File coruption on Raid array with IBM GXP75's!

[Technoid]

Well....

I have concluded that the Coruption does not occur on the boot drive on the 686B...

Have formatted the array and will conduct tests...

BTW the old Memory chipd was cl 2-3-3 and the new ones are 2-2-2.

The IBM suport mentioned vibrations...
Could vibrations cause datacoruption without sector errors?

Still waiting for a answer for that....

Doc. Mordrid,

How does the "utilize the PCI bus" Slider in promise utility work?

Now its set to "low"

[Dr Mordrid]

It adjusts the PCI utilization of the card. Low = low PCI use, High = high PCI use.

I keep mine at high now that the Santa Cruz is handling the audio at very low latencies.

Dr. Mordrid

[The PIT]

I would look at the memory timings as well and slow them down.
Dr Mordrid did you have probs with a live before.

[Dr Mordrid]

Would vibration cause problems with a HD? I'd be suprised if it didn't. The nature of these problems would be up for grabs.

It could manifest many ways depending on what's causing the vibration and/or what's affected by the harmonics. Read errors would be high up on the probability list. The data generated this way could very well be interpeted by the firmware as data corruption.

Did I have a problem with the Live before?

hehehe....search for my name & SBLive and see how many hits you get

I've had problems with that card since it came out. I bought a few to install in my lab and not one system would work right with it. All had some form of bus contention with the capture card(s), crackling or high rates of dropped frames (or all of the above).

With the exception of throwing one in for testing now and then those things have sat on the shelf since 2 weeks post purchase.

Dr. Mordrid

[The PIT]

Well it seems that crackling was due mainly to due to Microshaft leaving the debugging code in. Thats the only time I've suffered it and switching it off seems to fix 97% of peoples problems. The other two percent it's becuase they got the one of the various volume controls at 100% and there just picking up the noise from the hard drives spinning up. After all the inside of a computer is going to throw a lot of interferance around. The other 1% seem to have a genuine problem.
This is ignoring the freezing and some other probs the sblive can throw up.

[Wombat]

Vibration might cause some problems, but you can check out the specs on your drives. A common spec is how many G's a drive will survive. I think the 75GXP's are good for 1G while operating, and 10G's when unpowered.

[The PIT]

How much force is 1G? I doubt that normal vibration (I.E fans and bashing about plating UT) would effect todays hard disks. Put it on top of Deisal engine may upset your hard drive but I've knocked a running computer over and didn't get one bad sector.
Against that I've dropped a hard disk one foot onto the floor and it hit me half way down on during the descent and it killed the drive.

[Wombat]

Not to be pedantic, but a G isn't a unit of force. It's a unit of acceleration. The problem isn't how far or fast the drive is going, but how fast you stop it.

I'd say you were lucky when knocking over the computer didn't corrupt the HD. Maybe you weren't doing any disk accesses, and so the heads were tucked away somewhere safe. If the drive had been clicking away when the computer fell, I think you would have lost a sector or two.

If you think about it for a second, I don't know of any computer component that gets <B>suspended</B> in more foam than a hard drive.

[Wombat]

Incidentally, what kind of HD was it that you dropped on the floor? Maxtor?

There was a problem with Maxtor drives for a while where the heads would lock up, rendering the drive useless. If you REALLY needed the information on the drive, the solution was to take out the drive, hold it about 4 inches above a hard surface, and drop it, circuit-side down. That would jar the heads loose, and you should have been able to read the data one last time (but the heads would be damaging the disk constantly after this, so it wasn't like you fixed the drive).

[The PIT]

Yup but doesn't the damage occur when theres an impact. Perhaps the cider got hold of me to much now but you could probably send a laptop up with the space shuttle and back and it would still work and the forces would be greater than 1G.
If I'm wrong I'm going to say Hick Hick.

[Wombat]

1. Laptops are built to much higher tolerances.
2. They don't use the hard drives while the shuttle is taking off. Laptop HD's are good for a 100G change if they're in storage.
3. Space shuttles are only high-G during blast off. Even then, with pressure suits, a human can only take so much. I bet a fighter pilot routinely hits higher Gs in a sharp bank than an astronaut does. And pilots usually blank out between 10-15G's if I remember right. I forget if blacking out or redding out happens at different Gs

You got me curious enough to actually investigate:

<B>The IBM 60gxp</B>
Operating: 55G/2ms half-sine; .56G(rms) random vibration
Non-Operating: 400G/2ms half-sine; 1.04G(rms) random vibration
<B>The IBM 48gh</B> (laptop drive)
Operating: 150G/2ms (half sine); 0.67G(rms) random vibration
Non-operating: 700G/1ms, 3G(rms) random

[Technoid]

Update

Now I have tested almost everything!!

Set the drives to UDMA2, check

Set the memory to 3-3-3 100Mhz interleaving disabled!, check

Set the cpu to 1000mhz instead of 1333 by lovering bus from 266 to 200 (133 -->100), check

Disabled PCI bursting, and alot of other things in bios....

Disabled the SBLive.

All of these things and the system should be slower than when I had the original drives...

(and then it was a Duron 700 overclocked to 930)

Dit it fix it?

NO!

Am I iritated?

Yes!

Am I Surprised?

Not realy...

What will I do now?

Check the drives as single drive arrays,

Get the samsungs and test if they do behave diferently when my machine don't have all the above mentioned slow downs.

Mayby get a sis board and a real fasttrack?

[LAMFDTK]

If you think about it for a second, I don't know of any computer component that gets suspended in more foam than a hard drive

I just received a replaced drive today and it was in a weird plastic shell. No foam! Im saving the packing this time.

I had the array in 5 different mobos for benching. I think I got a brand new drive. its a june 01. 30.7GB Thailand. Is there a way to tell if its reburber?

[Technoid]

Yep!

It has an orange sticker on it saying "working serviced part"

I have two with those!

I'm happy, so happy, so happy, blöööööö!

[LAMFDTK]

No sticker here I guess I lucked out. I ran the IBM tests on the "other" HD and it passed. So I guess I am ready for a raid ghost. then a total rebuild of apps. I forgot to ghost the setup I had on this box. That will take me a week...

Back to 1995------>WD ?