PG5D2 w/Si3114 RAID 1 disk failed - now what???

I am running SBS 2003 on this board. The RAID 1 array stands behind a
shared data drive, not the system boot disk.

For several weeks I was getting timeout errors on the "disk" in the
event logs. Then on a reboot two weeks ago, an event log entry told
me that failure was "imminent". Then silence...

I rebooted yesterday and went into the SI BIOS (ctrl+S). My disk 0 in
the RAID array was "Dropped", and disk 1 was "Current". I guess the
RAID mirror array did its job and the OS kept going; a little
silently, but fine.

The array had two WD 250 GB disks. So I drove to Fry's and shopped.
Sorry, no 250 GB SATA disks today. So I bought a Seagate 300 GB.
Bigger should be OK, right?

I popped in the disk, in the place of the failed one, and went back
into the RAID BIOS. I thought I'd be offered a chance to Rebuild.
Nope. It first complained of conflicts. OK, fine. "Fix Conflicts".
That ran and now the new disk is "unavailable". I cannot select
Rebuild, as the BIOS tells me that no disk is available for
rebuilding. Huh? Is it because the software is braindead and is
looking for an exact duplicate of the old WD disk? That's 3 year old
technology, and hard to come by. Even if I find a new "250 GB" disk,
it may have a few sectors more or less. They have to expect such a
thing, right?

I am offered the chance to make the new drive a "spare", whatever that
is. The documentation is wholly worthless. It tells me that this
option will create a "spare disk". Thanks. Creating a "spare" and
telling it to use the maximum space occurs very fast. Way too fast to
have copied 250 GB of info anywhere. So it's obviously not doing what
I might expect.

I also installed the SI GUI tool and tried that. At one point it let
me select the new disk and say "Rebuild". At that point I got the
error: "Disk not failed". What the f*** does that mean? Of course
the failed drive was not there any more. I replaced it with the new
one. Silent documentation again. Is this SI RAID as bogus as it
seems to me???

So some specific questions:

1) I could imagine that deleting the RAID array and creating it again
could be a strategy. Random places on the net say that the deletion
might or might not destroy my data. I can't see why it would. The
"deletion" should affect a few bytes in a header. Random places say
that when I rebuild, it should offer me the choice of the source for
the data. (I.e. copy 1 to 0, or 0 to 1?) The documentation is
silent. Should I try a delete and rebuild?

2) What the heck is a spare? Does the RAID system do something with
it? Or is it just an honorary title?

3) Did I need to format the new disk, or something? Maybe a low-level
format?

This has skaken my faith in RAID altogether. Why did I bother buying
two disks in the first place? If I hadn't, there's a 50% chance I'd
still be running today. And if the disk failed, I would have
reinstalled or repaired the OS hours ago, rather than screw around
with menus that don't let me select the options I need.

Thanks!
Chuck

Just correcting the Subject line for archival purposes... (PG5D2 -
P5GD2)

On Fri, 05 May 2006 11:41:20 -0700, Charles Whitmer wrote:

I am running SBS 2003 on this board. The RAID 1 array stands behind a
shared data drive, not the system boot disk.

For several weeks I was getting timeout errors on the "disk" in the
event logs. Then on a reboot two weeks ago, an event log entry told
me that failure was "imminent". Then silence...

I rebooted yesterday and went into the SI BIOS (ctrl+S). My disk 0 in
the RAID array was "Dropped", and disk 1 was "Current". I guess the
RAID mirror array did its job and the OS kept going; a little
silently, but fine.

The array had two WD 250 GB disks. So I drove to Fry's and shopped.
Sorry, no 250 GB SATA disks today. So I bought a Seagate 300 GB.
Bigger should be OK, right?

I popped in the disk, in the place of the failed one, and went back
into the RAID BIOS. I thought I'd be offered a chance to Rebuild.
Nope. It first complained of conflicts. OK, fine. "Fix Conflicts".
That ran and now the new disk is "unavailable". I cannot select
Rebuild, as the BIOS tells me that no disk is available for
rebuilding. Huh? Is it because the software is braindead and is
looking for an exact duplicate of the old WD disk? That's 3 year old
technology, and hard to come by. Even if I find a new "250 GB" disk,
it may have a few sectors more or less. They have to expect such a
thing, right?

I am offered the chance to make the new drive a "spare", whatever that
is. The documentation is wholly worthless. It tells me that this
option will create a "spare disk". Thanks. Creating a "spare" and
telling it to use the maximum space occurs very fast. Way too fast to
have copied 250 GB of info anywhere. So it's obviously not doing what
I might expect.

I also installed the SI GUI tool and tried that. At one point it let
me select the new disk and say "Rebuild". At that point I got the
error: "Disk not failed". What the f*** does that mean? Of course
the failed drive was not there any more. I replaced it with the new
one. Silent documentation again. Is this SI RAID as bogus as it
seems to me???

