Uncorrectable Error on IDE, significant accumulation
Hi there.
I have a probably unusual question which is mainly directed to the linux-kernel IDE driver developers due to their experience with IDE disks.
I have a PC @home, which accumulates HDDs in it which die.
This Weekend my 120GB Maxtor begun to die with Uncorrectable Errors. I thought "OK, another damn Quality Hard Drive" But half a year ago I replaced a not so old 40GB MAxtor in it and before that a 20GB System Disk and a WD800JB has already 6 Secors remapped (smartctl -a). My friends call me the "Master of the HDD cemetary".
After that I realized, that 99% of my harddisk which die do this in that PC. I have a K7S5A Mainboard in it with SiS chipset. Kernel IDE Driver sis5513 is compiled in, UDMA switched on.
My question is, can a subtile Hardware error in the Mainboard exist so that HDDs is written bullshit to into some sektors which is discovered days or months later when reading/writing there?
Or can the Kernel provocate (in Hardware!) Uncorrectable errors with software?
I used Kernel 2.4.20-acX and since 2.5.69-mmX this one.
Or do I have simply bad luck in buying HDDs?
They are mounted good, the Wires are accurately folded together, 1" air above and under each disk, additional case fan...
It is my home server for mail/news/print/nfs and used as public ftp at lan-partys 3 or 4 times a year. For this purpose (pub ftp) there are two disks of three used with an LVM2 around all three. Statistically after each lan-party one drive dies. Not immedieately though.
Any opinions? Similair Experiences? Software? Hardware? Bad Luck?
Regards, Konstantin Kletschke
Signature · 5 lines
2.6.0-test1-mm2 Konstantin Kletschke <[email protected]>, <[email protected]> GPG KeyID EF62FCEF Fingerprint: 13C9 B16B 9844 EC15 CC2E A080 1E69 3FDA EF62 FCEF keulator.homelinux.org up 3:45, 20 users
Konstantin Kletschke wrote the following on 10/20/2003 03:27 PM :
Quoted 14 lines
Hi there.
I have a probably unusual question which is mainly directed to the linux-kernel IDE driver developers due to their experience with IDE disks.
I have a PC @home, which accumulates HDDs in it which die.
This Weekend my 120GB Maxtor begun to die with Uncorrectable Errors. I thought "OK, another damn Quality Hard Drive" But half a year ago I replaced a not so old 40GB MAxtor in it and before that a 20GB System Disk and a WD800JB has already 6 Secors remapped (smartctl -a). My friends call me the "Master of the HDD cemetary".
Quoted 5 lines
After that I realized, that 99% of my harddisk which die do this in that PC. I have a K7S5A Mainboard in it with SiS chipset. Kernel IDE Driver sis5513 is compiled in, UDMA switched on.
When did you purchase it ? Better : can you find the revision (black on white sticker glued in the corner of the motherboard near the last PCI slot) ?
There were electrical defects with the earliest K7S5A revisions (I returned one myself). Rev 0, 1, 2 and 3 are affected, I don't remember for 4-6 but IIRC rev 7 and above aren't affected : I have a rev 5 or 7 at home (don't remember which one it is) and it works perfectly.
LB.
KK> Any opinions? Similair Experiences? Software? Hardware? Bad Luck?
In my home computer there is a 6th hdd in 3 years. All didn't die but most did. On culprit was probably the power supply - I measured its voltage dropping significantly on some cdrom operations.
Signature · 1 line
Meelis Roos
Quoted 2 lines · Lionel Bouton
When did you purchase it ? Better : can you find the revision (black on
IIRC it is revision 3 or older. IIRC I downloaded BIOS files for revision 3.x or older, 1.3?
Quoted 4 lines
There were electrical defects with the earliest K7S5A revisions (I returned one myself). Rev 0, 1, 2 and 3 are affected, I don't remember for 4-6 but IIRC rev 7 and above aren't affected : I have a rev 5 or 7 at home (don't remember which one it is) and it works perfectly.
Hm, I have the Board Revision whih also remembers its BIOS Settings once after a while (once per month or so). Does that also belong to "electrical defects"?
