P5LD2-Deluxe and RAID 1 rebuild...how to stop

[R. J. Salvi]

Not having a P5LD2-Deluxe system in front of me to reference, I need to
guide an individual though *disabling* the Intel Matrix software (or RAID
BIOS) from prompting them to rebuild a RAID 1 aarray t bootup. This happens
every boot. Any help (or web resource) is appreciated. Thanks and Happy New
Year!

 

[Paul]

Not having a P5LD2-Deluxe system in front of me to reference, I need to
guide an individual though *disabling* the Intel Matrix software (or RAID
BIOS) from prompting them to rebuild a RAID 1 aarray t bootup. This happens
every boot. Any help (or web resource) is appreciated. Thanks and Happy New
Year!

When the Intel software wants to rebuild the RAID 1 array, it
means the two disks have become de-synchronized. For example, if
the computer boots, and one of the two disks is not ready, the
disk that is ready is marked as being an "orphan". At the
next opportunity, the software will want to copy the good
disk to the other disk, to resync the members of the RAID 1
array. So, that is what you are trying to stop.

You need to understand what is causing the bad status of
the array. Whether there is a run time error, where one
disk has become unresponsive, or whether there is an
error right at shutdown time, causing the software to
conclude the array is broken. The Intel software must
be marking the array as broken at some point during
each run, and that is why it is trying to rebuild each
time.

Check the Event Viewer or any other place the Intel documentation
suggests that errors are recorded, and see if there is a log
of the problems.

I don't have experience with RAID, and if this was my problem
and I had to get on with the job (i.e. I needed the computer
right now and could not wait hours for an off-line resync),
I would disconnect one of the two disks (the sick one, if you
know which is which), then run the machine with one disk.
The computer is still going to try to resync, when the
second disk is again plugged in.

I would be careful to ascertain which disk is most likely
to contain a good copy of the data on the array. If there
is any concern whatsoever, about the integrity of the array,
the very first thing to do is an incremental backup, so you
can use your existing backup media members to restore the
array if a problem develops. An incremental shouldn't take
too long to run, if backups are done regularly. Better to
be safe, than sorry.

You might also recommend getting a copy of the disk test
software from the disk manufacturer's web site. Since
the array already has a broken status, running the computer
with just one disk connected, and doing a read-only test
with one of those test tools, may highlight which disk has
a bad SMART status or other problem. That may speed up the
resolution of which disk to replace, if it is an actual
disk hardware problem.

HTH,
Paul

 

[R. J. Salvi]

When the Intel software wants to rebuild the RAID 1 array, it
means the two disks have become de-synchronized. For example, if
the computer boots, and one of the two disks is not ready, the
disk that is ready is marked as being an "orphan". At the
next opportunity, the software will want to copy the good
disk to the other disk, to resync the members of the RAID 1
array. So, that is what you are trying to stop.

You need to understand what is causing the bad status of
the array. Whether there is a run time error, where one
disk has become unresponsive, or whether there is an
error right at shutdown time, causing the software to
conclude the array is broken. The Intel software must
be marking the array as broken at some point during
each run, and that is why it is trying to rebuild each
time.

Check the Event Viewer or any other place the Intel documentation
suggests that errors are recorded, and see if there is a log
of the problems.

I don't have experience with RAID, and if this was my problem
and I had to get on with the job (i.e. I needed the computer
right now and could not wait hours for an off-line resync),
I would disconnect one of the two disks (the sick one, if you
know which is which), then run the machine with one disk.
The computer is still going to try to resync, when the
second disk is again plugged in.

I would be careful to ascertain which disk is most likely
to contain a good copy of the data on the array. If there
is any concern whatsoever, about the integrity of the array,
the very first thing to do is an incremental backup, so you
can use your existing backup media members to restore the
array if a problem develops. An incremental shouldn't take
too long to run, if backups are done regularly. Better to
be safe, than sorry.

You might also recommend getting a copy of the disk test
software from the disk manufacturer's web site. Since
the array already has a broken status, running the computer
with just one disk connected, and doing a read-only test
with one of those test tools, may highlight which disk has
a bad SMART status or other problem. That may speed up the
resolution of which disk to replace, if it is an actual
disk hardware problem.

HTH,
Paul

Thanks Paul. I know which of the disks is "degraded" and the data from the
healthy drive *is* backed up regularly, but there's an option in the RAID
BIOS that when selected -- and I forget the exact language -- prompts
Intel's Windows-based utility to rebuild the array on reboot. Since it's not
my machine and cannot be repaired any time soon, the owners have inquired
about deactivating the rebuild sequence when the box boots up.
Unfortunately, the owners don't have the time to allow for an array rebuild
just yet and I don't consider them proficient enough to open the box and
remove a drive.

The only quick and dirty fix I can think of is to uninstall Intel's
Windows-based Matrix software so it won't automatically prompt the rebuild
on reboots. Again, since I'm not sitting in front of the machine, I can't
recall if there might be another way to deactivate the rebuild process in
the RAID BIOS. Although it's a RAID 1 and the data is mirrored as opposed to
striped, I'm reticent to advise "deleting the array" since the possibility
of data loss exists. That's where I am...argh.

