Chkdsk w/ Hardware Raid 1

[Jisha]

Am I interpreting these Chkdsk reports correctly?

I've been experimenting with Hardware Raid 1 on my new w2k box (Promise
FastTrak TX4000)... and still pretty green with it.

I ran Chkdsk to see how the system was going to handle the Mirrored Pair...
To my surprise, it treated two volumes differently.

There is currently only one HDD pair with the basic w2k installation, split
into 2 volumes C: & D:

Chkdsk on C: produced only 1 report... no errors found.
Chkdsk on D: produced 2 reports, I presume for each side of the mirrored
pair... first part found and fixed some errors, second part error free.
I ran it again, this time only 1 error free report each for both C: & D:.

It appears that Chkdsk IS checking both halves of the mirror... I see the
HDD LED active for both drives... (which are on separate IDE channels)
but only produces separate reports if errors are found... otherwise just one
single report (since they would be identical anyway...)

Is this the correct interpretation?

Chkdsk reports are included below.

TIA,
Chris


************

C:=========================

Event Type: Information
Event Source: Winlogon
Event Category: None
Event ID: 1001
Date: 1/22/2004
Time: 1:34:01 PM
User: N/A
Computer: BIG
Description:
Checking file system on C:
The type of the file system is NTFS.
Volume label is SYSTEM.

A disk check has been scheduled.
Windows will now check the disk.
Cleaning up minor inconsistencies on the drive.
Cleaning up 13 unused index entries from index $SII of file 0x9.
Cleaning up 13 unused index entries from index $SDH of file 0x9.
Cleaning up 13 unused security descriptors.
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.

25165790 KB total disk space.
2544636 KB in 7993 files.
2032 KB in 488 indexes.
0 KB in bad sectors.
75346 KB in use by the system.
65536 KB occupied by the log file.
22543776 KB available on disk.

4096 bytes in each allocation unit.
6291447 total allocation units on disk.
5635944 allocation units available on disk.

Windows has finished checking your disk.
Please wait while your computer restarts.




d:=========================

Event Type: Information
Event Source: Winlogon
Event Category: None
Event ID: 1001
Date: 1/22/2004
Time: 1:34:01 PM
User: N/A
Computer: BIG
Description:
Checking file system on D:
The type of the file system is NTFS.
Volume label is PROGRAMS.

A disk check has been scheduled.
Windows will now check the disk.
Cleaning up 6 unused index entries from index $SII of file 0x9.
Cleaning up 6 unused index entries from index $SDH of file 0x9.
Cleaning up 6 unused security descriptors.
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.

33447329 KB total disk space.
111400 KB in 461 files.
164 KB in 110 indexes.
0 KB in bad sectors.
69341 KB in use by the system.
65536 KB occupied by the log file.
33266424 KB available on disk.

4096 bytes in each allocation unit.
8361832 total allocation units on disk.
8316606 allocation units available on disk.


Checking file system on D:
The type of the file system is NTFS.
Volume label is PROGRAMS.

A disk check has been scheduled.
Windows will now check the disk.
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.
Windows has checked the file system and found no problem.

33447329 KB total disk space.
111404 KB in 462 files.
164 KB in 110 indexes.
0 KB in bad sectors.
69341 KB in use by the system.
65536 KB occupied by the log file.
33266420 KB available on disk.

4096 bytes in each allocation unit.
8361832 total allocation units on disk.
8316605 allocation units available on disk.

 

[Jeremy Winston]

I've been experimenting with Hardware Raid 1 on my new w2k box (Promise
FastTrak TX4000)... and still pretty green with it.
I ran Chkdsk to see how the system was going to handle the Mirrored Pair...
To my surprise, it treated two volumes differently.
There is currently only one HDD pair with the basic w2k installation, split
into 2 volumes C: & D:
Chkdsk on C: produced only 1 report... no errors found.
Chkdsk on D: produced 2 reports, I presume for each side of the mirrored
pair... first part found and fixed some errors, second part error free.
I ran it again, this time only 1 error free report each for both C: & D:.
It appears that Chkdsk IS checking both halves of the mirror... I see the
HDD LED active for both drives... (which are on separate IDE channels)

Well of course both LEDs should light; both drives are being accessed.

but only produces separate reports if errors are found... otherwise just
one single report (since they would be identical anyway...)
Is this the correct interpretation?