So should I buy a K7S5A Pro now. It eats my old school SDRAM, thats why...
Konsti
Signature · 5 lines
2.6.0-test1-mm2 Konstantin Kletschke <[email protected]>, <[email protected]> GPG KeyID EF62FCEF Fingerprint: 13C9 B16B 9844 EC15 CC2E A080 1E69 3FDA EF62 FCEF keulator.homelinux.org up 27 min,
Quoted 3 lines · Meelis Roos
In my home computer there is a 6th hdd in 3 years. All didn't die but most did. On culprit was probably the power supply - I measured its voltage dropping significantly on some cdrom operations.
And do after this modification last longer?
Hm, I use the cdrom very seldom in this PC, as Lionel Bouton pointed out some K7S5As had electrical failures... I thought about PowerSuplly already, threw this though away because there is only
The Motherboard
cdrom (hardly used)
512MB RAM
3 HDDs
old intel AGP Graphics Card, never used.
connected.
Can a bad Powersupply or weak mainboard create Uncorrectable Errors on HDDs? Again only a question to experience of this community...
So should I buy a new board or a new Powersupply...
Konsti
Signature · 5 lines
2.6.0-test1-mm2 Konstantin Kletschke <[email protected]>, <[email protected]> GPG KeyID EF62FCEF Fingerprint: 13C9 B16B 9844 EC15 CC2E A080 1E69 3FDA EF62 FCEF keulator.homelinux.org up 32 min,
Konstantin Kletschke wrote the following on 10/20/2003 04:48 PM :
Quoted 6 lines
Hm, I have the Board Revision whih also remembers its BIOS Settings once after a while (once per month or so). Does that also belong to "electrical defects"?
Maybe... do you mean your CMOS settings are periodically reset ? In this case it usually is simply a worn-out battery.
Quoted 4 lines
So should I buy a K7S5A Pro now. It eats my old school SDRAM, thats why...
K7S5A Pro are dirt cheap these days, so it might be a good idea to buy a new one.
For the record, reallocated sectors are quite common with modern disks, I had some (3-4) on a one-year old 120Go Maxtor but the drive worked perfectly. You should install the smartmontools and run smartd configured to send you an e-mail each time a new defect is found. You have then one more opportunity to anticipate a drive going to fail by backuping and restoring on another drive.
Best regards,
Signature · 7 lines
Lionel Bouton - inet6 --------------------------------------------------------------------- o Siege social: 51, rue de Verdun - 92158 Suresnes / _ __ _ Acces Bureaux: 33 rue Benoit Malon - 92150 Suresnes / /\ /_ / /_ France \/ \/_ / /_/ Tel. +33 (0) 1 41 44 85 36 Inetsys S.A. Fax +33 (0) 1 46 97 20 10
Konstantin Kletschke wrote the following on 10/20/2003 04:53 PM :
Quoted 4 lines
Can a bad Powersupply or weak mainboard create Uncorrectable Errors on HDDs? Again only a question to experience of this community...
It certainly can temporarily (under load).
Try offloading some power strain by removing some peripherals (CD-ROM, non-mandatory disk drive) and see if it solves your problem.
I might be mistaken (don't know the exact behavior of drive electronics) but it seems unlikely that a bad PSU with underrated voltage could damage a drive (overrated voltage is another matter). Usually under spec PSUs fail to produce enough juice under load and the system simply becomes unstable.
Best regards,
Signature · 7 lines
Lionel Bouton - inet6 --------------------------------------------------------------------- o Siege social: 51, rue de Verdun - 92158 Suresnes / _ __ _ Acces Bureaux: 33 rue Benoit Malon - 92150 Suresnes / /\ /_ / /_ France \/ \/_ / /_/ Tel. +33 (0) 1 41 44 85 36 Inetsys S.A. Fax +33 (0) 1 46 97 20 10
Quoted 2 lines · Meelis Roos
Bad power supply can breaks disks like this. Bad mainborad or ide
OK, nice to know.
Quoted 1 line
cabling usually gives just CRC errors (like Abit KT7).
It is definately hardware.
