Hello Folks,

I have access to many drives.

The problem is like that: None of the drives on running systems failed on me so far (I stress test them for a day before committing data). I have been using computers 15+ years.

But almost half of my offline storage drives (some have not been used for a year, some 2 years) are failing one way or the other. Those drives were all have been burnt-in and tested properly before they were taken offline. They were just good drives, they have been stored properly (15C to 30C), they have not been subjected to external shocks or factors. But they just are failing. SATA/PATA mixed.

Does anyone know anything about demagnetisation on hard drives? Is this what i am facing?

Can't hard drives be used as offline storage?

Thanks!

Maria

In my past experience I got a lot of HDD failed which was disconnected for long time (couple of months) then problematic.
I got such type of HDD a lot where customer are saying the drive was good and after connecting the HDD after couple of month, hdd is making unusual sound, not detected etc, etc.
Most of the case I got SAMSUNG various model HDD, MAXTOR 6Y Series in those case.

_________________
__________
There is no substitute for education and experience
THANK YOU
SHAHI
shahi.mahbub@gmail.com

Hello,

The problem is with the time.
The platter's coating layer becomes too old, and the protection of the media is getting lost...
The surface slowly gets corrosion as well.

In this way, the drive is really reliable in the first some years, and not after this period.
It is partially depends on the usage, and partially not.

Better to use less than 2 year drives....

Regards,
Janos

shahij,

The drives i am using do not have anything common with models or manufacturers (WD, Seagate, Maxtor, etc). The only thing they have in common is that they have been sparingly used and they are just sitting there to hold offline data (backups, etc).

Janos,

These are new drives. They have not been used at all: Stress test for a day, burn-in, check drive health, store data, verify checksums, take offline. They are not old drives at all.

What i am saying is that: I bought a drive 2 years ago. That same day, i stress-tested the drive and all was okay. I loaded my data. And took it offline. As the name implies, offline storage. From then on, the drive has been used very sparingly (to take few files off, to verify checksums, etc)... once or twice maybe. Now i load the drive and it is all corrupted.

This is just one scenario. What i am saying is that almost half of my sparingly-used drives are rotten just like meat. In contrast, the drives that are constantly being used 24/7 are in very good health (and never failed on me, I guess it helps to use $200 PSU as opposed to el-cheapo $5 chinese crap).

I have backups of backups, i have not lost data in my entire life. But this is frustrating. I have so many failing drives. I have salvaged the data (they had backups anyway) but unfortunately they are all out of warranty.

The only thing all these drives have in common is that they have RARELY been used.

So the conlusion is that you should use hard drives as offline storage? Why has this not been mentioned anywhere at all?

I have two nearly good solution for you.

1. build up linux sw raid5 from more drives (3-10) and store the data on this array.
The linux (2.6.16 -) have a capability to safely recover the data on the fly, if the parity sectors is readable wich is corresponding to the unredable data sectors.
Power up the array only if you need the data.

2. is the combination of the #1 only two difference:
a, keep your storage on inside one linux box
b, use some additional hardware to fully protect your drives from power surges, voltage noise related damage or corruption, lightning, etc...

(I have designed the 2b hardware, but it is on beta state, so it is too early at the moment to announce it...
If you interested, keep watching this forum or feel free to dropping me an pm... )

Regards,
Janos

N.C,

I don't feel feel secure with any hard drive connected to power let alone the risk of burning multiple drives with a RAID solution while i could only lose a single drive. The point of offline backups... is that they are OFFLINE so that when the computer, house or universe explodes, they are safe. Considering that most hard drive failures are results of electrical malfunctions, i believe my theory is valid. Just because i did not lose a drive while electrically-connected, does not mean it won't happen to me.

We have come to believe that magnetic media degrades over time with use. Is this accurate?

I have not read anywhere that they will ROT if unused (that also includes a google paper concluding that with 100,000 drives they had, demagnetization was not an issue).

