Problems with SATA

[asavidge]

I had two 120Gb Maxtor SATA hdd set up in a mirrored array. The RAID is
done in hardware on the motherboard (Abit KV8 Pro).

The other night, during a disk-heavy operation (un-raring a large
archive) the PC slowed, and then locked up hard. When I hit the reset
button the RAID BIOS kicked in and said there was a disk fault on the
drive on channel 0, giving me the option to refresh the mirror or
continue to boot.

First time round I continued to boot, which sort of worked but was
really really slow. I never waited long enough for the OS to come up, I
could see it would take hours.

Second time round, I rebooted, it showed the same RAID BIOS message and
I selected to refresh the mirror. It started the refresh, displayed a
progress bar and then just sat there, for an hour, with no progress.

I have tried both drives, singly (having broken the mirror) and
together, in both the SATA channels, and I get variously:

1) The slow boot
2) A much faster looking boot, that unfortunately stops just as the
Windows XP splash with the progress bar. When it does this, there's an
audible click (could be power, could be hdd) and then reboots again.

Thinking (hoping) it was the onboard SATA I went out and bought a PCI
SATA card, and get exactly the same results.

I find it really hard to believe that both hard-drives have died at the
same time, but I'm otherwise at a loss to explain why the results of
using a different SATA interface.

Any ideas?

Tomorrow I will borrow an IDE drive from work, reinstall Windows on
that, or maybe use a linux boot CD and see if I can access the SATA
drives from an OS on a different disk.

 

[sbb78247]

I had two 120Gb Maxtor SATA hdd set up in a mirrored array. The RAID
is done in hardware on the motherboard (Abit KV8 Pro).

The other night, during a disk-heavy operation (un-raring a large
archive) the PC slowed, and then locked up hard. When I hit the reset
button the RAID BIOS kicked in and said there was a disk fault on the
drive on channel 0, giving me the option to refresh the mirror or
continue to boot.

First time round I continued to boot, which sort of worked but was
really really slow. I never waited long enough for the OS to come up,
I could see it would take hours.

Second time round, I rebooted, it showed the same RAID BIOS message
and I selected to refresh the mirror. It started the refresh,
displayed a progress bar and then just sat there, for an hour, with
no progress.

I have tried both drives, singly (having broken the mirror) and
together, in both the SATA channels, and I get variously:

1) The slow boot
2) A much faster looking boot, that unfortunately stops just as the
Windows XP splash with the progress bar. When it does this, there's an
audible click (could be power, could be hdd) and then reboots again.

Thinking (hoping) it was the onboard SATA I went out and bought a PCI
SATA card, and get exactly the same results.

I find it really hard to believe that both hard-drives have died at
the same time, but I'm otherwise at a loss to explain why the results
of using a different SATA interface.

Any ideas?

if it's a mirror (raid1) then what ever gets ****ed up on one drive gets
borked on the other. remember????

 

[john_doe]

On 22 May 2006 12:54:35 -0700, (e-mail address removed) wrote:

They're only joking about that motherboard raid controller stuff.

If you want raid, you need to spend big bucks to do it right.

http://www.newegg.com/Product/Produ...Ntk=all&N=0&minPrice=&maxPrice=&Go.x=0&Go.y=0

 

[Hackworth]

I had two 120Gb Maxtor SATA hdd set up in a mirrored array. The RAID is
done in hardware on the motherboard (Abit KV8 Pro).

The other night, during a disk-heavy operation (un-raring a large
archive) the PC slowed, and then locked up hard. When I hit the reset
button the RAID BIOS kicked in and said there was a disk fault on the
drive on channel 0, giving me the option to refresh the mirror or
continue to boot.

First time round I continued to boot, which sort of worked but was
really really slow. I never waited long enough for the OS to come up, I
could see it would take hours.

Second time round, I rebooted, it showed the same RAID BIOS message and
I selected to refresh the mirror. It started the refresh, displayed a
progress bar and then just sat there, for an hour, with no progress.

I have tried both drives, singly (having broken the mirror) and
together, in both the SATA channels, and I get variously:

1) The slow boot
2) A much faster looking boot, that unfortunately stops just as the
Windows XP splash with the progress bar. When it does this, there's an
audible click (could be power, could be hdd) and then reboots again.

Thinking (hoping) it was the onboard SATA I went out and bought a PCI
SATA card, and get exactly the same results.

I find it really hard to believe that both hard-drives have died at the
same time, but I'm otherwise at a loss to explain why the results of
using a different SATA interface.

Any ideas?

Tomorrow I will borrow an IDE drive from work, reinstall Windows on
that, or maybe use a linux boot CD and see if I can access the SATA
drives from an OS on a different disk.

Regrettably, onboard RAID is just about as good as onboard sound and onboard
video. You need a "real" RAID controller if you're serious about RAID.
This one has all the bells and whistles and the price is right:

http://www.newegg.com/Product/Product.asp?Item=N82E16816102063

Higher-end motherboards sometimes have a built-in hardware RAID controller
(with a Promise chipset, for example) in addition to the usual two or four
standard SATA ports controlled by the motherboard chipset. Such a built-in
hardware RAID controller counts as a "real" RAID controller. An example is
shown in the pictures of this eBay auction:

http://cgi.ebay.com/ATHLON-64-3700-...yZ114192QQssPageNameZWDVWQQrdZ1QQcmdZViewItem

 

[asavidge]

Thanks for all the reccomendations. I do work with the real deal
hardware in my job, so I'm well aware that the stuff built in for home
use is little more than a joke, but I was at least hoping that given a
failure on one drive that it would give me enough leeway for some
recovery, seemed like a good idea at the time.

I have an IDE drive here, on which I'm going to reinstall XP and cross
my fingers that I can mount one of the SATA drives.

