Possible Onbaord RAID Controller Failure

  • Thread starter Thread starter Jeff
  • Start date Start date
J

Jeff

Hi Everyone,

I have done a little searching; but, I haven't seen anything relevant to my
problem. I hope the post is appropriate here.

My problem is that I started trying to determine why my Windows XP computer
would not boot normally. I was able to boot into Safe Mode with Networking
for about five days, and then I couldn't even boot into safe mode.

After some initial troubleshooting, I was helped by Daave on the Window XP
General forum. He provided a great outline for troubleshooting which began
with making sure there was not malware which had found its way into the
system. My problem at that point was that when booting up via two different
vendor's .iso files, I would get to a point where the hard drives were not
recognized.

As a point of reference, the hard drives were two identical drives set up as
a RAID0 configuration.

I can provide many more details of my troubleshooting plan; but, maybe it
will be easier if I give you the major results.

First, the system that I am trying to troubleshoot has an onboard Promoise
RAID controller.

As a point of reference, if I run the Win XP setup disk, I get to the Setup
screen which lists the drives which are available for installation. There are
four instances of Unknown Devices, and no volumes available for the
installation of XP.

I have made sure to load current RAID drivers during the XP Set routine. Yet
the RAID configuration is not identified.

If I go into the RAID setup utility which runs after POST, and I delete the
RAID0 volume, and then I go into BIOS and change the setting for SATA as RAID
from Yes to No, and reboot the system.... which spins up the Win XP setup
disk... I enter the setup routine again.

This time... I have two volumes available for the installation.

This leads me to believe that either the RAID controller on the motherboard
is bad, or the drivers are not up-to-date for the RAID controller during Win
XP setup. I know that I have the most current drivers from ASUS/Promise, so
I am reasonably certain that they are not the problem.

Does anyone think that I am on the right track?

Replacing the motherboard probabaly isn't a good option. Do you know if
there is a way to disable the RAID controller on the motherboard? Are there
RAID controllers mounted to aftermarket cards that I can run as a standalone
from the motherboard? (I will investigate those options today; but, I thought
you might know off the top of your head.)

I will probably just go ahead and install a larger hard drive in a non-RAID
configuration, and try to get things back up and running. The added speed of
the RAID0 configuration isn't really necessary now as this is more of a
back-up system.

I will always be a little suspect, though, fearing that other components on
the board may be close to failure.

The original thread that I was working through with Daave is
http://groups.google.com/group/micr...ead/e4f5f555b3121946/4795274e8331c587?hl=en&q

Thanks for any help that can be offered.

Jeff
 
Jeff, you should have a option to disable the Raid Controller in the
Bios. They do have PCI Raid Controller Cards.
 
Hi Rich,

Thanks for your message.

I have disabled RAID in BIOS... I'm not familiar enough with this type of
hardware to know if there is something more that I need to do.

Do you think my observation is close to correct with respect to the onboard
controller being bad? Is there any way to check to know for sure?

Thanks, too, for your recommendation of the PCI Raid Controller Card.

Take Care

Jeff
 
Jeff said:
Hi Everyone,

I have done a little searching; but, I haven't seen anything relevant to my
problem. I hope the post is appropriate here.

My problem is that I started trying to determine why my Windows XP computer
would not boot normally. I was able to boot into Safe Mode with Networking
for about five days, and then I couldn't even boot into safe mode.

After some initial troubleshooting, I was helped by Daave on the Window XP
General forum. He provided a great outline for troubleshooting which began
with making sure there was not malware which had found its way into the
system. My problem at that point was that when booting up via two different
vendor's .iso files, I would get to a point where the hard drives were not
recognized.

As a point of reference, the hard drives were two identical drives set up as
a RAID0 configuration.

I can provide many more details of my troubleshooting plan; but, maybe it
will be easier if I give you the major results.

First, the system that I am trying to troubleshoot has an onboard Promoise
RAID controller.

As a point of reference, if I run the Win XP setup disk, I get to the Setup
screen which lists the drives which are available for installation. There are
four instances of Unknown Devices, and no volumes available for the
installation of XP.

I have made sure to load current RAID drivers during the XP Set routine. Yet
the RAID configuration is not identified.

If I go into the RAID setup utility which runs after POST, and I delete the
RAID0 volume, and then I go into BIOS and change the setting for SATA as RAID
from Yes to No, and reboot the system.... which spins up the Win XP setup
disk... I enter the setup routine again.

This time... I have two volumes available for the installation.

This leads me to believe that either the RAID controller on the motherboard
is bad, or the drivers are not up-to-date for the RAID controller during Win
XP setup. I know that I have the most current drivers from ASUS/Promise, so
I am reasonably certain that they are not the problem.

Does anyone think that I am on the right track?