So some specific questions:

1) I could imagine that deleting the RAID array and creating it again
could be a strategy. Random places on the net say that the deletion
might or might not destroy my data. I can't see why it would. The
"deletion" should affect a few bytes in a header. Random places say
that when I rebuild, it should offer me the choice of the source for
the data. (I.e. copy 1 to 0, or 0 to 1?) The documentation is
silent. Should I try a delete and rebuild?

2) What the heck is a spare? Does the RAID system do something with
it? Or is it just an honorary title?

3) Did I need to format the new disk, or something? Maybe a low-level
format?

This has skaken my faith in RAID altogether. Why did I bother buying
two disks in the first place? If I hadn't, there's a 50% chance I'd
still be running today. And if the disk failed, I would have
reinstalled or repaired the OS hours ago, rather than screw around
with menus that don't let me select the options I need.

Thanks!
Chuck

Charles Whitmer wrote:
I am running SBS 2003 on this board. The RAID 1 array stands behind a
shared data drive, not the system boot disk.

For several weeks I was getting timeout errors on the "disk" in the
event logs. Then on a reboot two weeks ago, an event log entry told
me that failure was "imminent". Then silence...

I rebooted yesterday and went into the SI BIOS (ctrl+S). My disk 0 in
the RAID array was "Dropped", and disk 1 was "Current". I guess the
RAID mirror array did its job and the OS kept going; a little
silently, but fine.

The array had two WD 250 GB disks. So I drove to Fry's and shopped.
Sorry, no 250 GB SATA disks today. So I bought a Seagate 300 GB.
Bigger should be OK, right?

I popped in the disk, in the place of the failed one, and went back
into the RAID BIOS. I thought I'd be offered a chance to Rebuild.
Nope. It first complained of conflicts. OK, fine. "Fix Conflicts".
That ran and now the new disk is "unavailable". I cannot select
Rebuild, as the BIOS tells me that no disk is available for
rebuilding. Huh? Is it because the software is braindead and is
looking for an exact duplicate of the old WD disk? That's 3 year old
technology, and hard to come by. Even if I find a new "250 GB" disk,
it may have a few sectors more or less. They have to expect such a
thing, right?

I would try disconnecting the replacement drive, going back into the SI
BIOS and see what it reports then. It may let you mark that drive as
failed or something and then let you rebuild onto the new disk after you
reconnect it. Maybe you need to add the new disk to the array or
something? I'm not very familiar with the Silicon Image RAID setup.


I am offered the chance to make the new drive a "spare", whatever that
is. The documentation is wholly worthless. It tells me that this
option will create a "spare disk". Thanks. Creating a "spare" and
telling it to use the maximum space occurs very fast. Way too fast to
have copied 250 GB of info anywhere. So it's obviously not doing what
I might expect.

Setting a drive up as a hot spare means it doesn't get used until one of
the other drives fails. Then the data will be mirrored onto the spare
which acts in place of the failed drive.

--
Robert Hancock Saskatoon, SK, Canada
To email, remove "nospam" from
Home Page: http://www.roberthancock.com/

Thanks for the reply. I will try some disconnecting and reconnecting
and see what happens. I now see that on a critical system this is the
kind of game you should play when setting up the system originally.
It would give you a chance to learn the shortcomings of the RAID user
interface.

On the spare disk: Interesting. So it's like having an extra drive
on the shelf that the RAID hardware can decide to pop in when needed.
What is strange at this point is: (1) The spare cannot be assigned to
the RAID set; and (2) They want a size, from 100 MB up through 100 GB,
HALF, or FULL, to be selected. If this is going to be useful, then
the RAID BIOS must have the freedom to assign the spare to any of its
logical RAID arrays. I am guessing that if I set any "size" less than
250 GB, that he will think the spare isn't big enough for this array.
It's a little strange, since he knows darn well how big the disk is.
If I tell it anything less than "FULL", what happens to the extra
space? Would he share it among multiple arrays?

Chuck

On Fri, 05 May 2006 23:30:23 GMT, Robert Hancock
wrote:

Charles Whitmer wrote:
I am running SBS 2003 on this board. The RAID 1 array stands behind a
shared data drive, not the system boot disk.

For several weeks I was getting timeout errors on the "disk" in the
event logs. Then on a reboot two weeks ago, an event log entry told
me that failure was "imminent". Then silence...

I rebooted yesterday and went into the SI BIOS (ctrl+S). My disk 0 in
the RAID array was "Dropped", and disk 1 was "Current". I guess the
RAID mirror array did its job and the OS kept going; a little
silently, but fine.

The array had two WD 250 GB disks. So I drove to Fry's and shopped.
Sorry, no 250 GB SATA disks today. So I bought a Seagate 300 GB.
Bigger should be OK, right?