Quoted 3 lines
Quoted 1 line
So should I buy a new board or a new Powersupply...
Test it with a tester first, I would suggest.
You wont get fast transient power leakage with conventional measure-tools...
You know what? I exchange Board AND PSU, I am too old for these copy actions with "UncorrectableError" all over my screen...
One of these devices causes my HDDs break definately if I am seeing that right here...
(my english becomes german, *lol*)
Konsti
Signature · 5 lines
2.6.0-test1-mm2 Konstantin Kletschke <[email protected]>, <[email protected]> GPG KeyID EF62FCEF Fingerprint: 13C9 B16B 9844 EC15 CC2E A080 1E69 3FDA EF62 FCEF keulator.homelinux.org up 56 min,
Quoted 3 lines · Lionel Bouton
Try offloading some power strain by removing some peripherals (CD-ROM, non-mandatory disk drive) and see if it solves your problem.
The risk ist too high, may be the PSU has a dry capacitor or whatsoever and the next drive breakes in two month with cdrom disabled. Because, there is only a CDROM Drive which could be disabled. No other components, no fat Graphics...
Quoted 2 lines
PSUs fail to produce enough juice under load and the system simply becomes unstable.
Well, the computer is most stable, it freezes sometimes, but I don't know if that was at the last lanparty (April) or thtereafter.
But as said, I am "convinced" and head to the store now. The machine is NOT relieable to me anymore...
Konsti
Signature · 5 lines
2.6.0-test1-mm2 Konstantin Kletschke <[email protected]>, <[email protected]> GPG KeyID EF62FCEF Fingerprint: 13C9 B16B 9844 EC15 CC2E A080 1E69 3FDA EF62 FCEF keulator.homelinux.org up 1:19, 6 users
Quoted 3 lines · Lionel Bouton
Maybe... do you mean your CMOS settings are periodically reset ? In this case it usually is simply a worn-out battery.
No, I replaced it already, it is also a known K7S5A bug, once a month it looses CMOS. BTW, near the last PCI Slot was a little square Button labeled with a "5".
Quoted 2 lines
K7S5A Pro are dirt cheap these days, so it might be a good idea to buy a new one.
Thats it, bought Board and PSU, for 55EUR...
Quoted 2 lines
perfectly. You should install the smartmontools and run smartd configured to send you an e-mail each time a new defect is found. You
Yes, absolutely...
Konsti
Signature · 5 lines
2.6.0-test6-mm4 Konstantin Kletschke <[email protected]>, <[email protected]> GPG KeyID EF62FCEF Fingerprint: 13C9 B16B 9844 EC15 CC2E A080 1E69 3FDA EF62 FCEF keulator.homelinux.org up 4:06, 8 users
Quoted 9 lines · Lionel Bouton
Konstantin Kletschke wrote the following on 10/20/2003 04:53 PM :
Quoted 4 lines
Can a bad Powersupply or weak mainboard create Uncorrectable Errors on HDDs? Again only a question to experience of this community...
It certainly can temporarily (under load).
Quoted 8 lines
Try offloading some power strain by removing some peripherals (CD-ROM, non-mandatory disk drive) and see if it solves your problem.
I might be mistaken (don't know the exact behavior of drive electronics) but it seems unlikely that a bad PSU with underrated voltage could damage a drive (overrated voltage is another matter). Usually under spec PSUs fail to produce enough juice under load and the system simply becomes unstable.
I don't know the specific drive electronics either. It has also been many years since i studied the electrical side of this stuff.
Undervolt, or voltage sag under load, is worse than overvolt. Overvolt can be caught with fuses or voltage regulators neither of which can do aught with an undervolt. If overvolt does get through things run just a bit warmer.
Undervolt on the other hand can cause VCC to drop to the level where it can no longer represent a logic high value but is instead in the range where it is neither 0 nor 1 and logic gates start spending too much time in the transition range where they act like amplifiers. When that happens not only do you get data corruption but things start heating up big-time and the current drain goes way up.
