Delayed Write Failed - could I have lost data?

Just ignore this fool. He doesnt have a ****ing clue about Win.

Arno wrote:
Short answer: You lost data and you will lose more
data until you find out what is wrong and correct that
problem.

Arno


In comp.sys.ibm.pc.hardware.storage
wrote:
Last night I was copying files from a Windows XP laptop to a USB
external hard drive. Then when I tried to safely remove the hard
drive, I got an error saying "Generic volume cannot be stopped". I've
seen this error before and it goes away if I restart Windows, but
this time I didn't want to restart, so I just put my laptop into
hibernation, and when I brought it out of hibernation, I was able to
safely remove the drive.

Then later I copied more files to the USB drive, and when I tried to
safely remove it again, I got the "Generic volume cannot be stopped"
error again. So I put the laptop into hibernation again, but when I
brought the laptop out of hibernation, Windows displayed a dialog box
saying:

"Delayed Write Failed - Windows was unable to save all the data for
the file e:\$Mft. The data has been lost. This error may be caused
by a failure of your computer hardware or network connection. Please
try to save this file elsewhere."

My USB drive is drive E, and the $Mft apparently stands for master
file table. That error seems to say that data was definitely lost,
but when I ran a chkdsk on the drive, I got the following output:

C:\Documents and Settings\Administratorchkdsk e: /f /v /r
The type of the file system is NTFS.
Volume label is Elements.

CHKDSK is verifying files (stage 1 of 5)...
File verification completed.
CHKDSK is verifying indexes (stage 2 of 5)...
Index verification completed.
Cleaning up minor inconsistencies on the drive.
CHKDSK is verifying security descriptors (stage 3 of 5)...
Cleaning up 12 unused index entries from index $SII of file 9.
Cleaning up 12 unused index entries from index $SDH of file 9.
Cleaning up 12 unused security descriptors.
Fixing mirror copy of the security descriptors data stream.
Security descriptor verification completed.
CHKDSK is verifying file data (stage 4 of 5)...
File data verification completed.
CHKDSK is verifying free space (stage 5 of 5)...
Free space verification is complete.

625131831 KB total disk space.
577839972 KB in 1884 files.
1024 KB in 437 indexes.
0 KB in bad sectors.
87383 KB in use by the system.
65536 KB occupied by the log file.
47203452 KB available on disk.

4096 bytes in each allocation unit.
156282957 total allocation units on disk.
11800863 allocation units available on disk.

That chkdsk output seems to indicate the drive only had minor
inconsistencies and problems with security descriptors. It doesn't
say anything about data errors. Is it possible that I still lost
data on the drive?

On Oct 19, 10:46*am, "Rod Speed" wrote:
That chkdsk output seems to indicate the drive only had minor
inconsistencies and problems with security descriptors. It doesn't say
anything about data errors. Is it possible that I still lost data on the drive?

Not if the files you copied are visible.

You mean it isn't possible for a file to still exist and have a few
corrupted or missing bits - it's either there and completely intact,
or it's corrupted and totally gone?

On Oct 19, 10:46*am, "Rod Speed" wrote:
That chkdsk output seems to indicate the drive only had minor
inconsistencies and problems with security descriptors. It doesn't say
anything about data errors. Is it possible that I still lost data on the drive?

Not if the files you copied are visible.

You mean it's not possible for a file to still be there and have a few
corrupted/missing bits - it's either there and completely intact, or
corrupted and totally gone?

wrote
Rod Speed wrote

That chkdsk output seems to indicate the drive only had minor inconsistencies
and problems with security descriptors. It doesn't say anything about data
errors. Is it possible that I still lost data on the drive?

Not if the files you copied are visible.

You mean it isn't possible for a file to still exist and have a few corrupted or missing bits

Not if chkdsk doesnt whine about that file.

- it's either there and completely intact, or it's corrupted and totally gone?

You can see the file in a state that chkdsk complains about and normally
attempts to fix if your system is configured to fix problems auto.

wrote
Tim Meddick wrote

What the messages basically meant were that your laptop had
been attempting to save data to the portable drive, but that in
your trying to "out smart" the warning that the "drive could NOT
be safely removed at this time" (or similar), there was corruption
of that data that was being written to the drive - *at that time*.

I thought that hibernation would force anything in the write buffer to
get written to the drive.

Yes, it should do.

Seems like it should - isn't hibernation pretty much like the power off state,

Yes.

so all buffers should get flushed just before entering hibernation.

Yes.

Also, what if I wasn't trying to "out smart" the warning? I frequently
close the lid on my laptop just before going to bed (and my laptop is
configured to hibernate Windows if the lid is closed).

Yeah, me too. Tho I dont use mine fulltime, I use a desktop full time
and close the lid on the laptop when I wont be using it for a while.

So what if I had moved some files to the USB
drive late at night and then closed the lid?

It should be fine.

