matched drives needed for Silicon Image SiI 3112 SATALink ?

[-RS-]

I have a Intel D845PEBT2 mobo with the on-board SATA mentioned in the
subject line.

One of the 2 year old Seagate Baracuda 200Gb SATA HDDs is reported as
"dropped" by the Silicon Image BIOS/setup.

I have been told that the only way to safely keep from losing data is to
swap in a fresh HDD for the dropped one, and REBUILD the RAID set.

It is hard to find the exact same Seagate 200Gb HDDs. How different can the
2nd SATA HDD drive be? The mobo manual and read me file and CD version of
the manual dont' say much.

Thanks,

-RS-

 

[Arno Wagner]

I have a Intel D845PEBT2 mobo with the on-board SATA mentioned in the
subject line.

One of the 2 year old Seagate Baracuda 200Gb SATA HDDs is reported as
"dropped" by the Silicon Image BIOS/setup.

I have been told that the only way to safely keep from losing data is to
swap in a fresh HDD for the dropped one, and REBUILD the RAID set.

It is hard to find the exact same Seagate 200Gb HDDs. How different can the
2nd SATA HDD drive be? The mobo manual and read me file and CD version of
the manual dont' say much.

Any HDD with the same interface and the same number or more of
sectors (!) should do. Note that seome 200GB HDDs will havw a
smaller number of sectors, so better get a 250GB one if you are
not sure.

However, before you rebuild:

Do a full backup (you know that RAID is not a replacement
for backup, right?)

Afterwards or before, but after the backup:

Find out what the isse with the dropped disk is. It may be transient.
It may also be something systematic that threatens the other
disk as well.

Arno

 

[-RS-]

Should I be able to boot, read data etc. from the non-dropped disk, by
itself? (So that I can backup data onto removable media.)

So far I have not been able to boot from it. This was a mirrored RAID setup.

Thanks,

-Ron-

 

[Arno Wagner]

Should I be able to boot, read data etc. from the non-dropped disk, by
itself? (So that I can backup data onto removable media.)

That would be the point of using RAID in the fist place....

So far I have not been able to boot from it. This was a mirrored RAID setup.

No? Then there is no real point to call this RAID, now is there.
In fact you should be able to use the degraded array as if it
was complete.

Arno

 

[-RS-]

Yes, that's what I am wondering ... where is the much-vaunted RAID security?
I was told that it just enables me to re-build the set (with a new
replacement SATA HDD), and not being able to boot from 1/2 the mirrored set
is par for the course. I gather you say no to that and I should be able to
boot from just the non-dropped drive?

When I try to boot normally, the system stalls on the "incomplete RAID set"
screen. If I go into the mobo's BIOS and change from RAID to just SATA boot,
it "tries" to boot. I get the Windows XP splash screen with the animated bar
for a while ... then it reboots and repeats this, ad nauseum.

-Ron-

 

[Arno Wagner]

Yes, that's what I am wondering ... where is the much-vaunted RAID
security? I was told that it just enables me to re-build the set
(with a new replacement SATA HDD), and not being able to boot from
1/2 the mirrored set is par for the course. I gather you say no to
that and I should be able to boot from just the non-dropped drive?

Well, with a real RAIDF controller (hardware or software) you can.
You miught have a very cheap "fakeraid" controller, that is really
a brain-damaged software RAID without the advantages. In that case
you may indeed have to rebuild with a new drive.

When I try to boot normally, the system stalls on the "incomplete
RAID set" screen. If I go into the mobo's BIOS and change from RAID
to just SATA boot, it "tries" to boot. I get the Windows XP splash
screen with the animated bar for a while ... then it reboots and
repeats this, ad nauseum.

As I said, you may actually have to get that second drive.
I really think people that put out this type of substandard
technology that makes empty promises and then stabs you in
the back should be barred from ever working as engineers again.

Arno

 

[-RS-]

I got a universal USB > HDD dongle & power supply thing and took out the
RAID HDDs and plugged them one at a time into it and read the contents on
another PC.

- First, I tried the drive that was plugged into the mobo's "secondary" SATA
socket, which I figured must correspond to what the Silicon Image setup
screen reported as drive "1" ... the "dropped" drive. I was able to read and
copy data off it. Unfortunately it had no newer files on it since April
2007.

- Next, I tried the drive that was plugged into the mobo's "primary" SATA
socket, which I figured must correspond to what the Silicon Image setup
screen reported as drive "0" ... the "current" drive. I was able to "see"
the drive in My Computer as Drive E ... however it had 0 bytes available and
0 bytes used !