Replacing the motherboard probabaly isn't a good option. Do you know if
there is a way to disable the RAID controller on the motherboard? Are there
RAID controllers mounted to aftermarket cards that I can run as a standalone
from the motherboard? (I will investigate those options today; but, I thought
you might know off the top of your head.)

I will probably just go ahead and install a larger hard drive in a non-RAID
configuration, and try to get things back up and running. The added speed of
the RAID0 configuration isn't really necessary now as this is more of a
back-up system.

I will always be a little suspect, though, fearing that other components on
the board may be close to failure.

The original thread that I was working through with Daave is
http://groups.google.com/group/micr...ead/e4f5f555b3121946/4795274e8331c587?hl=en&q

Thanks for any help that can be offered.

Jeff
Click to expand...

The first thing I'd like you to do, is tell us the motherboard
model number and the kind of Promise chip.

My immediate reaction is this. The majority of "RAID as an afterthought"
added to motherboards, are "soft" RAID devices. It is like simple IDE
controllers with a RAID overlay. It relies on the firmware Extended
INT 0x13, for booting the RAID during POST. And once booted, the
driver in the OS does something similar. The driver accesses the drives
as simple IDE devices, and does it in a way suitable for making it
look like RAID 0 (striped) or RAID 1 (mirrored). So it is the drivers
job, to decide which disk to access and when. The hardware chip itself
is not striping or mirroring on its own. There would be no register
in the hardware, declaring the stripe size. Only the driver knows.
The chip is dumb.

So if it was my hardware, a diagnostic would be to

1) Flip to non-RAID mode in the BIOS.
2) Boot with the hard drive manufacturer diagnostic. These are
available in floppy, CD, and perhaps even USB stick versions.
The diagnostic may use FreeDOS and run the diagnostic on top of
that. By doing so, no Windows OS needs to be working.
3) Usually, the diagnostic will be read-only. If the diagnostic
has options which do writes, it should warn you it is about
to wipe out all data.
4) Some hard drive manufacturers provide no diagnostics. I'm
not sure now, but Hitachi may not have any for download.

If the diagnostic passes without throwing errors, then you know
the hardware is not busted. (Check both disks, so you know both
cables are working.) Then the problem is at some other level.

1) CMOS problem. Clear CMOS (with the power cord on the computer unplugged).
Make sure you've written down the BIOS settings before hand, as
you'll need to restore them. Very few motherboards support
"profiles" and easy restoration.
2) BIOS firmware problem. The firmware is protected by a checksum,
so if the BIOS firmware was corrupted, it should be detected at
POST by the boot block. I see little reason to be re-flashing the
BIOS at this point. The boot block starts first, and verifies the
main BIOS (holding the Promise code) is OK.
3) Metadata content of disks. When you use the RAID Setup in the BIOS,
you're writing a 64K block (reserved sector) somewhere on the disk.
My experience with my own Promise, is I suspect it may even use
metadata when in non-RAID mode. But I haven't wasted the time
proving it. When you go into the BIOS and delete the RAID array,
that too should make a change to the metadata. Whether it completely
erases the area used, or writes "ordinary IDE" in the block is something
I don't know.
4) Windows driver problem.
5) Windows installation problem. Windows will not rewrite the MBR, if
there is already a signature in the first sector where the MBR is stored.
This causes a problem on the first reboot. I don't know if partitioning
or partition types (all partitions EXT2 or whatever), would prevent
Windows from starting over again, and putting a new partition somewhere.
I thoroughly erase disks, if I'm hitting a brick wall. DBAN from
dban.org is one such tool. Linux and "dd" using /dev/zero is a more
efficient erase, and only takes me seconds to prepare a disk, by
erasing the first few megabytes. With the RAID drives, I'd probably
erase them right up to the very end, to ensure I'd wiped the metadata
block.

When you see the "unknown devices" and there are four of them,
do they appear to be identical devices ? How many hard drives are
present in the computer in the first place ? When I'm installing
here at home, I like to unplug all drives not related to the
installation, so I'm only dealing with the boot drive or boot array.

So please post what kind of motherboard this is, how many drives
are present, and where they're connected.

Paul
 
First set for RAID enable in the BIOS,set for RAID to run with IDE devices
(cdrom,
other IDE hd,etc),set the BIOS to boot to xp cd-1st priority,set to boot to
RAID
2nd,save & exit BIOS,set-up youre RAID set in RAID utility,exit when
thru,press
F6 for RAID driver option,you need to run from the floppy 2X (once it asks
to use
drivers from floppy or xps),press enter the 3rd starting time,xp cd install
menu
will show after.
 
Hi Andrew,

I had done exactly what you suggested in your message prior to posting. With
the onboard controller and BIOS set to RAID, the hard drives were not
recognized.

Although I am confused by you saying that I need to run from the floppy 2X?

Thanks for your advice.

