How can I determine the physical disk that is failing?

[David Trusty]

Hi,

I'm seeing disk failures from
an L-class server.

The messages (GSP) say:

SOURCE DETAIL: 6=DISK SOURCE ID:0
PROBLEM DETAIL=3: functional failure

There are four drives in the machine.

How can I tell which one is bad?

Thanks in advance!

David

[Michael Tully]

I don't think GSP is going to tell you which disk it is.
Do you have any messages in your syslog (/var/adm/syslog/syslog.log)?
EMS event log? (/var/opt/resmon/log/event.log)

Michael
"When I have trouble spelling, it's called fat finger syndrome"

[David Trusty]

It doesn't come up far enough for me to see any O/S logs.

David

[Michael Tully]

At what point does it stop?

Perhaps you could try and start the machine without any LVM, or from the alternate disk if you have one.
From the BCH menu, select 'bo alt'

or
Stop the boot at the ten second sequence.
Interact with ISL ? yes
ISL> hpux -lm

Michael
"When I have trouble spelling, it's called fat finger syndrome"

[Eugeny Brychkov]

David,
1. can you boot machine into hpux?
2. are all disks installed HP-disks?
3. can you see all disks in 'ioscan -fn' CLAIMED and DEVICEs?
4. can you dd from these disks to /dev/null with no 'I/O error' message?
5. did you try Support Tools Manager (cstm, mstm, xstm)?
Eugeny

[Eugeny Brychkov]

David,
1. can you boot machine into hpux?
2. are all disks installed HP-disks?
3. can you see all disks in 'ioscan -fn' CLAIMED and DEVICEs?
4. can you dd from these disks to /dev/null with no 'I/O error' message?
5. did you try Support Tools Manager (cstm, mstm, xstm)?
Eugeny

[Bill McNAMARA_1]

dd if=/dev/dsk/cXtYdZ of=/dev/null

This will read the disk via devicefile and empty into the bit bucket. (you should see the led of the disk blink when it is reading if you don't understand the ioscan -fnkC disk output)

If there is an error on the disk, the command will throw out errors. Otherwise it will complete normally.

Later,
Bill

[Kyri Pilavakis]

I am dealing with a hard disk failure at this moment. I run following to check - the system needs to be up...

1 ??? ioscan ???fnC disk - Disks should all say CLAIMED . If not ?? NO-HW = faulty disk

Class I H/W Path Driver S/W State H/W Type Description
======================================================================
disk 0 10/0.4.0 sdisk CLAIMED DEVICE SEAGATE ST34573WC
/dev/dsk/c0t4d0 /dev/rdsk/c0t4d0
disk 2 10/0.6.0 sdisk CLAIMED DEVICE SEAGATE ST39236LC
/dev/dsk/c0t6d0 /dev/rdsk/c0t6d0
disk 3 10/4/4.5.0 disc3 CLAIMED DEVICE SEAGATE ST32550W
/dev/dsk/c1t5d0 /dev/rdsk/c1t5d0
/dev/floppy/c1t5d0 /dev/rfloppy/c1t5d0
disk 4 10/4/12.5.0 disc3 CLAIMED DEVICE SEAGATE ST32550W
/dev/dsk/c2t5d0 /dev/rdsk/c2t5d0
/dev/floppy/c2t5d0 /dev/rfloppy/c2t5d0
disk 5 10/12/5.2.0 sdisk CLAIMED DEVICE TOSHIBA CD-ROM XM-5401
TA
/dev/dsk/c3t2d0 /dev/rdsk/c3t2d0


2 ??? diskinfo /dev/rdsk/c?t?d? - Should provide good disk info

SCSI describe of /dev/rdsk/c0t4d0:
vendor: SEAGATE
product id: ST34573WC
type: direct access
size: 4194157 Kbytes
bytes per sector: 512


3 ??? dd if=/dev/rdsk/c?t?d? of=/dev/null bs=1024 count=1. This takes a long time so use it only on suspect drives.


Running these 3 commands tell me all the info I need to know.

[Tobias Hartlieb]

Hi,

if tte error occurs before HP-UX is launched, you can interrupt the boot process at BCH level.
Then type 'sea' (search) to check the connected devices. Afterwards, do 'sea ipl', to get devices, which also have a boot header. It might be helpful, to compare these results with the expected boot path: type 'path'

if you can find any bootable media, then start booting from this disk by typing either 'bo ' or 'bo pX' where X is the appropriate number from the 'sea ipl' output.
Answer "Interact with IPL?" with 'y'.
On ISL prompt, type 'ode'. Hopefully, ODE (Offline Diagnostic Environment) is installed.
If so, you will get the ODE prompt. From there, launche a kind of "ioscan" tool by typing 'run mapper2'. Compare this output, with the actually HW config in terms of disks..
(you can leave the tool by typing 'exit')

BTW, does any of the internal disks have a permanently LED lit? It might help to pull the disk out and plug it back in, when at BCH level...IF so, check the FW of this disk afterwards..

Also, it might help to check the Console history, by typing 'CL' on GSP prompt (CTRL-B on Console)..sometimes, you can see some messages from console here, which are in live mode disguised by the GSP alert! MAybe, the HW path of the failing disk is written somewhere..

Regards.

Tobias

[Bill McNAMARA_1]

also check root's email. There may be some diagnostics info there.

There may be some diagnostics/exercisers available from xstm/mstm also.

Later,
Bill