I now have Spinrite 6 diagnosing, doing its thing on that drive (that I
couldn't read) on another PC with SATA capability, It has been running for
9hrs and 15 min. ... is at 45% done ... and there is another 11hrs 10 min.
to go. So far it doesn't seem to have found any bad sectors. I really hope
Spinrite can do it's magic and make that drive readable so I can copy off
the data that was written since April !!

-Ron-

 

[Arno Wagner]

I got a universal USB > HDD dongle & power supply thing and took out the
RAID HDDs and plugged them one at a time into it and read the contents on
another PC.

- First, I tried the drive that was plugged into the mobo's "secondary" SATA
socket, which I figured must correspond to what the Silicon Image setup
screen reported as drive "1" ... the "dropped" drive. I was able to read and
copy data off it. Unfortunately it had no newer files on it since April
2007.

- Next, I tried the drive that was plugged into the mobo's "primary" SATA
socket, which I figured must correspond to what the Silicon Image setup
screen reported as drive "0" ... the "current" drive. I was able to "see"
the drive in My Computer as Drive E ... however it had 0 bytes available and
0 bytes used !

I now have Spinrite 6 diagnosing, doing its thing on that drive (that I
couldn't read) on another PC with SATA capability, It has been running for
9hrs and 15 min. ... is at 45% done ... and there is another 11hrs 10 min.
to go. So far it doesn't seem to have found any bad sectors. I really hope
Spinrite can do it's magic and make that drive readable so I can copy off
the data that was written since April !!

I doubt that. SpinRite is basically worthless today. You may get lycky
though. Still, if your data is valuable you should instead stop messing
with the drive immediately and have it recoverd by a professional
data recovey outfit.

Hmm. If you can read a drive at all then it does not have a RAID
superblock on disk or it is placed (sensibly) at the end of the
disk. Otherwise accessing the disk would not be possible without the
RAID controller or a compatible software.

Now the only way I can think of why the data stops in April is that
you actually lost the "1" disk in April (letting aside wrong settings
in the system clock). Then the boot would actually have failed
because the second drive died and you would have worked since April
on a degraded (i.e. 1-disk) array.

This shows one major point about RAID: Without integrity
monitoring it is basically worthless. If you can work with a
failed disk for 2 months without noticing, then there is something
seriously wrong. Maybe you did overlook a boot-message? Did
you install the monitoring software that doubtlessly came with the
controller?

Arno

 

[-RS-]

You're right, I had seen an error message about "incomplete RAID set" ...
but do not boot this system often and I must admit that I am guilty of
ignoring it because ... the system was still booting !! (Sad but true)

You mention that the drives should NOT be readable outside of the RAID
environment/controller ? Is it possible that is why the "current" drive
appears to have 0 bytes ... and that actually it is fine ... and all I need
to do is try a rebuild with a fresh drive?

-Ron-

 

[Arno Wagner]

You're right, I had seen an error message about "incomplete RAID set" ...
but do not boot this system often and I must admit that I am guilty of
ignoring it because ... the system was still booting !! (Sad but true)

Well, condolences on your loss, but you also have learned
something valuable...

You mention that the drives should NOT be readable outside of the RAID
environment/controller ? Is it possible that is why the "current" drive
appears to have 0 bytes ... and that actually it is fine ... and all I need
to do is try a rebuild with a fresh drive?

Sorry, no. The thing I was talking about is that depending on disk
layout some RAID controllers cause the disks to be unreadable, while
others do not have that effect. Since you could read one disk just fine,
yours is of the second variant.

The idea is as follows: You have to store a RAID supoerblock
(describes RAID level, geometry and other things) somewhere on
disk. The easy option is to do that at the beginning of the disk
and to present the following sectors shifted (sector number
minus the ones you took) to the OS. Such RAID drives cannot be
read without the controller or some trickery. The other option
is to store the info at the end of the disk and just present
the disk to the OS as being a bit smaller. If the RAID set is
a RAID 1, then you can read the individual disks unraided, just
as you did for the "failed" drive. Most OSes will not mind that
the disk has silently gotten larger.

I really thing your "0" drive failed completely, which is also
supported by you not being able to boot off it.

Arno

 

[Folkert Rienstra]

The symptom you have indicates a problem somewhere early in the filestruc-
ture administration. If by now SpinRite hasn't recovered any sector it likely
won't recover anything. If the drive had no pending bad sectors to begin with
when you started SpinRite then the whole SpinRite exercise is futile anyway.

I doubt that.

SpinRite is basically worthless today.
You may get lycky though.

In other words, you haven't got a clue, babblebot.

Still, if your data is valuable you should instead stop messing
with the drive immediately and have it recoverd by a professional
data recovey outfit.