After that I guess it's back to DVD backups, or steal a proper NAS from
work.

Wish me luck.

 

[asavidge]

Some good news. With the IDE drive containing a clean install of XP one
of the disks functions perfectly as a D: and E: drive (I had a 30Gb
partition and a 130Gb partition, making 160Gb, not 120Gb as originally
stated.)

I'm stoked about this because at the worst I can get my photos, video
and mp3s onto DVD. At best, I'll be able to get the 30Gb partition
booting again, and I won't have to rebuild the entire OS back to where
I like it.

The other drive is weird... with it in makes the new IDE install boot
r-e-a-l-l-y slowly and CPU usage flashes regularly in a pattern of
2-100-2-88-2-88-2-88-2-88-0-0% you can see the regular sawtooth pattern
in the performance graph. When it's high, the mouse locks up, when it's
low it's fine. These are about 5 second intervals between each spike.
Every time the CPU spikes, the disk thrashes.

Leave it 30 minutes and it calms down, idles at 2% CPU and shows just
the 30Gb partition on the SATA drive in My Computer. As long as you
don't try and involve the 130Gb partition in any way (Disk management
shows it, but doing anything with it starts the spiking CPU again for
another 30 minutes.)

Given that one hard drive looks entirely recoverable my interest in the
second is purely scientific. It seems unlikely to me that the physical
media is damaged, given that one partition is fine on the drive. Does
that sound reasonable? I can't imagine I have a half-functioning
physical disk because of actual physical malfunction. It seems far more
likely to me that there's something so wrong with the logical layout
that it's causing Windows to poo itself.

Still, my confidence in h/w RAID is diminished, but I'm not sure how
much of my situation has been caused by the RAID and how much it's been
saved by it. Hard to tell really... maybe without it I'd be looking at
some gruelling data recovery in the 5 second gaps between disk
thrashing... who knows?

 

[JAD]

what are your virtual memory settings?

 

[David Maynard]

Some good news. With the IDE drive containing a clean install of XP one
of the disks functions perfectly as a D: and E: drive (I had a 30Gb
partition and a 130Gb partition, making 160Gb, not 120Gb as originally
stated.)

I'm stoked about this because at the worst I can get my photos, video
and mp3s onto DVD. At best, I'll be able to get the 30Gb partition
booting again, and I won't have to rebuild the entire OS back to where
I like it.

The other drive is weird... with it in makes the new IDE install boot
r-e-a-l-l-y slowly and CPU usage flashes regularly in a pattern of
2-100-2-88-2-88-2-88-2-88-0-0% you can see the regular sawtooth pattern
in the performance graph. When it's high, the mouse locks up, when it's
low it's fine. These are about 5 second intervals between each spike.
Every time the CPU spikes, the disk thrashes.

Leave it 30 minutes and it calms down, idles at 2% CPU and shows just
the 30Gb partition on the SATA drive in My Computer. As long as you
don't try and involve the 130Gb partition in any way (Disk management
shows it, but doing anything with it starts the spiking CPU again for
another 30 minutes.)

Given that one hard drive looks entirely recoverable my interest in the
second is purely scientific. It seems unlikely to me that the physical
media is damaged, given that one partition is fine on the drive. Does
that sound reasonable? I can't imagine I have a half-functioning
physical disk because of actual physical malfunction. It seems far more
likely to me that there's something so wrong with the logical layout
that it's causing Windows to poo itself.

Can't say for sure without analyzing it but your presumption that a
physical malfunction would automatically barf the whole drive is incorrect.
A physical media defect, for example, would affect only those locations
falling in the defective area.

Still, my confidence in h/w RAID is diminished, but I'm not sure how
much of my situation has been caused by the RAID and how much it's been
saved by it. Hard to tell really... maybe without it I'd be looking at
some gruelling data recovery in the 5 second gaps between disk
thrashing... who knows?

You left out another possibility: potentially made worse by user. In
particular, trying to sync the mirror before knowing what was faulty.

 

[asavidge]

Fair enough, but there was no way to find out, without shelling out for
a new drive there and then. The RAID BIOS gave precious little
information and knowing XP I was inclined to check cheaper options
first (like try to get something to boot and then chkdsk.)

 

[David Maynard]

Fair enough, but there was no way to find out, without shelling out for
a new drive there and then. The RAID BIOS gave precious little
information and knowing XP I was inclined to check cheaper options
first (like try to get something to boot and then chkdsk.)

Well, both drives in a mirror are complete, operational, copies so you
could disable the array and work with each one separately to see which is
(most) intact, do data recovery, diagnostics, or whatever (like boot), and
then reestablish the mirror once sorted out. Of course, if it's because one
of the drives failed you'd be reestablishing the mirror with a new drive.

Now, I don't know that it corrupted things any worse than it already was
since if it was truly a defective drive then mirror operations would likely
have failed anyway and that might explain why it was so slow booting:
repeated attempts trying to write on a bad drive.

 

[Carlos]

(e-mail address removed) wrote:

Now, I don't know that it corrupted things any worse than it already was
since if it was truly a defective drive then mirror operations would
likely have failed anyway

I'm a little confused... isn't this why we use RAID?... to be able to
have a safe backup in case a drive goes bad?

 

[David Maynard]

I'm a little confused... isn't this why we use RAID?... to be able to
have a safe backup in case a drive goes bad?

RAID is not an alternative to, or substitute for, backups.

However, the issue was the OP forcing a mirror sync in the middle of the
catastrophe without knowing, or checking, what was bad or good.

 

[john_doe]

I'm a little confused... isn't this why we use RAID?... to be able to
have a safe backup in case a drive goes bad?

This is why you back up

 

[Squibbly]

This is why you back up

i think those SATA cables are so flimsy