opensever cmd=0x000000C8 on fixed disk
tully
different and badtrk does not seem to find anything. What does the
cmd C8 used for? do we need a new one here?
example below:
wd0: Error on fixed disk dev 1/42, block=8049847, cmd=0x000000C8
status=0x00000040, LBA sector=16792639...
Bela Lubkin
C8 is the "READ using DMA" command. These are read errors on the disk.
When you say the blocks are "usually" different, what do you mean? Have
you actually seen some where the block numbers were the same?
Modern drives do auto-sparing; that is, if they find a bad sector, they
replace it for you without having to bother the operating system (most
of the time). There are a few different scenarios:
- Bad sector detected on a write: the drive can silently spare it out,
OS never needs to know a thing.
- Bad sector detected on a read: the drive will attempt many times to
read the data, seeking from different sides, trying different head
current strengths, whatever the particular drive is designed to do. If
it eventually reads successfully, it can assign a spare sector, write
the data there, pass it back to the OS without informing it about the
problem.
- Bad sector detected on a read, and data was unrecoverable: then the
drive _has_ to tell the OS about the problem. It will probably also
replace that sector with a spare (but the spare will contain garbage, or
[hopefully] all 0's or some other such intentional "blank" pattern).
You're seeing this problem on reads, which makes sense. The fact that
it's actually coming through to the OS means that the sectors it's
finding are totally unreadable.
If a `badtrk` scan finds nothing, either there are no other bad sectors
at the time, or (if there are any) the drive is retrying and eventually
getting a good read.
And then if you get more bad sectors later, whatever is wrong is an
ongoing deterioration; as if, for instance, something has come loose
inside the drive and is bouncing around in there, scratching more and
more places on the media.
There's one thing to check before replacing the drive. It's using Ultra
DMA. Make sure that you have it attached with a proper 80-pin UDMA IDE
cable. IDE connectors are 40 pins wide. UDMA drives require an 80-pin
cable where the separating pins are ground lines. Operating a UDMA
drive with a 40-pin cable can cause all sorts of bizarre behavior.
If that isn't the problem, you need to replace the drive. Do not
overwrite your last few backups. Make a new full backup. Replace the
drive, restore onto it. The older backups are important because you may
later notice that some files are corrupt; you will want several backups
from various time ranges to have the best possible chance of being able
to find a good copy.
>Bela<
tully
Yes, I have seen the same block repeat, here is a snipet
Oct 11 23:33:53 wd0: Error on fixed disk dev 1/42, block=8049842,
Oct 11 23:33:53 wd0: Error on fixed disk dev 1/42, block=8049843,
Oct 11 23:41:34 wd0: Error on fixed disk dev 1/42, block=8049842,
Oct 12 14:52:54 wd0: Error on fixed disk dev 1/42, block=6902005,
Oct 12 14:52:54 wd0: Error on fixed disk dev 1/42, block=6902006,
Oct 12 15:30:37 wd0: Error on fixed disk dev 1/42, block=8049842,
Oct 12 15:30:37 wd0: Error on fixed disk dev 1/42, block=8049843,
Oct 12 22:52:21 wd0: Error on fixed disk dev 1/42, block=6902005,
Oct 12 22:52:21 wd0: Error on fixed disk dev 1/42, block=6902006,
Oct 12 22:57:20 wd0: Error on fixed disk dev 1/42, block=6902005,
i will not replace the drive at this time (its a mirror backup system)
but i DID replace the 40 pin ide with an 80 pin cable. I was under the
assumption that i could get away with and old 40 pin cable since this
is an old confuser (before UDMA 66 days). Oddly enough the second drive
in this confuser has not had any errors and it still has a 40 pin cable, but
that one is connected to a cd-rom drive as well..... hmm, time will tell.
<chris>
> When you say the blocks are "usually" different, what do you mean? Have
> you actually seen some where the block numbers were the same?
Bela Lubkin
> Yes, I have seen the same block repeat, here is a snipet
> Oct 11 23:33:53 wd0: Error on fixed disk dev 1/42, block=8049842,
> Oct 11 23:33:53 wd0: Error on fixed disk dev 1/42, block=8049843,
> Oct 11 23:41:34 wd0: Error on fixed disk dev 1/42, block=8049842,
> Oct 12 14:52:54 wd0: Error on fixed disk dev 1/42, block=6902005,
> Oct 12 14:52:54 wd0: Error on fixed disk dev 1/42, block=6902006,
> Oct 12 15:30:37 wd0: Error on fixed disk dev 1/42, block=8049842,
> Oct 12 15:30:37 wd0: Error on fixed disk dev 1/42, block=8049843,
> Oct 12 22:52:21 wd0: Error on fixed disk dev 1/42, block=6902005,
> Oct 12 22:52:21 wd0: Error on fixed disk dev 1/42, block=6902006,
> Oct 12 22:57:20 wd0: Error on fixed disk dev 1/42, block=6902005,
>
> i will not replace the drive at this time (its a mirror backup system)
> but i DID replace the 40 pin ide with an 80 pin cable. I was under the
> assumption that i could get away with and old 40 pin cable since this
> is an old confuser (before UDMA 66 days). Oddly enough the second drive
> in this confuser has not had any errors and it still has a 40 pin cable, but
> that one is connected to a cd-rom drive as well..... hmm, time will tell.
If it is only using UDMA33, a 40-pin cable is supposed to be adequate.
An 80-pin cable does give better signal fidelity, so maybe it will help.
It's also possible the old cable was just bad (recently damaged,
somehow), and that replacing it will help even if you only use a 40-pin
replacement.
But... most likely the drive itself is going bad.
When a drive auto-spares a bad block, the replacement block comes out of
some sort of pool the drive reserved for itself at low-level format time
(probably done by the manufacturer). If the drive is constantly finding
more and more bad blocks, that pool will eventually run out and the
drive will have no choice but to report errors to the operating system.
That's another scenario where you would get repeated instances of the
same block. It basically means your drive is now a high-tech doorstop.
The cost of a replacement drive must be far less than the cost of
dealing with it if it finishes dying. Replace the drive.
>Bela<