I might see the "delayed write error" when I bring it out
of hibernation the next day - that should not happen.

Yes, but one of the downsides with using Win like that, setting it to
hibernate when you shut the lid, is that eventually the system will
\need a full reboot when its got its tiny little brain rather scrambled.

Thats likely what happened with your delayed write error message
and maybe with the inability to safely remove the USB drive too.

This data corruption was limited to the file that was being
written at the time the PC lost power due to the end of the
hibernate sequence. And also, the MFT would have lost it's
validity due to the fact it could not be updated after the loss.

I'm curious how large the Windows write buffer is

Thats configurable.

- just wondering if any possible corruption could be limited
to a partial file, or multiple files if the buffer is large enough.

Yes it can and it doesnt need to be that large, it isnt that
uncommon to have more than one file being written at the
same time and you can get a delayed write failure in that
situation, usually on a mains failure with a desktop etc.

Thats one area where laptops handle that situation much better auto.

Fortunately I know the last several files I moved over to the drive
- most of them are files I downloaded off the internet, so I can
redownload those just in case. But the last file I moved over
was a text file I had been editing, and it appears to be OK...

Yeah, all the evidence you reported supports the idea that the
copy to the USB drive worked fine and that the only problem
was some glitch with the hibernate that saw the system shut
down before the last of what was written to the drive had got
out to the drive. That was in fact a bug in XP initially, it could
do that at times, not wait long enough after the last of the
updating of stuff to the drives on the file system status had
been written to the drives before actually shutting down, so
you could see a delayed write failure at times. Just status info
like whether the drive was 'dirty' or not and stuff like that, not
stuff that the user was actually deliberately writing to the drive.

The dirty status is what sees the drive do an automatic chkdsk
on startup when the system hasnt been shut down properly,
usually due to a mains flick on a desktop system etc or someone
just turning it off at the wall instead of shutting down properly.

unless the corruption has manifested itself in truncating the end
of the file - I don't quite remember what was at the end of the file.

Its very likely fine and all you saw was a hibernate glitch.

In comp.sys.ibm.pc.hardware.storage wrote:
On Oct 19, 10:13?am, "Tim Meddick" wrote:
What the messages basically meant were that your laptop had been attempting
to save data to the portable drive, but that in your trying to "out smart"
the warning that the "drive could NOT be safely ?removed at this time" (or
similar), there was corruption of that data that was being written to the
drive - *at that time*.

I thought that hibernation would force anything in the write buffer to
get written to the drive.

You cannot force data on a drive that gives you an error.
One way around that is to also store the buffer-state and contents
in the hibernation device. But that would make things slow, so
I guess MS decided against it. This may also be a valid
reson not to use hibernation at all.

Seems like it should - isn't hibernation
pretty much like the power off state, so all buffers should get
flushed just before entering hibernation. Also, what if I wasn't
trying to "out smart" the warning? I frequently close the lid on my
laptop just before going to bed (and my laptop is configured to
hibernate Windows if the lid is closed). So what if I had moved some
files to the USB drive late at night and then closed the lid? I might
see the "delayed write error" when I bring it out of hibernation the
next day - that should not happen.

Indeed. But if there is a storage problem it _will_ happen.
One of the reasons why hibernation is actually not that
good a thing, at least with this treatment of the filesystem
buffer.

This data corruption was limited to the file that was being written at the
time the PC lost power due to the end of the hibernate sequence. ?And also,
the MFT would have lost it's validity due to the fact it could not be
updated after the loss.

I'm curious how large the Windows write buffer is - just wondering if
any possible corruption could be limited to a partial file, or
multiple files if the buffer is large enough.

Basically all available memory as a worst-case.

Fortunately I know the
last several files I moved over to the drive - most of them are files
I downloaded off the internet, so I can redownload those just in
case. But the last file I moved over was a text file I had been
editing, and it appears to be OK... unless the corruption has
manifested itself in truncating the end of the file - I don't quite
remember what was at the end of the file.

Well, one way around this is to do a clean shutdown instead of
that half-baked "hibernation". Another is to do a safe remove
of the USB storage.

Arno
--
Arno Wagner, Dr. sc. techn., Dipl. Inform., CISSP -- Email:
GnuPG: ID: 1E25338F FP: 0C30 5782 9D93 F785 E79C 0296 797F 6B50 1E25 338F
----
Cuddly UI's are the manifestation of wishful thinking. -- Dylan Evans

In comp.sys.ibm.pc.hardware.storage wrote:
On Oct 19, 10:46?am, "Rod Speed" wrote:
That chkdsk output seems to indicate the drive only had minor
inconsistencies and problems with security descriptors. It doesn't say
anything about data errors. Is it possible that I still lost data on the drive?

Not if the files you copied are visible.

You mean it isn't possible for a file to still exist and have a few
corrupted or missing bits - it's either there and completely intact,
or it's corrupted and totally gone?