I popped in the disk, in the place of the failed one, and went back
into the RAID BIOS. I thought I'd be offered a chance to Rebuild.
Nope. It first complained of conflicts. OK, fine. "Fix Conflicts".
That ran and now the new disk is "unavailable". I cannot select
Rebuild, as the BIOS tells me that no disk is available for
rebuilding. Huh? Is it because the software is braindead and is
looking for an exact duplicate of the old WD disk? That's 3 year old
technology, and hard to come by. Even if I find a new "250 GB" disk,
it may have a few sectors more or less. They have to expect such a
thing, right?

I would try disconnecting the replacement drive, going back into the SI
BIOS and see what it reports then. It may let you mark that drive as
failed or something and then let you rebuild onto the new disk after you
reconnect it. Maybe you need to add the new disk to the array or
something? I'm not very familiar with the Silicon Image RAID setup.


I am offered the chance to make the new drive a "spare", whatever that
is. The documentation is wholly worthless. It tells me that this
option will create a "spare disk". Thanks. Creating a "spare" and
telling it to use the maximum space occurs very fast. Way too fast to
have copied 250 GB of info anywhere. So it's obviously not doing what
I might expect.

Setting a drive up as a hot spare means it doesn't get used until one of
the other drives fails. Then the data will be mirrored onto the spare
which acts in place of the failed drive.

Thanks for the help. Problem now solved.

It all comes down to the useless documentation. They tell you how to
set up the array, but never elaborate on the steps to follow in the
event of a failure.

Here's what I needed to do:

Step 0: Select the new disk and "Delete RAID Array". Somehow using
their various tools, i.e all the options in the BIOS and then in the
GUI tool, the new disk was changed into some kind of array on its own
and therefore unavailable for anything.

Step 1: "Create RAID Array" as type "Spare" with the new disk. (This
is the essential missing step.)

Step 2: "Rebuild Array". Only with the new disk set as a spare am I
allowed to rebuild. This is very strange, but it worked!

Chuck
PS - Motherboard type was P5GD2 Deluxe.

On Sat, 06 May 2006 07:24:33 -0700, Charles Whitmer wrote:

Thanks for the reply. I will try some disconnecting and reconnecting
and see what happens. I now see that on a critical system this is the
kind of game you should play when setting up the system originally.
It would give you a chance to learn the shortcomings of the RAID user
interface.

On the spare disk: Interesting. So it's like having an extra drive
on the shelf that the RAID hardware can decide to pop in when needed.
What is strange at this point is: (1) The spare cannot be assigned to
the RAID set; and (2) They want a size, from 100 MB up through 100 GB,
HALF, or FULL, to be selected. If this is going to be useful, then
the RAID BIOS must have the freedom to assign the spare to any of its
logical RAID arrays. I am guessing that if I set any "size" less than
250 GB, that he will think the spare isn't big enough for this array.
It's a little strange, since he knows darn well how big the disk is.
If I tell it anything less than "FULL", what happens to the extra
space? Would he share it among multiple arrays?

Chuck

On Fri, 05 May 2006 23:30:23 GMT, Robert Hancock
wrote:

Charles Whitmer wrote:
I am running SBS 2003 on this board. The RAID 1 array stands behind a
shared data drive, not the system boot disk.

For several weeks I was getting timeout errors on the "disk" in the
event logs. Then on a reboot two weeks ago, an event log entry told
me that failure was "imminent". Then silence...

I rebooted yesterday and went into the SI BIOS (ctrl+S). My disk 0 in
the RAID array was "Dropped", and disk 1 was "Current". I guess the
RAID mirror array did its job and the OS kept going; a little
silently, but fine.

The array had two WD 250 GB disks. So I drove to Fry's and shopped.
Sorry, no 250 GB SATA disks today. So I bought a Seagate 300 GB.
Bigger should be OK, right?

I popped in the disk, in the place of the failed one, and went back
into the RAID BIOS. I thought I'd be offered a chance to Rebuild.
Nope. It first complained of conflicts. OK, fine. "Fix Conflicts".
That ran and now the new disk is "unavailable". I cannot select
Rebuild, as the BIOS tells me that no disk is available for
rebuilding. Huh? Is it because the software is braindead and is
looking for an exact duplicate of the old WD disk? That's 3 year old
technology, and hard to come by. Even if I find a new "250 GB" disk,
it may have a few sectors more or less. They have to expect such a
thing, right?

I would try disconnecting the replacement drive, going back into the SI
BIOS and see what it reports then. It may let you mark that drive as
failed or something and then let you rebuild onto the new disk after you
reconnect it. Maybe you need to add the new disk to the array or
something? I'm not very familiar with the Silicon Image RAID setup.


I am offered the chance to make the new drive a "spare", whatever that
is. The documentation is wholly worthless. It tells me that this
option will create a "spare disk". Thanks. Creating a "spare" and
telling it to use the maximum space occurs very fast. Way too fast to
have copied 250 GB of info anywhere. So it's obviously not doing what
I might expect.

Setting a drive up as a hot spare means it doesn't get used until one of
the other drives fails. Then the data will be mirrored onto the spare
which acts in place of the failed drive.