 

[Paul]

Thanks Paul. I know which of the disks is "degraded" and the data from the
healthy drive *is* backed up regularly, but there's an option in the RAID
BIOS that when selected -- and I forget the exact language -- prompts
Intel's Windows-based utility to rebuild the array on reboot. Since it's not
my machine and cannot be repaired any time soon, the owners have inquired
about deactivating the rebuild sequence when the box boots up.
Unfortunately, the owners don't have the time to allow for an array rebuild
just yet and I don't consider them proficient enough to open the box and
remove a drive.

The only quick and dirty fix I can think of is to uninstall Intel's
Windows-based Matrix software so it won't automatically prompt the rebuild
on reboots. Again, since I'm not sitting in front of the machine, I can't
recall if there might be another way to deactivate the rebuild process in
the RAID BIOS. Although it's a RAID 1 and the data is mirrored as opposed to
striped, I'm reticent to advise "deleting the array" since the possibility
of data loss exists. That's where I am...argh.

One problem I see here, is anything suggested should really
be tested. You probably don't have the gear to exactly
reproduce their system config, otherwise you'd have an
answer by now. And tossing out ideas that happen to fail
for the individual won't make you any friends. (If you
have a motherboard with ICH5R, ICH6R, or ICH7R, plus two
blank disks, that would be a close enough environment
for the purposes of this experiment.)

One thing I'm curious about, is the Intel RAID BIOS does
its thing on the RAID BIOS screen. But the Main BIOS page
lists "Primary", "Third", and "Fourth" Master and Slave
drives. Is it possible that going into the appropriate
BIOS entry on the Main BIOS page, would allow you to
disable the degraded drive ? I really would have expected
that the presence of the RAID BIOS, would make the
disks "disappear" from the Main BIOS page.

Deleting the array, or disabling the RAID BIOS, should
really be tested, before being suggested as a solution.
If the array is used to boot the computer, there might
only be grief waiting for you with those options. Too
many things can go wrong...

As I understand it, disabling RAID in the BIOS, actually
causes the enumeration of the Southbridge to change, as
seen by Windows. Thus, if you have an ICH7R, if you
disable Southbridge RAID in the BIOS, in Windows the
Southbridge should list itself as ICH7. That causes
a "Catch22" situation, and is the reason there is a
"RAID Ready" procedure listed here. By enabling RAID
right away, and installing the driver, the OS is ready
when the migration is eventually done. Otherwise, the
difference in enumeration (ICH7 versus ICH7R) will prevent
a RAID driver from being installable when needed. And you
can imagine how Windows feels about an enumeration change.

ftp://download.intel.com/support/chipsets/iaa_raid/IAAR_Quick_Start.pdf

Post back what happens

Good luck,
Paul

 

[R. J. Salvi]

One problem I see here, is anything suggested should really
be tested. You probably don't have the gear to exactly
reproduce their system config, otherwise you'd have an
answer by now. And tossing out ideas that happen to fail
for the individual won't make you any friends. (If you
have a motherboard with ICH5R, ICH6R, or ICH7R, plus two
blank disks, that would be a close enough environment
for the purposes of this experiment.)

One thing I'm curious about, is the Intel RAID BIOS does
its thing on the RAID BIOS screen. But the Main BIOS page
lists "Primary", "Third", and "Fourth" Master and Slave
drives. Is it possible that going into the appropriate
BIOS entry on the Main BIOS page, would allow you to
disable the degraded drive ? I really would have expected
that the presence of the RAID BIOS, would make the
disks "disappear" from the Main BIOS page.

Deleting the array, or disabling the RAID BIOS, should
really be tested, before being suggested as a solution.
If the array is used to boot the computer, there might
only be grief waiting for you with those options. Too
many things can go wrong...

As I understand it, disabling RAID in the BIOS, actually
causes the enumeration of the Southbridge to change, as
seen by Windows. Thus, if you have an ICH7R, if you
disable Southbridge RAID in the BIOS, in Windows the
Southbridge should list itself as ICH7. That causes
a "Catch22" situation, and is the reason there is a
"RAID Ready" procedure listed here. By enabling RAID
right away, and installing the driver, the OS is ready
when the migration is eventually done. Otherwise, the
difference in enumeration (ICH7 versus ICH7R) will prevent
a RAID driver from being installable when needed. And you
can imagine how Windows feels about an enumeration change.

ftp://download.intel.com/support/chipsets/iaa_raid/IAAR_Quick_Start.pdf

Post back what happens

Here are -- what I think are -- the only options I can possibly suggest:

A.) Uninstall Intel's Windows-based matrix software. This could work and
it's the least intrusive, but I don't know how tightly it's integrated with
the RAID BIOS. However, if it's listed in the startup menu, it could be
"rem'd out" with MSCONFIG. If it's a service, I'd be hesitant disabling it.
B.) Delete the array from the RAID BIOS. Too risky even though all necessary
boot and system files are on the good drive.
C.) Deactivate RAID from the main BIOS. See B.
D.) Unplug the power and SATA cable from the offending drive. In no way do I
want a neophyte poking around inside the box.

If I get *my* hands on the box, I'll try some of the above possibilities.
However, and in the interests of self-preservation, I've decided against
suggesting anything except "proper repair" because ultimately, anything I
suggest that doesn't work or causes data loss, I'll get blamed for. ;-)

Thanks again Paul.