More Hard Drive Hijinks

[Gurm]

So I have this Seagate hard disk, which if you recall from my last update was running poorly.

So it:

- Runs too darn hot (50'C vs. 30'C for the other drive in the machine)
- Runs very slow (18M/sec. vs. 26M/sec. for the other drive, equal data, Seagate SHOULD be faster)

It's also intermittantly corrupting data and once in a while it makes that clicky clicky noise (never good).

BUT... EZ Recovery finds no errors, and its SMART warnings haven't gone off even once.

I wonder what gives? Clearly I'm getting the data off, verifying that it's all ok, and then wiping the drive and RMA'ing it with Seagate, but shouldn't errors pop up more consistently? Or shouldn't SMART be triggered? What's up?

- Gurm

[VJ]

Hmm, it could be something mechanical (bad bearing ?) causing it to run too hot; this in turn could put the disk in a slower mode as a safety-precaution (to prevent it from overheating)...

Just for reference, my Seagate 10K.6 idles at 35°C, my IBM UltraStar 36lzx runs a bit hotter, but nothing close to 50°C.

Have you contacted tech support ?
Either way, I'd try to RMA it, explaining what the problem is. (let them figure it out )

Jörg

[SpiralDragon]

RMA it.... smart us not as smart as they claim it is.... i had a drive that was obviously dead... yet smart keept showing it in good condition.... smart is not that relaiable..... and to be honest seagate drives are the ones i had the worst experiances with...

[rugger]

Maybe you SHOULD STOP BLOODY USING SANDRA SISOFT AS THE BENCHMARK. Sandra Sisoft is crap as a hard drive benchmark, unreliable and unpredictable.

Get hdtach, or Winbench 99, and run the hard drive surface performance tests. From there, you will probably be able to see if it is working correctly.

BTW, there are no slow modes for hard drives. The drive is only be able to work correctly at the speed it was designed for.

Also, get hdtemp to read off the smart values for the drive and paste them here. Then we can see if the drive is doing stuff that dieing drives ussually do.

[Wombat]

Originally posted by rugger
Maybe you SHOULD STOP BLOODY USING SANDRA SISOFT AS THE BENCHMARK.

That has nothing to do with the data corruption or noise.

[rugger]

Wombat, you are right,

But at least with good software tools, a more accurate diagnosis can be made.

Its like trying to diagnose electrical problems without a multimeter. Possible, but a silly idea.

[The PIT]

I like what http://www.storagereview.com/ say about hard drive benchmarks. Basically none of them are brilliant.

[a_h]

I agree with rugger, a close look at the smart values would be a help. Since your seagate works without data loss it's not necessary that the smart thresholds are reached - it's more interesting how close they come to them.

Sorry if I misunderstand you, but if you're really getting corrupt data, rma it.

What about cooling your hdd with a fan?

PS: Storagereview is a great site for hdds. Always recommended.

PPS: the seagates are known for their quite realistic & honest smart values, so please no wonders if some other hdd from any other manufacturer show way better results as a new seagate.

[Gurm]

I don't understand the SMART values anyway. The threshhold for things is like 50, and the actual value is 200... isn't that OVER the threshhold? Erm, hold on.

Here it is...

- Gurm

[rugger]

Ok,

The raw values for smart are ussually the most useful when self-diagnoising a hard drive. They are specific to the drive, but ussually match some internal counter measuring the drive's operation.

> Raw read error rate 67 6 00000086FE0Bh CR OC PR ER
> Hardware ECC recovered 67 0 00000086FE0Bh OC ER EC

The raw error rate is high, but that is ok because hardware ECC is fixing all the errors (hard drive read heads are never 100% accurrate, they generate a fuzzy picture that the drive error correction uses to generate the real sectors) Many drives report the after ECC read error rate for this value, but the seagate is reporting the pre-ECC read error rate.

> Reallocated sector count 100 36 000000000000h CR OC EC SP
> Current pending sector count 100 0 000000000000h OC EC
> Off-line uncorrectable sector count 100 0 000000000000h EC

This indicates that the drive is not suffering from bad sectors (as the raw values are all zero). If the reallocated sector count becomes high (or increases too much over time), the drive is going bad. If the current pending sector count is non-zero, the drive is waiting for a write operation on the bad sector(s) to reallocate the sector.

> Spin up retry count 100 97 000000000000h CR OC EC
> Spin up time 100 0 000000000000h CR OC

The drive is not having problems spinning the drive up, so the bearings are not causing any functional problems.

> Seek error rate 81 30 000007184F9Eh CR OC PR ER

This is an unusual value, but I think it ok because a google search reveals most seagate hard drives report a high seek error rate.

> Ultra ATA CRC error rate 200 0 000000000001h OC PR ER EC SP

A non-zero value here is a worry. I would replace the cable you are using with another one, just to be sure. Both of my drives report zero for this value, even after months of usage. It could be a fluke that caused this error though.

> Write error rate 100 0 000000000000h

Ussually this will be 0 unless the servo information is screwed up. When a drive writes to a sector, it cannot tell if it has been written correctly. (but ECC will fix these errors anyway unless the sector is bad)

> Temperature 42 0 00000000002Ah OC SP

Drive seems to be at 42C, which is perfectly fine. Some drives may appear hotter then other models, simply because of the location of the thermister/thermo-diode.

> Start/stop count 100 20 00000000000Dh OC EC SP
> Power-on time count 95 0 00000000138Dh OC EC SP
> Device power cycle count 100 20 0000000001E9h OC EC SP

These are just informational.

> Attribute 202 100 0 000000000000h OC EC SP

No idea what this is

Ok, from reading the smart values, and given your syptoms, the most likely problems are:

1) Memory timing is too aggressive or bad memory, resulting in occasional corruption that you see on the drive at a later date.
2) The cable may be bad.

I suggest you try:

1) Replacing the IDE cable your seagate hard drive is using for another one.
2) Run memtest86 for several hours, to make sure your memory/cpu/motherboard subsystem isn't causing your corruption problems. I have seen this happen before, someone blames their hard drive for corruption when it is really the memory that is problem.
3) Run winbench 99 STR test to make sure your drive is performing properly (post a picture of the STR graph if you are unsure of how it should run)

I wouldn't worry too much about the occasional click yet, it may just be thermal recalibration. I also can't rule out the drive is bad, but it seems unlikely.

[Kurt]

A little Prime95 torture test too...

[Jon P. Inghram]

Turn the acoustic management off.