Which manufacturer mentions the drives must be used? On the contrary, they mention that their drives except Enterprise ones should not be used in 24/7 systems.

What is there to use for offline storage then? CDs?

I have no use for archived data. But I cannot throw them away in case they are needed some day.

I cannot be the only person with a problem like this.

I am understand you completely, i am as paranoid like you.
This is why i am designing some protective hardwares.

Keeping the data is always like some type of race.
Race with the time or race with the environments.

All type of storage have some risk inside.
At this time, i think the offline redundant raid is the best, but not the perfect solution.

I have aready seen a lot of data lost and a lot of type of data lost.

If you use continously one hdd, this is the problem. (bearing, sa damage, headcrash)
If you use the hdd as usually, like some hours per day, this is the problem. (ramp problems, electrical problems, landing zone problems)
If you keep the drive offline, this is the problem, and so on.... (surface problems, and the superparamagnetic effect)

It is really a race....

Regards,
Janos

I have here an old 20 Megabyte drive (1986) in a very old system (Olivetti M24) that i sometimes boot (once every 5 years), and all my data is still there
Its an old Seagate. I think these were made to last.


Dobre

_________________
Murphy was an optimist

Datarecovery in Belgium, Holland, France and Germany
Datarecoverytools http://www.drtools.eu

Janos, what about running 'dd' to /dev/null automatically once the drive is booted?

That would read entire drive and eliminate demagnetisation issue. Or do we need writes to the drive as well?

Dobre, due to densities, i think those drives would not be affected by demagnetisation. I had some very old infrequently accessed drives (<1 GB) and they worked fine until being discarded. Problems seem to have started with 100GB upwards.

I think, this is a little more complicated.

If you only read, the drive writes back to refresh the sector, if the sector needs to be ECC corrected. (and of course it depends on manuf/model as well.)
But, if somethig little comes out from the coating layer, the full readout can clear the place.....

The best is to rewrite all the drive with itself, but anyway, i am still wote for the sw raid instead.
If you use 10 drive, you can use 9 drive's capacity, so, the 1/10 is enoug for being safe.
This is not too expensive, i think....

Janos

Yes, and if i have right, the asnwer is about the superparamegnetic effect.
The lower density drives not effected, because the magnetic domains have more place to fit.

I am really interested about this "lost with time" on the new perpendicular hdds...

For the storing, and avoiding the superparamagnetic, personally i like the hitachi's TAR solution. (Thermal Assistant Recording)

We will see....

Janos

I have about 10+ drives like this. Nowhere i read something similiar and no manufacturer makes a mention of this. I did not lose data (backup of backups). What if one day you wake up to find out you lost 2TB of your data coz your drives got rotten like meat?

All drives out of warranty... most were with 1 year warranty anyway. I feel i am cheated. All these drives have been very good taken care of. The problem is... I cannot take a bunch of drives and go attack a distributor and ask for RMA... the drives are from different manufacturers.

I read all of this forum... and many others. Nobody makes a mention of this or cites similiar problems.

No technology is perfect in the world. All we have come in the world and one day we will die. We are not sure about us that when we will die. HDD is a creature of human, no manufacturer can confirm you how long it will last, just can assure you about a new HDD for replacement instead of the existing dead one.

you have to go every step with your own risk.
Realize REALITY.

_________________
__________
There is no substitute for education and experience
THANK YOU
SHAHI
shahi.mahbub@gmail.com

I think you are missing the point here.

RUNNING DRIVES -> Good
OFFLINE DRIVES -> Rotten

And yes, the manufacturers do mention appropriate warnings: For example, their non-Enterprise drives must not be used 24/7. BUT THEY DO NOT MENTION THAT THE HARD DRIVE SHOULD NOT BE KEPT UNUSED. On the contrary, everywhere the hard drives are advertised to replace Tapes, Optical Media, etc for offline storage.

Show me a pointer where it is mentioned that a hard drive should not be left unused. Yet, i have 10+ drives that are rotten.

