Write Error Count keeps going up on Samsung 2TB

T

TheSpoon

Limp Gawd
Joined
Jun 6, 2004
Messages
431
I have 2 HD204UI drives in my system in RAID-0. During heavy disk activity, the write error count (raw value) goes up. Now it's at 848 on one drive and 1055 on another. Bad drives? Nothing to worry about?

Switching SATA cables had no effect.

Edit: Gah, I think this is happening because of squished SATA cables (by the side panel). I opened the panel and the errors stopped. Cable management is dangerous.
 
Last edited:
Holy shit.. I did some more investigating, and the problem does seem to have to do with cables. But with power cables, not data cables. My two HDs are connected with this cable:



I started running a 64GB dd write. Whenever I push on that loop in the cable with my finger (simulating the back panel closing) my write speeds drop from ~200MB/s to nearly zero, and the number of write errors increases. Wacky!

Now I hope I'll be able to rearrange them to avoid this...
 
It seems to have worked! No more write errors, even with closed panel. I did this (sorry for blur):



Well it's 2738 write errors later but I learned a lesson. Check all cabling before declaring a drive defective.
 
Crazy, that's one of those things that if someone asked me to check the power cable I'd say duhh of course I checked it. :p
 
vsboxerboy said:
Crazy, that's one of those things that if someone asked me to check the power cable I'd say duhh of course I checked it. :p
Click to expand...

Yeah I know I feel incredibly lucky to have discovered this. It would have sucked ass to RMA the drives only to have the issue persist.
 
The write errors returned! Though the cable doesn't seem overly pressed anymore. So I don't know what's wrong with em. But now I'm going to unlearn the lesson about checking cables.
 
Maybe the pulling of the cable has knocked somethign loose like a connector or the board isn't in place properly. That MIGHT explain why pressing the cable did something ??
 
Yeah that's possible. Though it's odd that both drives have write errors because the pressures on their respective connectors are quite different. They'd have to be pretty fragile. Ugh.
 
hm, that is weird - maybe that whole power cable arm is fubar - can you try with a different set of connectors?
 
Oh I did. They're now on a different cable than they were on originally. Oooh I just got a UDMA CRC Error too!
 
This gets stranger. I opened the panel again, and the write errors went away. But the closed panel didn't press on the connectors at all anymore.

So I did this experiment:

Close the panel at the bottom, but leave it open at the top. This way the panel doesn't touch any cables at all. Join the top of the panel to the top of the case with metal scissors (the panel and the case are both steel), like so:


... every time I do that, my write speed goes to near zero and write errors climb. I remove the scissors, write speed is restored. WHAT THE FUCK.
 
That's an interesting problem (not for you...). Anything touching the case that shouldnt? Motherboard standoff or anything else inside or outside?
 
Well the motherboard is ON standoffs of course, but there are no loose standoffs anywhere or anything like that... In one spot, I tied all of my SATA data cables with a cable tie that is metal wire covered in plastic. I hooked it to a metal hooking thingy on the motherboard tray.

There are quite a few reports online of write errors with this particular drive. I realize now that when I was pushing that loop with my finger, I was pushing the power cable closer to the drive. It's like.... my power cable induced a sub-harmonic fluctuation in the subspace resonance field encasing the drive's platters, or something.
 
Thats really possible we used to have power cables interfere with data cables on plant floors. The drive and/or shielding could be terrible.
 
Falcon_CMH said:
Thats really possible we used to have power cables interfere with data cables on plant floors. The drive and/or shielding could be terrible.
Click to expand...

Oh, that's interesting to hear. Is there a chance it really is the PSU's fault then? It's an HX1000, which is a pretty popular choice around here I think.

I also have another HD in there -- a 320GB WD, which has no problem. Though older HDs use different storage technologies. So I don't know how much that matters.
 
Well I would say what your desribing seems like an issue with the drive not the PSU. Try relocating it in your case away from all power but its own power connection
 
Yeah I was thinking of relocating them to the 5.25" bays. I'll be out of the country next week and I'll do that when I get back.
 
Falcon_CMH said:
Thats really possible we used to have power cables interfere with data cables on plant floors. The drive and/or shielding could be terrible.
Click to expand...

Except the computer power supply is using DC. The hard drive should present a fairly constant load during operation, so not much change in current.

Faraday's law tells us that the induced voltage is equal to the time derivative of the magnetic flux - unless the cable is being moved constantly, or the system is using AC power (or some weird switched DC), the induced EMF is effectively 0.

Shielding would prevent against electrostatic effects, however, I wouldn't expect the SATA data lines to radiate very far (they're surrounded by grounds, which acts as a shield)so shielding shouldn't be a factor at all.

OP - sounds like you have some weird grounding things going on. See if you can get some nylon hardware to mount the drives. If possible, take them completely out of the case and put them on the table, then see if the errors continue.
 
cymon said:
OP - sounds like you have some weird grounding things going on. See if you can get some nylon hardware to mount the drives. If possible, take them completely out of the case and put them on the table, then see if the errors continue.
Click to expand...

Well the errors go away (either entirely or almost entirely) when I just leave the case door open. The drives were also working fine in my previous case -- an Antec P182. So there is definitely something peculiar about this case (a Silverstone RV02). Or at least the combination of this case and these drives. How do you track down weird grounding things? I have a multimeter, which allows me to test resistance. There is very little resistance between the drive's outer shell and my motherboard tray. So it would appear the drives are grounded. The annoying thing is that the symptoms also vary somewhat with time. Right now my case is closed and I'm getting very few write errors (though not zero) -- not enough to slow down write speeds.
 
i say there's a short somewhere. you complete the circuit when you close the door.

when these things happen, I just take the motherboard and everything out and check for shorts. It's not very surgical but it works.
 
i swear im going through the same thing... lol

except it's where i close my door and my system fails to boot. i get stuck at the OS splash screen. perhaps shorting is an issue. im gonna go thru a thorough debug.

its either the door shorting something or my PSU cables not doign well while bending. I want to say it has to do with bending because I took my hand and pushed them in like a case door would and I have boot issues. ugh.
 
Last edited:
dmo580 said:
i swear im going through the same thing... lol

except it's where i close my door and my system fails to boot. i get stuck at the OS splash screen. perhaps shorting is an issue. im gonna go thru a thorough debug.

its either the door shorting something or my PSU cables not doign well while bending. I want to say it has to do with bending because I took my hand and pushed them in like a case door would and I have boot issues. ugh.
Click to expand...

What case and HDs?
 
You must log in or register to reply here.
Back
Top