Loosing data after (W2K pro) shutdown

[Silence Seeker]

I recently built a PC comprised of the following:
- ASUS P4P800 Deluxe motherboard / P4-2.6CGHz
- Samsung SP1614N 160GB HDD
- Windows 2000 Professional

The HDD is connected to the motherboard through the VIA VT6410 RAID
controller, although it is the only drive in the system (I did so
because the VIA RAID controller can operate in the HDD's UDMA6
(ATA133) mode while the other (Intel) IDE controller can only operate
up to UDMA5 (ATA100).

Initially all worked just fine - I installed quite a few applications
and used them happily - until one day I couldn't boot because the
registry's SOFTWARE hive was corrupted.

I wasn't sure what caused this (first time ever happening to me - I
use quite a few W2K Pro workstations at work - none exhibit this
problem), so I decided to re-install the OS and everything else.

Again, for a while all worked just fine - until one day I turn on the
PC and while it booted completely - it reported the installed
ZoneAlram firewall lost its settings.

This is when I realized that there must be somthing problematic in my
HARDWARE configuration.

I (re)searched the newsgroups for "W2K shutting down too quickly" and
got the first hint: Power to the HDD may be turned off before the
HDD's own write-cache gets a chance to flush (and the SP1614N has a
pretty impressive cache: 8MB).

So, I was looking for a way to configure "something" in the system to
remedy this situation. I couldn't find one in the BIOS (if you know of
such in the P4P800 Delux, please advise).


But I then found some posting mentioning the "write cache enable"
option in the disk properties (in Device Manager). But when I
attempted to try this option I discovered that it is... greyed out (in
the disabled state)!

Now I have several questions:
1. why? Why is the "write cache enable" option greyed out
(and disabled)?
2. Is it possible to enable it? If so, how?
3. Should I attempt to enable it? Will this solve the cache flushing
problem or only worsen the problem? (see KB Article 281672:
http://support.microsoft.com/?id=281672)

BTW, the VIA VT6410 RAID controller is identified by Device Manager as
a "SCSI and RAID controller", leading to identifying the Samsun
SP1614N HDD as a "SCSI Disk Device" - despite the fact that it is
actually an IDE/ATAPI drive. Perhaps this can serve as a clue?

Any idea how to solve this problem?

Many thanks!
Sam

 

[DL]

Did win2k have largelba enabled prior to Format?
I read a couple of MVP posts that suggested possible data corruption if a
large disk was formated to below 137gb, then largelba enabled and the free
space subsequently formated to another partition(s)

 

[Psi-Tau Paladin]

(e-mail address removed) (Silence Seeker) wrote in

I (re)searched the newsgroups for "W2K shutting down too quickly" and
got the first hint: Power to the HDD may be turned off before the
HDD's own write-cache gets a chance to flush (and the SP1614N has a
pretty impressive cache: 8MB).

I had this problem with my CUV4X-D mobo with W2K where there was data
corruption if I shut down too quickly. The easiest way I found to remedy
this was to log out before I shut down. This caused the all the current
open files to be written to disk and then prevented the problem from
reoccuring. The 8MB cache shouldn't make a difference because I was using
a 2MB 5400rpm WD at the time.

 

[Silence Seeker]

OK, after some research I think that I got a few more clues.

Loggin out before shutting down is a workaround that has been
mentioned, but I would rather find a solution that allows me to
shutdown in a single step.

I am currently testing the system with the first of the following
applied:

KB 182086: http://support.microsoft.com/support/kb/articles/Q182/0/86.ASP
Set the value to "1" in:
HKLM\SYSTEM\CurrentControlSet\Control\Session Manager\Memory
Management

KB 146092: http://support.microsoft.com/support/kb/articles/Q146/0/92.asp
Set the value to "20,000" [ms] in:
HKLM\SYSTEM\CurrentControlSet\Control\WaitToKillServiceTimeout

Shutdown seems to take forever (I have 1GB RAM), but once I select
"Start|Shut down", I go to sleep and leave the system do its course
until it shuts itself down. So far so good...

Still, I would like to find a more elegant solution - if such one
exists.

In the meanwhile, I have been wondering (perhaps an expert can answer
this):

If write caching was never enabled, why did I experience the problem
in the first place? Could it be that the "write caching" in Device
Manager refers to the one provided by the OS, while the on-disk 8MB
cache is another cache which the OS has no control over?

Thanks,
Sam

 

[Ray Kasprzak]

OK, after some research I think that I got a few more clues.

Loggin out before shutting down is a workaround that has been
mentioned, but I would rather find a solution that allows me to
shutdown in a single step.