Jeff
 
Hi Paul,

Paul said:
The first thing I'd like you to do, is tell us the motherboard
model number and the kind of Promise chip.

My immediate reaction is this. The majority of "RAID as an afterthought"
added to motherboards, are "soft" RAID devices. It is like simple IDE
controllers with a RAID overlay. It relies on the firmware Extended
INT 0x13, for booting the RAID during POST. And once booted, the
driver in the OS does something similar. The driver accesses the drives
as simple IDE devices, and does it in a way suitable for making it
look like RAID 0 (striped) or RAID 1 (mirrored). So it is the drivers
job, to decide which disk to access and when. The hardware chip itself
is not striping or mirroring on its own. There would be no register
in the hardware, declaring the stripe size. Only the driver knows.
The chip is dumb.

So if it was my hardware, a diagnostic would be to

1) Flip to non-RAID mode in the BIOS.
2) Boot with the hard drive manufacturer diagnostic. These are
available in floppy, CD, and perhaps even USB stick versions.
The diagnostic may use FreeDOS and run the diagnostic on top of
that. By doing so, no Windows OS needs to be working.
3) Usually, the diagnostic will be read-only. If the diagnostic
has options which do writes, it should warn you it is about
to wipe out all data.
4) Some hard drive manufacturers provide no diagnostics. I'm
not sure now, but Hitachi may not have any for download.

If the diagnostic passes without throwing errors, then you know
the hardware is not busted. (Check both disks, so you know both
cables are working.) Then the problem is at some other level.

1) CMOS problem. Clear CMOS (with the power cord on the computer unplugged).
Make sure you've written down the BIOS settings before hand, as
you'll need to restore them. Very few motherboards support
"profiles" and easy restoration.
2) BIOS firmware problem. The firmware is protected by a checksum,
so if the BIOS firmware was corrupted, it should be detected at
POST by the boot block. I see little reason to be re-flashing the
BIOS at this point. The boot block starts first, and verifies the
main BIOS (holding the Promise code) is OK.
3) Metadata content of disks. When you use the RAID Setup in the BIOS,
you're writing a 64K block (reserved sector) somewhere on the disk.
My experience with my own Promise, is I suspect it may even use
metadata when in non-RAID mode. But I haven't wasted the time
proving it. When you go into the BIOS and delete the RAID array,
that too should make a change to the metadata. Whether it completely
erases the area used, or writes "ordinary IDE" in the block is something
I don't know.
4) Windows driver problem.
5) Windows installation problem. Windows will not rewrite the MBR, if
there is already a signature in the first sector where the MBR is stored.
This causes a problem on the first reboot. I don't know if partitioning
or partition types (all partitions EXT2 or whatever), would prevent
Windows from starting over again, and putting a new partition somewhere.
I thoroughly erase disks, if I'm hitting a brick wall. DBAN from
dban.org is one such tool. Linux and "dd" using /dev/zero is a more
efficient erase, and only takes me seconds to prepare a disk, by
erasing the first few megabytes. With the RAID drives, I'd probably
erase them right up to the very end, to ensure I'd wiped the metadata
block.

When you see the "unknown devices" and there are four of them,
do they appear to be identical devices ? How many hard drives are
present in the computer in the first place ? When I'm installing
here at home, I like to unplug all drives not related to the
installation, so I'm only dealing with the boot drive or boot array.

So please post what kind of motherboard this is, how many drives
are present, and where they're connected.

Paul
Click to expand...

Finally... A few successes!

First to answer a few of your questions.

The motherboard is an Asus Socket 478, P4C88-E Deluxe motherboard. This is
the board that came with the Compaq X09 system.

If the documentation that I have for the motherboard is correct, the Promise
controller is a PDC 20378 controller. After reading through the manual more
closely, it appears that the way the HDDs were cabled on the original system
(and still are) the RAID control is actually by the Intel ICH5R chipset. (I
was confused by the references I saw to the Promise controller... now I
understand.)

If you're interested, here's a link to the Asus page which lists the user
manual download:
http://dlcdnas.asus.com/pub/ASUS/mb/sock478/P4C800E-DX/e1347b_p4c800-e_deluxe.pdf.
(It should be the last item in the list, if the link works correctly.) Here's
another link to the installation guide for the Promise controller:
http://support.asus.com/technicaldocuments/PDC20378.pdf

There were two hard drives present in the system at the time the problem
began. They were configured as RAID0. Each was attached to a connector
labeled SATA on the motherboard (one labelled SATA1 and the other labelled
SATA2).

After reading the documentation, I believe that the RAID configuration was
actually controlled by the Intel ICH5R chipset. I found that information on
page 5-33 of the motherboard manual.