I've never seen a hard drive die this way. I had at least 5 drives that were offline for 5+ years, and there was no issues with them when plug them in...

I would just keep drives in a dry cool place and I am sure they will be fine after 5 years or so.

If you are saying they are failing, then what exactly is the failure? Are they all same model number?

_________________
Dmitry

dmitry, no they are not the same make and model. But one thing they have in common is that they all are > 100 GB drives and they have been very rarely used. Mostly Seagates and WDs.

They are kept in a temperature controlled place (which is my room). And i leave in a hot country, so it is always temperature controlled. No magnets, no nothing... beside the drives that are kept next to them work fine (they were frequently used).

Few topics below, you can see, probably a failed head. I have some like these. I have some with corrupted SMART tables, some with lots of bad sectors, some with bazillions of weak sectors...

But not a single DEAD drive though.

Really old drives worked fine though until being discarded. I have a 10+ years old 2.6 GB WD that works fine though and that hasn't been touched for almost 10 years.

We won't know the cause of your failed drives, until you find the actual failure mode of some (several?) of those drives, but your comment below reminded me of something that I had read:


There have been several studies into disk drive failure modes, including those affecting non-operating (storage) conditions. One specific failure mode which may fit for a hot country with air conditioning, is head corrosion (e.g. not enough airflow [due to not being used] allowing moist, non-moving air to affect the heads). Perhaps that is why the disk drives which are being used nearby, are ok - they are warmer (because they are spinning) and they have moving air inside the HDA...

For examples of other failure modes, you can read this study (see link to PDF near top of the page):

http://portal.acm.org/citation.cfm?id=1044961

As I said, until you know the actual root cause of some of your failed disk drives, we can only guess... Anyway, I hope you find the article interesting and useful.

No dead drives. Few head failures. Most have bad sectors / weak sectors. On some drives, weak sectors seem to get better with use (they are not turned into bad sectors). This leads me to believe this is a magnetisation issue. Important thing to note is that all these drives were in TOP HEALTH before they were taken offline. And all these have less than 100 hours use.

Vulcan, thank you for the link. Unfortunately it requires subscription, so i cant get it. If you have it, would you please be kind enough to post?



I am sorry but how will be different for any other country? Since we have air conditioning, the humidity is less. So in reality, this is better than places without A/c where humidity and moisture will be more. My storage conditions are better than manufacturer specifications for my drives.

Yes, hot country, i should lay my drives under the sun... and perhaps salt them too!

Thanks for the contribution!

Hi,


My suspicion is that the "bad/weak sectors" may be due to bad heads, especially when you said this...:


For such new drives, the super-paramagnetic effect ("demagnetization") would not be expected - hence I was looking for other problems which can cause apparent weak sectors, but without being a problem with the media itself. Unless you find the real cause of some of these drive problems (e.g. change the head stack and see if the problems were due to head degradation of some kind) then everything else is a guess... We need data (no pun intended ) about the real cause of the problem...


Oh! Sorry about that - I must be lucky to have access to that link I will try to find a different public link... [Edit: Found one, see below]



I have seen humidifiers being used with a/c units, which is where my suspicions were... Do you have a long-term humidity recorder, where the disk drives are, just to be sure?


Unfortunately you cannot know that Without a gas chromatograph to check for airborne chemical ions, it is impossible to rule-out something in your environment, which causes a problem when allowed to stay around the disk drive heads for long periods. For example, I know of one specific chemical which causes GMR (i.e. read) heads to become desensitized over time (it affected one of my customers.) This is why I said that we're all guessing, until there is some data about what the problem is, with your disk drives e.g. demagnetism, or heads, or bearings etc. and I don't like guessing

One "common factor" with these disks is your storage environment, but as I said, that's just a guess, and no-one can confirm or deny that, without finding the true cause of the read failures with one of your "problem drives"...

Good luck!

Edit: New link for the study into disk drive failure modes http://hebb.mit.edu/people/jfmurray/publications/Hughes2004.pdf