Hmm. If you can read a drive at all then it does not have a RAID
superblock on disk or it is placed (sensibly) at the end of the disk.

Right, now think hard, babblebot. Hum a little more, if you must.

Otherwise accessing the disk would not be possible without the
RAID controller or a compatible software.

No! Really? He's using SpinRite6, babblebot.
You know, SpinRite, that software that you know nothing about?
That one?

 

[Folkert Rienstra]

Well, condolences on your loss, but you also have learned something valuable...


Sorry, no. The thing I was talking about is that depending on disk
layout some RAID controllers cause the disks to be unreadable, while
others do not have that effect. Since you could read one disk just fine,
yours is of the second variant.

The idea is as follows: You have to store a RAID supoerblock
(describes RAID level, geometry and other things) somewhere on disk.

The easy option is to do that at the beginning of the disk

And there is plenty of space between sector 0 (MBR) and the
first partition bootblock to not be a problem for that either.

and to present the following sectors shifted (sector number
minus the ones you took) to the OS.

And this is the easiest option? Yeah right, of course it is.

Such RAID drives cannot be read without the controller or some
trickery. The other option is to store the info at the end of the
disk and just present the disk to the OS as being a bit smaller.

Babblebot, clueless as always.
The OS only sees the partitions and is completely oblivious of the
real size of the drive. Most partitioners do not even use the full
drive anyway, just leaving the partial end cylinder unused.

If the RAID set is a RAID 1, then you can read the individual
disks unraided, just as you did for the "failed" drive.

And so you can if the unused space at the beginning is used.

Most OSes will not mind that the disk has silently gotten larger.

Smaller, larger, babblebot can't make his mind up, as always.

I really thing your "0" drive failed completely, which is also
supported by you not being able to boot off it.

Or just an important block in the file system administration
failed or was simply overwritten.

 

[-RS-]

Just an update. Spinrite finished Friday afternoon and the log didn't seem
to show much. I tried to read the drive via the USB dongle and it looked
exactly as before (0 bytes, etc.). So it does not look good ... :-(

Have not tried to boot from it using the orginal motherboard yet. I don't
expect that will work. Will try Monday.

-Ron-

 

[Folkert Rienstra]

Just an update. Spinrite finished Friday afternoon and the log didn't seem
to show much. I tried to read the drive via the USB dongle and it looked
exactly as before (0 bytes, etc.).

As expected. What does the filesystem checking utility say.

So it does not look good ... :-(

Have not tried to boot from it using the orginal motherboard yet. I don't
expect that will work.

No change: no work.

Will try Monday.

-Ron-

Badly quoted text cleaned up.
If you don't expect someone to read the mess you made, snip it.

 

[John Turco]

As expected. What does the filesystem checking utility say.


No change: no work.

Badly quoted text cleaned up.
If you don't expect someone to read the mess you made, snip it.

<edited, for brevity>

Hello, Folkert:

Okay, so, why didn't >you< "clean up" the rest of the original poster's
"mess" -- as I, myself, just did? :-J


Cordially,
John Turco <[email protected]>

 

[-RS-]

Just an update: I was able to get the data recovered by a data recovery
service. I have been busy and have not yet put the RAID system back
together to see if it was the controller that was at fault for the actual
failure. If the system seems stable, I will reinstitute the RAID setup and
be much more aware of any error messages !.

-Ron-

 

[Arno Wagner]

Just an update: I was able to get the data recovered by a data recovery
service.

Smar choice IMO, if the data was worth the cost.

I have been busy and have not yet put the RAID system back
together to see if it was the controller that was at fault for the actual
failure. If the system seems stable, I will reinstitute the RAID setup and
be much more aware of any error messages !.

I think the RAID is so-so. It likely works ok, but its
warnings and diagnostics are obviously not really usable.

Better get a RAID controller card that has management software and
would give you warnings and diagnostics while the system is running.

Arno

 

[Folkert Rienstra]

Smar choice IMO, if the data was worth the cost.

Much more likely it was a very expensive way to correct a logical error.

I think the RAID is so-so. It likely works ok, but its
warnings and diagnostics are obviously not really usable.

None are if the warnings are ignored by default.

Better get a RAID controller card that has management software and
would give you warnings and diagnostics while the system is running.

Arno

[snip more poorly quoted text}

 

[-RS-]

Does anyone know if it is likely Ghost 7.5 will work with a striped SATA
RAID 0 set that is read from its motherboard which uses an on-board Silicon
Image RAID chip? The goal is to de-stripe the SATA RAID 0 pair, using
Ghost, and move everything to a single HDD, which will probably be connected
to one of the motherboard's un-used IDE ribbon cables. Thanks !

-Ron-