Taking a step back... I attached the SATA hard drives to the SATA RAID1 and
SATA RAID2 connectors on the motherboard. I then had to enable the Promise
controller in BIOS. During POST I was given the option to configure the RAID
array using the promise controller. Following reboot, the XP setup began
again, I pressed F6 to load drivers, and when I was asked, I loaded the
Promise Fastrak RAID drivers. At the completion of this the two hard drives
were recognized (as a single volume). Great event!

Why did this happen? I was using the Promise driver when the SATA drives
were attached to the two connectors which interface with the Intel RAID
controller.

This leads me to believe that if I load the correct Intel drivers for the
RAID controller during XP setup, I will be able to see the drives during XP
install... I am going to try that next if I can locate the correct drivers.

I'll keep you posted.

Jeff
 
Hi Everyone,

Well... it appears that things are running normally again.

It's a long story; but, the biggest problem with the hard drive installation
issue has to do with me not understanding that there are two RAID controllers
on the Asus motherboard. When I read about the Promise controller when I
first started through the manual, I "assumed" that was the only RAID
controller, and that was the driver I attempted to install during the Win XP
setup. That's why XP setup never recognized the array.

I now know that I can configure RAID volumes through either the Intel
chipset, or the onboard Promise controller. I know which BIOS settings need
to be changed and which drivers to install during the XP setup routine.

I do have a question about the drives, though...

I initially set up a pair of identical SATA HDDs as RAID0 with the Promise
controller, and with the cabling correctly terminated. I thought I would
experiment and verify that I could set the drives up as RAID0 using the Intel
RAID controller. I changed the cabling, made changes to BIOS, made sure the
RAID array was configured properly, and restarted the XP setup. Setup
recognized the volume; however, when setup attempted to format the volume, it
hung.... I had to do a hard restart to exit setup.

I ended up removing the drives from the system and moving them to another
system where I made sure each HDD showed a single block of unallocated space.
(On one of the hard drives there was a volume that had to be deleted. Why one
and not the other?)

Other than not knowing what caused the original problem, I feel pretty
comfortable with where things stand. I've learned a lot....

Thanks again.

Jeff
 
Jeff said:
Hi Everyone,

Well... it appears that things are running normally again.

It's a long story; but, the biggest problem with the hard drive installation
issue has to do with me not understanding that there are two RAID controllers
on the Asus motherboard. When I read about the Promise controller when I
first started through the manual, I "assumed" that was the only RAID
controller, and that was the driver I attempted to install during the Win XP
setup. That's why XP setup never recognized the array.

I now know that I can configure RAID volumes through either the Intel
chipset, or the onboard Promise controller. I know which BIOS settings need
to be changed and which drivers to install during the XP setup routine.

I do have a question about the drives, though...

I initially set up a pair of identical SATA HDDs as RAID0 with the Promise
controller, and with the cabling correctly terminated. I thought I would
experiment and verify that I could set the drives up as RAID0 using the Intel
RAID controller. I changed the cabling, made changes to BIOS, made sure the
RAID array was configured properly, and restarted the XP setup. Setup
recognized the volume; however, when setup attempted to format the volume, it
hung.... I had to do a hard restart to exit setup.

I ended up removing the drives from the system and moving them to another
system where I made sure each HDD showed a single block of unallocated space.
(On one of the hard drives there was a volume that had to be deleted. Why one
and not the other?)

Other than not knowing what caused the original problem, I feel pretty
comfortable with where things stand. I've learned a lot....

Thanks again.

Jeff
Click to expand...

Generally speaking, the reserved sector on RAIDs, don't take the same
format. So you cannot expect to move a RAID0 from the Promise to the
Intel, and have it remain a RAID0. You'd need to use the Intel BIOS
or Intel RAID Management software, to put new reserved sectors (metadata)
on the drives.

You can move a Promise RAID0, from one Promise controller to another.
Apparently, there is some level of similarity between metadata, on
the Promise products.

Tomshardware did some compatibility testing, where they tried to move
hardware from one RAID solution to another. And that is the nature of
the results - some companies support some level of compatibility within
their own products, and even then, there are no guarantees. Something
you should learn from that, is that if your motherboard dies, your
RAID array may not be easy to move to another computer. Perhaps
a PCI RAID card, which can be unplugged from the motherboard,
makes more sense in that case. Or, making frequent backups
of the array on a single drive somewhere, so you can restore
to the array once it has been moved and prepared on another
computer. Either that, or be a wizard at data recovery :-)

What may have happened in your Intel case, is you didn't press F6
and install a RAID driver ? Maybe the WinXP installer was running,
using the default Microsoft driver or something.

Paul
 
Hi Paul,

I wasn't trying to move the array... I wanted to reconfigure the hard drives
that had been part of one array and make a new array from them. Perhaps the
metadata you reference needed to be deleted before they would work properly
with the other type of controller.

Lessons learned....

Thanks for your input.

Jeff
 
Back
Top