I am currently testing the system with the first of the following
applied:

KB 182086: http://support.microsoft.com/support/kb/articles/Q182/0/86.ASP
Set the value to "1" in:
HKLM\SYSTEM\CurrentControlSet\Control\Session Manager\Memory
Management

KB 146092: http://support.microsoft.com/support/kb/articles/Q146/0/92.asp
Set the value to "20,000" [ms] in:
HKLM\SYSTEM\CurrentControlSet\Control\WaitToKillServiceTimeout

Shutdown seems to take forever (I have 1GB RAM), but once I select
"Start|Shut down", I go to sleep and leave the system do its course
until it shuts itself down. So far so good...

Still, I would like to find a more elegant solution - if such one
exists.

In the meanwhile, I have been wondering (perhaps an expert can answer
this):

If write caching was never enabled, why did I experience the problem
in the first place? Could it be that the "write caching" in Device
Manager refers to the one provided by the OS, while the on-disk 8MB
cache is another cache which the OS has no control over?

Thanks,
Sam


(e-mail address removed) (Silence Seeker) wrote in message

I recently built a PC comprised of the following:
- ASUS P4P800 Deluxe motherboard / P4-2.6CGHz
- Samsung SP1614N 160GB HDD
- Windows 2000 Professional

The HDD is connected to the motherboard through the VIA VT6410 RAID
controller, although it is the only drive in the system (I did so
because the VIA RAID controller can operate in the HDD's UDMA6
(ATA133) mode while the other (Intel) IDE controller can only operate
up to UDMA5 (ATA100).

Initially all worked just fine - I installed quite a few applications
and used them happily - until one day I couldn't boot because the
registry's SOFTWARE hive was corrupted.

I wasn't sure what caused this (first time ever happening to me - I
use quite a few W2K Pro workstations at work - none exhibit this

problem), so I decided to re-install the OS and everything else.

Again, for a while all worked just fine - until one day I turn on the
PC and while it booted completely - it reported the installed
ZoneAlram firewall lost its settings.

This is when I realized that there must be somthing problematic in my
HARDWARE configuration.

I (re)searched the newsgroups for "W2K shutting down too quickly" and
got the first hint: Power to the HDD may be turned off before the
HDD's own write-cache gets a chance to flush (and the SP1614N has a
pretty impressive cache: 8MB).

So, I was looking for a way to configure "something" in the system to
remedy this situation. I couldn't find one in the BIOS (if you know of
such in the P4P800 Delux, please advise).


But I then found some posting mentioning the "write cache enable"
option in the disk properties (in Device Manager). But when I
attempted to try this option I discovered that it is... greyed out (in
the disabled state)!

Now I have several questions:
1. why? Why is the "write cache enable" option greyed out
(and disabled)?
2. Is it possible to enable it? If so, how?
3. Should I attempt to enable it? Will this solve the cache flushing
problem or only worsen the problem? (see KB Article 281672:
http://support.microsoft.com/?id=281672)

BTW, the VIA VT6410 RAID controller is identified by Device Manager as
a "SCSI and RAID controller", leading to identifying the Samsun
SP1614N HDD as a "SCSI Disk Device" - despite the fact that it is
actually an IDE/ATAPI drive. Perhaps this can serve as a clue?

Any idea how to solve this problem?

Many thanks!
Sam

Check and see if your virtual memory setting is set for programs or system
cache. Using system cache with a sata drive will often cause exactly the
problem you are having.

 

[Silence Seeker]

Check and see if your virtual memory setting is set for programs or system
cache. Using system cache with a sata drive will often cause exactly the
problem you are having.

Ray, thank you very much for your reply. Indeed, the problem has not
been solved yet, despite setting the ClearPageFileAtShutdown to 1 as
KB182086 suggested.

Which suggests that the problem (and solution) must be elsewhere.

In fact, when I turned on my PC today, it went through CHKDSK and
reported over 50 "cross linked" and "orphaned" files (whatever that
means).

Also, every once in a while Windows Explorer freezes (consuming no CPU
at all!) and the only way to get out of this is by ending the
non-responding process (using task manager). As a result all currently
open folders close and Explorer seems to restart (at least giving me
the system back to normal). These incidents are nicely logged in Event
Viewer as "The shell stopped unexpectedly and Explorer.exe was
restarted", Source: Winlogon.

Now... What is "virtual memory setting is set for programs or system
cache"? When I select "Performance Options" in the "Advanced" tab in
"System Properties", I only get options for:
1. virtual memory *size*
2. Application response

The latter lets me choose between "Optimize performance for:"
Applications or Background services. Is that what you meant? If not,
could you point me please to the correct location?