Everything I've read says hardware RAID hides the true HD config from
the OS, so as far as the OS is concerned, it's dealing with a single
drive.

I have no explanation for why CHKDSK would generate two separate reports
for D:, unless you ran it twice.

As I understand it, the only way to get CHKDSK to see each drive in the
mirror separately is to break the mirror.

-Jeremy

 

[Jisha]

In-line comments:

Everything I've read says hardware RAID hides the true HD config from
the OS, so as far as the OS is concerned, it's dealing with a single
drive.

yeah... that's what I thought too... which was why I was curious about how
system would handle it...

I'm trying to figure out HOW I'll know about any disk problems on a
particular disk if the OS truly sees the disk-pair as one HDD.
When I run Chkdsk (or Defrag) what's actually happening "behind the scene"
.... the OS thinks it's Checking or Defragging one Volume, but there are
actually two... so the RAID controller _must_ be playing some additional
role here...

I have no explanation for why CHKDSK would generate two separate reports
for D:, unless you ran it twice.

sorry... it didn't generate 2 reports in the sense of 2 separate entries in
the event log...
It ran once for D:, but the report was "twice as long" as normal...
indicating 2 separate instances of a Chkdsk run with 2 separate
summaries.... (presumably for each half of the mirror) which were in turn
merged into one report in the event log... . ( I've re-attached that report
below. )

It sure looks to me that the reason an expanded report was generated was
because one half of the mirror HAD file-system errors, and the other half
didn't... which really doesn't make sense to me from the perspective that
both drives are supposed to be identical... errors and all I presume... (and
which, if correct, then raises a reliability question of _why_ one drive had
file-system errors, and the other did not.)

But, what about bad sectors and other physical drive problems not
necessarily related to data being written to the drive by the RAID
controller? In this sense, it's perfectly reasonable that there could be
differences between the status of each half of the mirror... with the case
of one of the drives failing as an extreme example. So, in this case, what
should I expect to see in the event log?

Perhaps it's this particular RAID controller card... maybe it's "smart"
enough to check both disks when a Chkdsk is called for... and takes over the
job of writing to the event log?

Like I said, I'm new to a RAID setup... so I can't be sure I'm interpreting
this properly...


************
Chkdsk report for D: from event log:

Event Type: Information
Event Source: Winlogon
Event Category: None
Event ID: 1001
Date: 1/22/2004
Time: 1:34:01 PM
User: N/A
Computer: BIG
Description:
Checking file system on D: <<<<<<<<<<<<<<<=============================
The type of the file system is NTFS.
Volume label is PROGRAMS.

A disk check has been scheduled.
Windows will now check the disk.
Cleaning up 6 unused index entries from index $SII of file 0x9.
Cleaning up 6 unused index entries from index $SDH of file 0x9.
Cleaning up 6 unused security descriptors.
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.

33447329 KB total disk space.
111400 KB in 461 files.
164 KB in 110 indexes.
0 KB in bad sectors.
69341 KB in use by the system.
65536 KB occupied by the log file.
33266424 KB available on disk.

4096 bytes in each allocation unit.
8361832 total allocation units on disk.
8316606 allocation units available on disk.


Checking file system on D: <<<<<<<<<<<<<<<=============================
The type of the file system is NTFS.
Volume label is PROGRAMS.

A disk check has been scheduled.
Windows will now check the disk.
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.
Windows has checked the file system and found no problem.

33447329 KB total disk space.
111404 KB in 462 files.
164 KB in 110 indexes.
0 KB in bad sectors.
69341 KB in use by the system.
65536 KB occupied by the log file.
33266420 KB available on disk.

