External Hard Drives Failing?

[Bluebottle]

Lets start with the numpty question.  I have a Seagate external hard drive that Fedora is now complaining about telling me it is in the process of dying (neither Ubuntu, nor windows say the same thing) as a few bad clusters (150ish) have appeared.  Do I absolutely have to replace the ruddy thing?

I suspect I know the answer (see below for further clues).

The second question is why is this happening?  The big chunky Seagate was bought* at the start of the year to replace a WD that had the same problem after 3 weeks use.  Both on a Windows/Fedora dual boot.  My suspicion is that some grinding occurred (and not in a good way) during an unexpected Windows shutdown.  I do not remember the same problem before but...maybe.  Any other possible reasons?

TIA

[tiermat]

Is it one of the "Inverted T" Seagates or is it a slab one? If the former then it is monumentally bad design that is causing the problem. Consumer grade HDDs are designed to be laid flat, not stood or their end or side (as Elonex found out to their cost) and Seagates are the worst offender in this respect.

Obv if it is the slab type then this is irrelevant, apart from the fact that Seagates are pants, and the only reason I own one at the moment is that they are probably the easiest to get that work properly with a Wii softmod.  The other one I have, which is currently attached to this computer is a WD Elements drive and is fantastic!

All in all I would back up the data you have on there, try doing a full re-format (which should mark the dodgy blocks) and try it out.  Don't trust it to anything which you value though.

[fuaran]

The hard drive could be getting too hot. Does the case have any sort of fan?
You can get software to monitor the hard drive temperature, eg Speedfan. Though I'm not sure how well it supports USB drives.

[inc]

Consumer grade HDDs are designed to be laid flat, not stood or their end or side (as Elonex found out to their cost) and Seagates are the worst offender in this respect.

I don't know how believable HDD reliability stories are, when I bought my Qnap, direct from them,  they recommended Seagate as they had the least problems in their experience and it stands on it's side.

[tiermat]

Fair point inc, that may well be their experience, or it may be sales flannel.  From personal experience (a not insubstantial time in IT) when it comes to CONSUMER grade HDDs one year's best is next years rubbish (you only have to look at what are commonly known as DeathStar drives to show that, when made by IBM they truely earned their name, when Hitachi bought out that division of big blue, but carried on using the name and technology they started to, and still do, lead the market in reliable consumer grade HDDs)

Anyways back to the point I was going to make, propmted by fuaran, HDD temp is part of the SMART sense set, so should be able to be read by speedfan et al.  The fact that Fedora is showing errors is telling me that it is able to read the SMART data via USB.

FWIW the inverted T drive I have at home will not mount on fedora, it just refuses due to it being so bobbinsed, but windows mounts it fine, and will allow you to copy files to it, just don't expect that file to be in any useful form when you come to need it!

[TimO]

If you've got any doubt whatsoever about the reliability of the disc (which I think you have), the first thing you do is backup any data which you don't want to loose.  Then you start to experiment, and procrastinate etc

It's working now, so backed it up immediately.  Do not continue to use it if you've got nowhere to back the data up to, turn it off, and go and find something else to copy the data to, and then power it back on again.

Generally with hard discs, the more you use them, the worse they'll become if they're failing.

[ian]

Fact is, drives fail. They're incredible bits of engineering built to impressive tolerances, but it's the nature of probability and these fine engineering tolerances that they will fail over time. Frankly, it doesn't much matter on the manufacturer, or orientation, or the colour of the case - and assume that anything you read on the internet about the reliability of certain drives and manufacturers is bollocks, because it is. It's best to assume that any drive can and will fail and have appropriate back-ups. Bad luck and statistics mean some users have lots of drives fail, and others have none.

Once a drive starts to misbehave, assume it's dead but just hasn't realised.

[Ivo]

Once you've made a copy of all your data on this disk, you don't have to discard it but you can store it somewhere else so you have some sort of back-up (albeith not 100% secure) in case of fire or burglary.

[alexb]

We transfer 100+GB files using portable hard drives all the time at work. We're pretty much filling alomst the whole drive, then wiping it and doing it again, so not quite the typical usage.

We've found that cheap drives are just not an economy you want to countenance.

We've also found Iomega drives to be pretty poor.

So far Western Digital seem pretty reliable.

YMMV as always.

[ian]

Once you've made a copy of all your data on this disk, you don't have to discard it but you can store it somewhere else so you have some sort of back-up (albeith not 100% secure) in case of fire or burglary.

Hmm, it's not a back-up if it's failing.

Go outside and dismantle it with a hammer. It's fun and probably the most practical use for it.

Drives are cheap. 1.5 terabytes will set you back less than £50.

[TimO]

I've taken some of our failed drives apart, and it's alarming how much dirt the filters inside them can contain.  Given how much debris has come off of the disc's surface, it's also amazing that the drives worked for as long as they did before finally deciding to utterly fail.

[Biggsy]

We've also found Iomega drives to be pretty poor.

Iomega is just the enclosure, the HDD will be a different make.

Like Ian, I can't believe there can be much difference in reliability between any of the major brands of the actual hard drives (currently: Seagate, WD, Samsung and Hitachi).  Why would there be, and how could one continue to be successful if it was consistently worse than another?  They're all fiercely competetive, with similar technology and prices at any one time.

[nuttycyclist]

...
YMMV as always.

Indeed.   We had a long state at work with drives of a particular make failing after having been in use for a period of time, especially if the PC was rebooted (helpdesk was swamped after a power cut!)

It was obvious that one make in particular was at fault.

But maybe it was only one batch of one make as the manufacturer is a big one and seems reliable in the PCs made by a different company?

(Names withheld on purpose)

[ian]

It was obvious that one make in particular was at fault.

Except it isn't. Statistically, this could simply be bad luck. Someone else could have hundreds of the same drive and have none of them fail. Does that make them more reliable? No. You simply can't say anything meaningful about reliability from the information available.

This is why I said most internet comments about disk reliability are bollocks.

[nuttycyclist]

The volumes I'm talking about did give us the statistical implication.    Thousands of same manufacturer PCs installed at same time across multiple sites.  About 3 common HDD suppliers spread across those.  1 of those consistently failed under known circumstances (power cycle after x months of use), swapping for HDD from different supplier prevented recurrence.

Personally I think it may have been a combination of dodgy batch/computer firmware/similar, as the HDD manufacturer is well known and still going strong.  Hence me not naming.

[TimO]

Even the most reputable manufacturers occasionally have batch problems, and this is why generally people recommend that if possible you don't buy all the discs for a RAID device in one go, but buy them from a number of suppliers, since it reduces the risk of multiple simultaneous failures caused by a bad batch.

[Biggsy]

Even if it's more than a bad batch, it won't be more than a bad model - and models don't hang around for long.  No successful manufacturer repeatedly brings out bad models.