Signature · 5 lines
________________________________________________________________ J.W. Schultz Pegasystems Technologies email address: [email protected] Remember Cernan and Schmitt
Quoted 5 lines · Konstantin Kletschke
Quoted 3 lines
K7S5A Pro are dirt cheap these days, so it might be a good idea to buy a new one.
Thats it, bought Board and PSU, for 55EUR...
While at this point...
The new K7S5A Pro behaves strange.
When lilo comes up, it gets keyboard input containing of 4-6 lines "t:t:t:t:t:t:t:t:t:t:t:t:"...
When hitting backspace whole line gets cleared, enter boots default then. WTF?
even with no keyboard plugged in. My first thought was disabling "usb-keyboard support for dos" but... only a usb printer, ethernet and serial modem are plugged in...
Regards, Konsti
Signature · 5 lines
2.6.0-test6-mm4 Konstantin Kletschke <[email protected]>, <[email protected]> GPG KeyID EF62FCEF Fingerprint: 13C9 B16B 9844 EC15 CC2E A080 1E69 3FDA EF62 FCEF keulator.homelinux.org up 5:15, 7 users
Quoted 9 lines · Konstantin Kletschke
The new K7S5A Pro behaves strange.
When lilo comes up, it gets keyboard input containing of 4-6 lines "t:t:t:t:t:t:t:t:t:t:t:t:"...
When hitting backspace whole line gets cleared, enter boots default then. WTF?
even with no keyboard plugged in. My first thought was disabling "usb-keyboard support for dos" but... only a usb printer, ethernet and serial modem are plugged in...
Kernel configuration optins aren't going to affect LILO unless it's some strange data being left over from before reboot. Complete guess, but could it be some kind of BIOS test data that isn't getting cleared?
Signature · 1 line
Ian.
Quoted 11 lines · Ian Hastie
Quoted 9 lines · Konstantin Kletschke
The new K7S5A Pro behaves strange.
When lilo comes up, it gets keyboard input containing of 4-6 lines "t:t:t:t:t:t:t:t:t:t:t:t:"...
When hitting backspace whole line gets cleared, enter boots default then. WTF?
even with no keyboard plugged in. My first thought was disabling "usb-keyboard support for dos" but... only a usb printer, ethernet and serial modem are plugged in...
Quoted 1 line
Kernel configuration optins aren't going to affect LILO
Oh well. *8} It's that funny thing called the BIOS isn't it. Doh!!
: USB Function for DOS
: Enable this item if you plan to use the USB ports on this mainboard in a DOS : environment.
Well assuming it implies some kind of 16bit related access mode then I suppose it should be disabled anyway. However I didn't see a specific "usb-keyboard support for dos" in the manual. What happens if you disable "USB Function Support" completely?
Signature · 1 line
Ian.
Quoted 9 lines · Ian Hastie
: USB Function for DOS
: Enable this item if you plan to use the USB ports on this mainboard in a DOS : environment.Well assuming it implies some kind of 16bit related access mode then I suppose it should be disabled anyway. However I didn't see a specific "usb-keyboard support for dos" in the manual. What happens if you disable "USB Function Support" completely?
I meant your quoted excerpt. I did not remember the correct spelling of the option, remembered only something like "USB function for dos".
I thought also it must be some garbage left from BIOS or so, but when using arrow keys to select another item, lilo puts something like "2.6.0-test6-mm4mm4mm4mm4mm4mm4mm4mm4mm4mm4mm4mm4mm4mm4mm4mm4mm4"... onto the screen. One backspace erases whole 5 lines.
I flashed BIOS cause I thought ECS might have fixed it (if it is its fault) from January version to august version, same error.
I was fed up now and installed grub, which works like a charme.
Well, completely nokernel issue and I should write a bug report to the lilo folks...
Or is it a BIOS error still which grub copes around with?
Regards, Konsti
Signature · 5 lines
2.6.0-test6-mm4 Konstantin Kletschke <[email protected]>, <[email protected]> GPG KeyID EF62FCEF Fingerprint: 13C9 B16B 9844 EC15 CC2E A080 1E69 3FDA EF62 FCEF keulator.homelinux.org up 2:46, 4 users