounds like yours is successfully relocating the data tracks, so far, but it never gets healthy again and if this is what it is, you can expect it to get worse and even fail completely at any time.

Reply | Quote

Just because you haven’t used it doesn’t mean it’s OK. Not being used may have caused the bearings to seize.
It is time to retire your disk before it does so on it’s own. Saves having to re-build in the middle of important gaming activity. ;-))

cheers, Paul

Reply | Quote

Thanks to Gerald, Byron and PT for your prompt responses–

1. Gerald, your suggestion about a power-down setting is good, and the possibility occurred to me, as well. I’ll try to determine what the jumpers on the drive are set to enable, as well as whether I have set my Windows and/or CMOS to power down– and post a response, soon. I keep a running system log to keep track of such things, but I need to double check against the actual settings, anyway.

More weight attaches to the power-down hypothesis, since the loud click and whirring of a restart occurs when no activity has occurred for some time. But I cannot make a categorical statement the restart occurs only after inactivity, and so this is a watchful waiting period..

BTW, the SR6 readings were flawless, but I’m on notice the lubricant issue hangs over this drive.

2. Byron, your information is very useful, in case this turns out to be something other than a power-down behavior. Do you have a link for Scott Moulton?

3. PT, thanks for the warning– three of the four drives have been in storage for more than five years, but under reasonably good temp and humidity control. However, I know I have pushed the lube envelope. Here, I plead distraction, and have not had time to build the system yet for which the drives had been purchased.

Reply | Quote

Myharddrivedied.com is his website. I heard him talk about drives on the myharddrivedied podcast he does @ podnutz.com. Also on ComputerAmerica a while ago but I don’t think there’s any way to get those podcasts anymore.

Reply | Quote

LATEST FINDINGS ON HARD DRIVE CUT-OFFS

This past weekend, with some time to go into the machine deeply, I examined everything, starting with the DC connections via Molex (white nylon) connectors.

Immediately, I discovered the “old faithful” Molex on my system was not consistently tight. In one case– the SCSI HD at issue– I found actual play between two connectors, and this was just sensitive enough to account for the “spontaneous” cut-off, and restart.

Of course, long before posting here, I had checked and rechecked the connectors.

However, my technique was faulty. Having forced the two connectors together firmly, I simply left them alone, on the presumption they would stay as tightly-connected as they seemed— I did not also attempt to wiggle them afterward for play. Had I done so, I would have noticed the play in the SCSI power connector.

That left me with two options– replace the connector, hoping for a better fit, or do some Midnight Engineering. I chose a plastic-based bag tie to pull the two connectors toward each other, and twisted the tie, leaving the connectors tightly and permanently connected. Appropriate technology, and it works fine– no further cut-offs of any kind.

Sorry to disappoint those who expected a real bear hunt.

BTW, all the attention paid to my system did reveal some errors in one of my IDE drives which I use for data. Seagate’s Seagate Tools, v. 2.07, the Long Test, revealed one error on the 400 gb HD, and once fixed, I left the drive on SpinRite 6 overnight. After SR6, the plan is to run through chkdsk on boot for a full system consistency check. Since this drive has some really important data, it was fortunate to have noticed the problem. This IDE never has been in storage, and has been used for about 12,900 power-on hours (POH), just a pup.

Reply | Quote

Hi Alphaa,

That is great news! I am glad you fixed the problem. I love your “Midnight Engineering” tag. It reminds me of something I did to the Pentium 2 I earlier mentioned. I found a Slot 1 P3 processor with the correct specs to run on my motherboard about eighteen months ago. My slot 1 “stand” is of the variety that requires “caps” to be placed on top to securely lock the P3 processor card into place. The structure of the P2 card did not require the caps. I found these almost impossible to find anywhere, and saw the loose play in the slot as unacceptable. My “Midnight Engineering” solution was the purchase of appropriate length plastic locking straps (the kind that utilizes a ratchet type system, so once you thread the end through the clip, the only means of removal is to cut the strap). I used one strap on each of the two posts to lock the P3 card into the slot, being careful to not place the card or the slot posts in a bind. It worked like a champ, and the P3 processor is still chugging along running those old games. Everytime I check it, all is well.

Thanks for posting back your fix. You never know who that will help down the road.

Reply | Quote

12,900 power-on hours (POH), just a pup

A year and a half is only just run in!

cheers, Paul

Reply | Quote