Usually you can get
1) missing file
2) present file partial data (anuthing between 0 Bytes there and the last
byte missing)
3) full data.

What you typically cannot get is corrypted data. It either is
there and correct or (partially) missing. Of course, if you edited a
file, the edits can be partially missing either in cronological order
or in sequence from the start of the file, depending on the editors
update strategy.

Arno
--
Arno Wagner, Dr. sc. techn., Dipl. Inform., CISSP -- Email:
GnuPG: ID: 1E25338F FP: 0C30 5782 9D93 F785 E79C 0296 797F 6B50 1E25 338F
----
Cuddly UI's are the manifestation of wishful thinking. -- Dylan Evans

John John MVP wrote:
The data was in the write cache, it wasn't flushed to the disk so it
was lost when the computer was rebooted.

Why the hell doesn't Windows flush the write cache when it goes into
hibernation?


--

Reply in group, but if emailing add one more
zero, and remove the last word.

In comp.sys.ibm.pc.hardware.storage Tom Del Rosso wrote:

John John MVP wrote:
The data was in the write cache, it wasn't flushed to the disk so it
was lost when the computer was rebooted.

Why the hell doesn't Windows flush the write cache when it goes into
hibernation?

While I think the Windows is (still) at best a toy, in all fairness
if the disk refuses to take data, flushing the write-buffer (no, it
is not a cache) is impossible.

Arno

--
Arno Wagner, Dr. sc. techn., Dipl. Inform., CISSP -- Email:
GnuPG: ID: 1E25338F FP: 0C30 5782 9D93 F785 E79C 0296 797F 6B50 1E25 338F
----
Cuddly UI's are the manifestation of wishful thinking. -- Dylan Evans

Tom Del Rosso wrote
John John MVP wrote

The data was in the write cache, it wasn't flushed to the disk so it was lost when the computer was rebooted.

Why the hell doesn't Windows flush the write cache when it goes into hibernation?

It does. It can also write some status stuff to the drive before hibernating
and can shut down too quickly before it gets written to the drive, particularly
if there is some delay writing that status stuff to the druve.

Arno wrote:
In comp.sys.ibm.pc.hardware.storage Tom Del Rosso
wrote:

John John MVP wrote:
The data was in the write cache, it wasn't flushed to the disk so
it was lost when the computer was rebooted.

Why the hell doesn't Windows flush the write cache when it goes into
hibernation?

While I think the Windows is (still) at best a toy, in all fairness
if the disk refuses to take data, flushing the write-buffer (no, it
is not a cache) is impossible.

Does the disk refuse because it has to spin up first? Windows should be
able to delay the shutdown in that case.


--

Reply in group, but if emailing add one more
zero, and remove the last word.

On 10/25/2011 11:39 PM PT, Tom Del Rosso typed:

The data was in the write cache, it wasn't flushed to the disk so it
was lost when the computer was rebooted.

Why the hell doesn't Windows flush the write cache when it goes into
hibernation?

Why the frak do we have write cache? Is it for speed?
--
"Ants can lift up to 50 times their own weight. And your monitor is
missing. Time to bring out the bugspray." --BBspot's Geek Horoscopes
(2/28/2003)
/\___/\ Ant @ http://antfarm.ma.cx (Personal Web Site)
/ /\ /\ \ Ant's Quality Foraged Links: http://aqfl.net
| |o o| |
\ _ / If crediting, then use Ant nickname and AQFL URL/link.
( ) If e-mailing, then axe ANT from its address if needed.
Ant is currently not listening to any songs on this computer.

Rod, looks like you're right about the data being fine and it was just
a weird hibernation glitch. Since I had gotten a "delayed write
failed" error, I just assumed that the drive's policy was set to
"Optimize for performance". Then right after I started this thread, I
went out of town for over a week with my work laptop, so I couldn't
check my personal laptop. Now I'm finally home and was able to check
on what the policy was set to - it is set to "Optimize for quick
removal", which means the data must have been written correctly.

Ant wrote
Tom Del Rosso wrote

The data was in the write cache, it wasn't flushed to the disk so it was lost when the computer was rebooted.

Why the hell doesn't Windows flush the write cache when it goes into hibernation?

It does.

Why the frak do we have write cache? Is it for speed?

Yep. Otherwise you'd have to wait till the sector you are writing to moves under the heads etc.

wrote:

Rod, looks like you're right about the data being fine and it was
just a weird hibernation glitch. Since I had gotten a "delayed
write failed" error, I just assumed that the drive's policy was set to
"Optimize for performance". Then right after I started this thread, I
went out of town for over a week with my work laptop, so I couldn't
check my personal laptop. Now I'm finally home and was able to
check on what the policy was set to - it is set to "Optimize for quick
removal", which means the data must have been written correctly.

Thanks for that detail, too rare IMO.