Puzzling HD problem

[ras2]

(Sorry about the subject; I couldn't think of a more descriptive one).

I have a K7S5A (rev. 3.1) motherboard with a 4gb. Quantum Bigfoot
(master) and a 30gb. IBM DTLA-307030 (slave) on the primary IDE-channel.
The IBM has started to behave rather oddly:

The first sign of problems were long pauses after disk reads in Windows;
in games, some sound would load and the game would stall for a while.
Quitting the game could take over five minutes and once it finally
finished, opening the Start menu could take ages. Reinstalling Windows
seemed to help for a while and then the same problem started happening
again. Under Linux, I started getting BadCRC S.M.A.R.T. errors from the
drive.

I finally figured out that turning DMA off would stop the errors from
happening and it worked well enough (if slowly) for a while in both
Windows and Linux.

Then the superblock on one of the ext2 partitions got ruined. It was
fixed easily enough and I didn't lose any files (the filenames of
some pictures just got clobbered) and the drive worked okay for some
days until it suddenly half-way disappeared after the computer had
been idle for a few hours (no reading or writing had taken place on
the drive in that time); the temperature monitor didn't get any readings
from it and I had problems getting a file from one of the FAT32
partitions.

On reboot, it turned out that the superblocks on two ext2 partitions
had evaporated and that none of the FAT32 partitions had any FSINFO
signatures (no files on the ext2 partitions got wrecked this time either,
but the names of files and directories in the root of one partition all
got clobbered).

I ran the IBM Disk Fitness Tool and got mixed results: With the Quick Test
I first got an "Excessive Shock" diagnosis and on the next two runs the
drive passed. With the Advanced Test, I got two "Bad Cable" results and
on the third run the drive passed. S.M.A.R.T. reported the drive as OK
(but then, aside from those errors I mentioned, it always says the drive
is OK).
So that wasn't very conclusive.

When I booted the system yesterday, the system reported the drive as
"YB]-\T\A=387838 8 8 8 8 8 8 8 8 8 8 8 8, ATA DISK drive"
(and none of the partitions could be mounted) but when I rebooted a few
minutes ago, it was reported correctly and the ext2 partitions can be
mounted without problems now (the FAT32 can't, but I haven't fixed the
FSINFO problem either).

At no point have I had any reports about bad blocks from fsck.

All these problems seemed to start shortly after I'd connected an old
Amiga HD on the secondary IDE connector for a short while (~5 minutes).
I don't know why that would cause them, but the timing seemed rather
suspicious.


What I'd like to know is if this sounds like a normal way for a drive
to die or if it perhaps could be the motherboard or the cable or something?
I know of the history of the GXP75 drives, but it seems like they usually
die rather more dramatically than this.


-R.

 

[Rod Speed]

(Sorry about the subject; I couldn't think of a more descriptive one).

I have a K7S5A (rev. 3.1) motherboard with a 4gb. Quantum Bigfoot
(master) and a 30gb. IBM DTLA-307030 (slave) on the primary
IDE-channel. The IBM has started to behave rather oddly:

The first sign of problems were long pauses after disk reads in Windows;
in games, some sound would load and the game would stall for a while.
Quitting the game could take over five minutes and once it finally
finished, opening the Start menu could take ages. Reinstalling Windows
seemed to help for a while and then the same problem started happening
again. Under Linux, I started getting BadCRC S.M.A.R.T. errors from the
drive.

I finally figured out that turning DMA off would stop the errors from
happening and it worked well enough (if slowly) for a while in both
Windows and Linux.

Then the superblock on one of the ext2 partitions got ruined. It was
fixed easily enough and I didn't lose any files (the filenames of
some pictures just got clobbered) and the drive worked okay for some
days until it suddenly half-way disappeared after the computer had
been idle for a few hours (no reading or writing had taken place on
the drive in that time); the temperature monitor didn't get any readings
from it and I had problems getting a file from one of the FAT32
partitions.

On reboot, it turned out that the superblocks on two ext2 partitions
had evaporated and that none of the FAT32 partitions had any FSINFO
signatures (no files on the ext2 partitions got wrecked this time either,
but the names of files and directories in the root of one partition all
got clobbered).