BTW, my HDD (Samsung SP1614N) is not a SATA drive.
There is a SATA version of it (Samsung SP1614C), but it is quite
possible that internally they are identical. If you are curious, you
can compare specs at:

SP1614N (ATA-7 ) : http://tinyurl.com/36wta
SP1614C (SATA 1.0) : http://tinyurl.com/yshsv

Again, any tip or hint would very appreciated, since I am still
groping in the dark.

Thanks,
Sam

 

[Silence Seeker]

Did win2k have largelba enabled prior to Format?
I read a couple of MVP posts that suggested possible data corruption if a
large disk was formated to below 137gb, then largelba enabled and the free
space subsequently formated to another partition(s)

OMG!!! I think you are right. When I first read you reply (thank
you!), I thought it doesn't apply to my case because I installed
W2K-SP4 (slipstreamed) on this HDD and so all must be well.

But then I encountered KB 305098:
http://support.microsoft.com/default.aspx?scid=kb;EN-US;305098

Which suggests that even if I have SP4, I must manually change (or
add) the EnableBigLba registry value to 1 in the following registry
subkey:
HKLM\SYSTEM\CurrentControlSet\Services\atapi\Parameters

Sure enough, I checked the registry and that value isn't there!

Now... I don't have my 160GB HDD in one big partition. It was divided
prior to formatting/installing so that C: is 16GB and D: is 117GB (I
left some unused space in an unformatted partition). Does what you say
still apply in my case?

Also, you said that "...possible data corruption if a large disk was
formated to below 137gb, then largelba enabled and the free space
subsequently formated to another partition(s)", which suggests that
the large LBA (in W2K's registry?) should be enabled from the start.
But how is it possible to enable large LBA in W2K if W2K is not
installed yet?

Could you explain please what I am missing?

Thanks,
Sam

 

[DL]

I cannot explain 'what yr missing' I am merely repeating what I saw posted
The kb 305098 article does seem to indicate possible problems, and a
solution
Notes: bullet point 5

 

[Silence Seeker]

I cannot explain 'what yr missing' I am merely repeating what I saw posted
The kb 305098 article does seem to indicate possible problems, and a
solution
Notes: bullet point 5

DL, thank you very much for your tips and insight. You have already
helped me focusing on the right direction, so even if you can't answer
some of my questions I am already grateful.

As for bullet point 5, it says:

"If you install a copy of Windows 2000 that includes SP3 (SP3
integrated) on a large hard disk that has already been preformatted by
using a 48-bit LBA-enabled operating system, the ATAPI subsystem may
report hard-disk space greater than that which is addressable without
the 48-bit LBA support (larger than approximately 137 GB) during the
text-mode portion of Setup. In this case, the hard disk's partition
table information has already been created. To fix the incorrect disk
information, delete the partition by using either a disk partitioning
utility or by deleting and then re-creating the partition during the
text-mode portion of Setup. After you create the new partition, quit
Setup by pressing F3, and then restart the Windows installation
process. The ATAPI subsystem now correctly shows approximately 137 GB
of hard disk space."

Now... I may be misunderstaning something here, but this seems like a
"catch 22" situation (if I follow the above quoted bullet): If I
delete the problematic partition as suggested, I loose my OS and with
it I loose the "EnableBigLba registry value" that I just set. Then I
re-create the partition during the text-mode portion of Setup. But
this "text-mode portion of Setup" is the same exact one with which I
created the partition that I just deleted (and which allegedly is the
source of my problems). That is, it has no knowledge (yet) of the
"EnableBigLba registry value" - at least not after I re-install W2K
and add this registry value...

Do you see the problem here? I need to fully install W2K in order to
enable "Big LBA", but by that time it's already too late...

Or did I misunderstand the quoted bullet? Any idea how to proceed from
here?

Thanks,
Sam

 

[DL]

I also saw the catch 22.
You can create a slipstreamed win2k install to include latest SP. That way
the biglba would be enabled at install.
On my sys, in which the win2k installation disk was pre biglba, allthough I
had installed SP's prior to fitting a large HD. So when I fitted a large
disk all was OK I did notice, on checking the MS kb article, that one of the
reg.key values that supposedly had to be enabled/configured, to enable
biglba, was missing.
My sys worked ok, but I did manually - actually used I believe a Maxtor
utility - to write the key value in.
good luck

 

[Silence Seeker]

For the record, after creating the value EnableBigLba as the KB 305098
article suggested, all the problems disappeared. It has been over 10
days and still counting...