4096 bytes in each allocation unit.
8361832 total allocation units on disk.
8316605 allocation units available on disk.

 

[Jisha]

One more thing I failed to notice before...

The 2 parts of the Chkdsk report show slightly differing available space and
allocation units... further suggesting separate summaries for each half of
the mirror... (that's a good thing, I think...)

First Part:
33266424 KB available on disk.

4096 bytes in each allocation unit.
8361832 total allocation units on disk.
8316606 allocation units available on disk.


Second Part:
33266420 KB available on disk.

4096 bytes in each allocation unit.
8361832 total allocation units on disk.
8316605 allocation units available on disk.



.... so, am I understanding this correctly?

TIA,
Chris

 

[Jeremy Winston]

I'm trying to figure out HOW I'll know about any disk problems on a
particular disk if the OS truly sees the disk-pair as one HDD.

I have a mobo with an on-board Promise FastTrak100 Lite ctrtl w/a pair
of HDs in a RAID1 mirror. When a disk goes bad, the Promise FastCheck
monitoring util reports the array is critical & advises me to replace
the failed drive ASAP. It's not hot swap, but it is plug'n'play-- all
I have to do is reboot and go into the ctrlr's config during boot-up,
determine which drive has failed, shut down, replace the drive, reboot
and let the ctrlr rebuild the array.

When I run Chkdsk (or Defrag) what's actually happening "behind the scene"
... the OS thinks it's Checking or Defragging one Volume, but there are
actually two... so the RAID controller _must_ be playing some additional
role here...

But my point was that I don't think the CHKDSK s/w should know about the
pair of disks in h/w RAID. I.e., the ctrlr hides that fact. If there
are problems with either disk, the monitoring util that comes with the
ctrlr should report them to you independent of CHKDSK.

sorry... it didn't generate 2 reports in the sense of 2 separate entries in
the event log...
It ran once for D:, but the report was "twice as long" as normal...
indicating 2 separate instances of a Chkdsk run with 2 separate
summaries.... (presumably for each half of the mirror) which were in turn
merged into one report in the event log... . ( I've re-attached that report
below. )

It sure looks to me that the reason an expanded report was generated was
because one half of the mirror HAD file-system errors, and the other half
didn't... which really doesn't make sense to me from the perspective that
both drives are supposed to be identical... errors and all I presume... (and
which, if correct, then raises a reliability question of _why_ one drive had
file-system errors, and the other did not.)

If you had software-based RAID, I could understand the separate reports.

As for errors on one disk in h/w-based RAID, my understanding is the controller
marks corresponding sectors on both drives bad if it finds the sector bad on
either.

But, what about bad sectors and other physical drive problems not
necessarily related to data being written to the drive by the RAID
controller? In this sense, it's perfectly reasonable that there could be
differences between the status of each half of the mirror... with the case
of one of the drives failing as an extreme example. So, in this case, what
should I expect to see in the event log?

There likely will be different sectors bad on each drive, but see above.

Perhaps it's this particular RAID controller card... maybe it's "smart"
enough to check both disks when a Chkdsk is called for... and takes over the
job of writing to the event log?

Could be. I'm no expert. My experience is limited to the FastTrack100 Lite.

Like I said, I'm new to a RAID setup... so I can't be sure I'm interpreting
this properly...

Maybe a RAID guru can jump in at this point.

-Jeremy

 

[Jisha]

Hey Jeremy,
Thanks for taking the time to respond again.
....more in-line comments:

When a disk goes bad, the Promise FastCheck
monitoring util reports the array is critical & advises me to replace
the failed drive ASAP.

Oh yeah... right...

If there are problems with either disk, the monitoring util that
comes with the ctrlr should report them to you independent of
CHKDSK.

Oh... yeah... right again...
Tested that by unplugging a drive to simulate a failure...
....boy, did I get yelled at by the controller!

As for errors on one disk in h/w-based RAID, my understanding is the controller
marks corresponding sectors on both drives bad if it finds the sector bad on
either.

.....ahh... of course... makes sense...