I ran the IBM Disk Fitness Tool and got mixed results: With the Quick Test
I first got an "Excessive Shock" diagnosis and on the next two runs the
drive passed. With the Advanced Test, I got two "Bad Cable" results and
on the third run the drive passed. S.M.A.R.T. reported the drive as OK
(but then, aside from those errors I mentioned, it always says the drive
is OK).
So that wasn't very conclusive.

When I booted the system yesterday, the system reported the drive as
"YB]-\T\A=387838 8 8 8 8 8 8 8 8 8 8 8 8, ATA DISK drive"
(and none of the partitions could be mounted) but when I rebooted a few
minutes ago, it was reported correctly and the ext2 partitions can be
mounted without problems now (the FAT32 can't, but I haven't fixed the
FSINFO problem either).

At no point have I had any reports about bad blocks from fsck.

All these problems seemed to start shortly after I'd connected an old
Amiga HD on the secondary IDE connector for a short while (~5 minutes).

You may have a flakey cable and trying that drive was enough
to disturb the bad connection in the cable enough to make the
problem visible. Thats usually due to one of the prongs that
pierce the ribbon cable having got bent over when one of the
connectors was crimped onto the cable. Try another cable.

It can also be due to a bad power connector too. The metal
tunnels the pins go into can open up over time and not make
a good connection for the power. You wouldnt normally
expect the exact symptoms you are getting tho and you
would normally be able to hear the drive spinning up and
down with a relatively noisy drive like that, so this is less likely.

I don't know why that would cause them,
but the timing seemed rather suspicious.

Yeah, its unlikely to be a coincidence.

What I'd like to know is if this sounds like a normal way for a drive to
die or if it perhaps could be the motherboard or the cable or something?

Most likely the cable.

I know of the history of the GXP75 drives, but it seems
like they usually die rather more dramatically than this.

Not always.

 

[ras2]

I have a K7S5A (rev. 3.1) motherboard with a 4gb. Quantum Bigfoot
(master) and a 30gb. IBM DTLA-307030 (slave) on the primary
IDE-channel. The IBM has started to behave rather oddly:

[...]
You may have a flakey cable and trying that drive was enough
to disturb the bad connection in the cable enough to make the
problem visible. Thats usually due to one of the prongs that
pierce the ribbon cable having got bent over when one of the
connectors was crimped onto the cable. Try another cable.

I didn't think I had any extra cables, but I found one on the bottom of
a box of spares:
3 Quick Tests and 2 Advanced Tests with the IBM DFT all passed the drive
and Linux doesn't complain anymore when I turn DMA on, so it seems like
there may be hope.
One odd thing is that I checked the MD5 sums of a 900+ and a 700+ mb.
file and it seemed to not even take half as long as it used to (before
the problems started).

Now I just need to figure out how to repair those FSINFO signatures for
the FAT32 partitions, so I can see if Windows works too.

It can also be due to a bad power connector too. The metal
tunnels the pins go into can open up over time and not make
a good connection for the power. You wouldnt normally
expect the exact symptoms you are getting tho and you
would normally be able to hear the drive spinning up and
down with a relatively noisy drive like that, so this is less likely.

yes, it sounds okay. Or, it sounds like it's always done, at least.


Thanks.


-R.

 

[Rod Speed]

I didn't think I had any extra cables, but I
found one on the bottom of a box of spares:

3 Quick Tests and 2 Advanced Tests with the IBM DFT
all passed the drive and Linux doesn't complain anymore
when I turn DMA on, so it seems like there may be hope.

Yeah, looks like the cable may well have been the problem.

The other problem with ribbon cables is that some
people do tend to just yank on the ribbon itself to get
the connector out of the drive connector and that isnt
very good for the crimped connector connection.

One odd thing is that I checked the MD5 sums of a 900+
and a 700+ mb. file and it seemed to not even take half
as long as it used to (before the problems started).

Most likely you were getting significant numbers of
CRC errors being retried with the old cable and not
anymore and thats why its noticeably faster now.

Now I just need to figure out how to repair those
FSINFO signatures for the FAT32 partitions, so
I can see if Windows works too.

Dunno, dont use FAT32 like that myself.