Crash
processing for a virtual partition is similar to the crash processing
of a non-vPars OS: the OS is quiesced, portions of memory are written
to disk, and in the case of vPars, resources are released to the vPars
Monitor.
When the vPars Monitor crashes, a vPars Monitor
dump is created. By default, kernel dumps are not saved.
When there is a HPMC or MCA or when a TOC is issued,
the virtual partitions are launched for crash processing. On PA-RISC,
this occurs after saving the vPars Monitor dump, and on Integrity,
this occurs before saving the vPars Monitor dump.
When you enter the crash user interface:
To let the crash processing
continue, do nothing.
To enter the crash user
interface, press any key on the console.
 |
| |
|
 | NOTE: HP recommends that you let crash processing continue. |
|
| |
|
Crash User Interface |
|
If you enter the crash user interface, you will
see messages similar to the following on the console:
Virtual Partition Activity at Time of Crash
partition 0 (vpar1): active
partition 1 (vpar2): active
partition 2 (vpar3): inactive
The active partitions will be invoked to perform
crash handling. A soft reset will then be generated
to allow additional debugging.
TO OVERRIDE THIS BEHAVIOR, PRESS A KEY WITHIN 10 SECONDS....
CRASH PROCESS STOPPED.
Crash Command Menu
1. Examine memory contents
2. Continue with default crash handling
3. Cause Monitor crash dump to different device
4. Soft reset the machine (memory preserved)
5. Hard reset the machine (memory not preserved)
6. Launch partition 0 (vpar1) for crash processing
7. Launch partition 1 (vpar2) for crash processing
Enter number (1-7): |
The menu choices mean:
displays memory from <starting address> for <n> 32-bit words. For example:
Enter Address: 0x1000 4
0x00001000 0x00000000 0x1200a000 0xaa400000 0x00000000
Enter Address: quit |
continues with the default
crash handling
(PA-RISC only) allows
you to chose an alternate device to which the vPars Monitor dump is
written. The alternate device must contain the pre-allocated file /stand/vpmon.dmp. The file vpmon.dmp is automatically created in/standof a partition’s
boot disk by the vPars startup script.
soft resets the current
hard partition[4].
hard resets the current
hard partition.
boots the specified virtual
partition for crash processing
boots the specified virtual
partition for crash processing
If you chose to invoke a virtual partition for
crash processing or to examine memory contents, you will be returned
to this menu after those actions are completed (assuming no new crash
event is encountered).
Directory Location and Filenames |
|
On PA-RISC, the vPars Monitor dump is written
to the pre-existing file called /stand/vpmon.dmp.
On Integrity, the vPars Monitor dump is written
to a file called vpmon.dmp that is created in
the EFI partition of the vPars Monitor boot disk. The file will be
at fsN:/efi/hpux/vpmon.dmp.
When the virtual partition that owns the vPars
Monitor boot disk is booted, the following files are created automatically
in /var/adm/crash/vpar (where n is a number representing the nth occurrence
of a dump):
| vpmon.n | copy of the executable image of the vPars Monitor at the
time of the dump |
| vpmon.dmp.n | copy of the
vPars Monitor dump file |
| summary.n | an analysis
of the crash including PIM info for each processor |
vPars Monitor Dump Analysis Tool |
|
Because the vPars Monitor is not a HP-UX kernel,
you cannot use a kernel dump analysis tool to examine a vPars Monitor
dump file. Contact your HP Support Representative to analyze the vPars
Monitor dump file.
Kernel Dumps |
|
If a TOC (transfer of control) or HPMC (High Priority
Machine Check) for the entire hard partition is generated, a kernel
dump will not automatically be saved to /var/adm/crash for those partitions that have not previously had a kernel dump
occur. You can save their dumps to /var/adm/crash by performing the following on each of those virtual partitions:
(11i
v2 and above) Beginning with HP-UX 11i v2 and therefore vPars A.04,
the savecrash processing has changed. Instead of copying the kernel
file that was in use during the crash, the directory /stand/crashconfig is copied. Therefore, prior to executing the crashconf and savecrash steps below, create the /stand/crashconfigdirectory using
# kconfig -s crashconfig
Or if the kernel configuration used in the last
boot is different from the current kernel configuration, use the -c
option. For example, if the saved kernel configuration is named kc.custom,
the command is
# kconfig -c kc.custom crashconfig
For more information on using the kconfig command, see the manpages kconfig(5) and kconfig(1M)
Obtain
of list of dump devices, noting the DEVICE and OFFSET information:
# crashconf -v
DEVICE OFFSET(kB) SIZE(kB) LOGICAL VOL. NAME
-------------- ---------- --------- ------------- --------------
31:0X022000 314208 4194304 64:0X000002 /dev/vg00/lvol2 |
The DEVICE is 31:0X022000, and the OFFSET is 314208.
Map
the device number from the DEVICE information
to a device file:
# ls -l /dev/dsk /dev/disk | grep "31.*022000"
brw-r----- 1 bin sys 31 0x022000 Oct 13 2001 c2t2d0 |
The corresponding device file is /dev/dsk/c2t2d0.
Using
the OFFSET information and the device file,
save the dump to /var/adm/crash:
# savecrash -r -f -D /dev/dsk/c2t2d0 -O 314208 |
Printable version