Maybe a RAID guru can jump in at this point.

Sure... I'd love that!


Thanks again,
Chris

 

[Leonard Severt [MSFT]]

Am I interpreting these Chkdsk reports correctly?

I've been experimenting with Hardware Raid 1 on my new w2k box
(Promise FastTrak TX4000)... and still pretty green with it.

I ran Chkdsk to see how the system was going to handle the Mirrored
Pair... To my surprise, it treated two volumes differently.

There is currently only one HDD pair with the basic w2k installation,
split into 2 volumes C: & D:

Chkdsk on C: produced only 1 report... no errors found.
Chkdsk on D: produced 2 reports, I presume for each side of the
mirrored pair... first part found and fixed some errors, second part
error free. I ran it again, this time only 1 error free report each
for both C: & D:.

It appears that Chkdsk IS checking both halves of the mirror... I see
the HDD LED active for both drives... (which are on separate IDE
channels) but only produces separate reports if errors are found...
otherwise just one single report (since they would be identical
anyway...)

Is this the correct interpretation?

Chkdsk reports are included below.

TIA,
Chris


************

C:=========================

Event Type: Information
Event Source: Winlogon
Event Category: None
Event ID: 1001
Date: 1/22/2004
Time: 1:34:01 PM
User: N/A
Computer: BIG
Description:
Checking file system on C:
The type of the file system is NTFS.
Volume label is SYSTEM.

A disk check has been scheduled.
Windows will now check the disk.
Cleaning up minor inconsistencies on the drive.
Cleaning up 13 unused index entries from index $SII of file 0x9.
Cleaning up 13 unused index entries from index $SDH of file 0x9.
Cleaning up 13 unused security descriptors.
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.

25165790 KB total disk space.
2544636 KB in 7993 files.
2032 KB in 488 indexes.
0 KB in bad sectors.
75346 KB in use by the system.
65536 KB occupied by the log file.
22543776 KB available on disk.

4096 bytes in each allocation unit.
6291447 total allocation units on disk.
5635944 allocation units available on disk.

Windows has finished checking your disk.
Please wait while your computer restarts.




d:=========================

Event Type: Information
Event Source: Winlogon
Event Category: None
Event ID: 1001
Date: 1/22/2004
Time: 1:34:01 PM
User: N/A
Computer: BIG
Description:
Checking file system on D:
The type of the file system is NTFS.
Volume label is PROGRAMS.

A disk check has been scheduled.
Windows will now check the disk.
Cleaning up 6 unused index entries from index $SII of file 0x9.
Cleaning up 6 unused index entries from index $SDH of file 0x9.
Cleaning up 6 unused security descriptors.
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.

33447329 KB total disk space.
111400 KB in 461 files.
164 KB in 110 indexes.
0 KB in bad sectors.
69341 KB in use by the system.
65536 KB occupied by the log file.
33266424 KB available on disk.

4096 bytes in each allocation unit.
8361832 total allocation units on disk.
8316606 allocation units available on disk.


Checking file system on D:
The type of the file system is NTFS.
Volume label is PROGRAMS.

A disk check has been scheduled.
Windows will now check the disk.
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.
Windows has checked the file system and found no problem.

33447329 KB total disk space.
111404 KB in 462 files.
164 KB in 110 indexes.
0 KB in bad sectors.
69341 KB in use by the system.
65536 KB occupied by the log file.
33266420 KB available on disk.

4096 bytes in each allocation unit.
8361832 total allocation units on disk.
8316605 allocation units available on disk.

Chkdsk and Windows do not know about the hardware mirror drive. They do
not know that the drive even exists because the raid controller doesn't
report it. If it doesn't show up in Disk Managment then Windows does not
know about it. The raid controller is what knows about the second drive.
When chkdsk makes changes to the first drive it is the raid controller
that makes those same changes to the mirror drive. Same thing with when
you defrag the primary drive, the raid controller makes those same
changes to the mirror drive.

Leonard Severt

Windows 2000 Server Setup Team