Report defects through your support channel. Use the following instructions
to collect data to submit with your problem report.
Run the hpvmcollect command to gather information
about the guest before modifying any guest. Preserve the state of the VM Host
and Integrity VM to best match the environment when the VM Host failed.
If multiple guests are running, run the hpvmcollect command
for guest that was running at the time.
After the hpvmcollect archive is stored
on the VM Host, reboot the guest that caused the VM Host to crash.
Run the hpvmcollect command on the guest
again. Include this information in the hpvmcollect archive
from the VM Host.
Report the information through your support channel.
Table 11–1 describes the options to the hpvmcollect command.
Table 11-1 Options to the hpvmcollect Command
| Option | Description |
|---|
| —P vm-name | Specifies the virtual machine name, where vm-name is
the name of the virtual machine. |
| —p vm-number | Specifies the virtual machine number, where vm-number is
the number of the virtual machine. |
| -s host | Specifies a hostname to receive the archive, which is copied using
the scp command. Verify that you can log in to the host
without a password. |
| —n crash-dump | Specifies the number of crash dumps to copy to the archive. By default,
the hpvmcollect command copies the latest crash dump directory
(based on the bounds file). This option can be used only with the -c option. |
| -d dir | Specifies a target directory in which to create the hpvmcollect_archive directory. |
| -b report-number | Specifies the archive name with the specified label. If an archive
with the same name exists, it is renamed by appending a time stamp to the
original name before the new archive is created. |
| -c | Includes the latest crash dump directory in the archive. This option
is used if the guest or the VM Host fails or hangs. |
| —f | Forces an archive to be overwritten, if it exists, rather than renamed
with an appended time stamp. |
| —h | Displays the help message for the hpvmcollect command. |
| —l | Leaves the collected information in a directory rather than in an archive
file. The directory name follows the same naming convention as the archive
name. |
If the VM Host hangs, generate a crash dump using the TC command
on the VM Host console. When the VM Host crashes, it tries to dump a predefined
set of memory pages into the crash dump area, including those that belong
to Integrity VM. This is crucial to collecting a successful crash dump to
analyze Integrity VM problems.
The hpvmcollect command is a shell script that can
be run on either the VM Host or the guest to gather system information, log
files, Integrity VM logs, and configuration files for later analysis.
Because the hpvmcollect command collects generic
Integrity VM and HP-UX operating system and system information, it may not
collect all the information needed to analyze the source of the problem. Make
sure that all the relevant information is included in the collection. For
example, if the guest is running an Oracle® application, include the Oracle
application log files and configuration.
By default, the hpvmcollect command creates a directory
called hpvmcollect_archive in your current directory,
and copies and collects all the Integrity VM and VM Host information. For
example, to gather information for a guest named compass1 on
the VM Host, enter the following command:
# hpvmcollect -P compass1 |
This command creates a directory called hpvmcollect_archive in
your current directory (if it does not already exist) and then collects information
about the VM Host crash dump. The information is then put into a tar file
format (if there is a crash dump) or tar.gz file
format (if there is no crash dump). Do not modify the guest configuration
before running the hpvmcollect command.
If you do not want to archive the collection into tar.gz but
simply want to examine the contents of the collection, use the -l option
to leave the contents as they are.
If the VM Host failed, use the -c option
to collect crash dump files as well. Because the -c option
collects the latest crash dump, use the -n option to specify
a crash dump number.
Use the -d option to specify a different
directory in which to store the hpvmcollect_archive.
For example, to collect information about compass1,
enter the following command:
# hpvmcollect -c -n 21 -d /tmp/hpvm_collect_archive compass1 |
This command collects information about the guest called compass1 using
crash dump number 21. The final archive is under /tmp/hpvm_collect_archive directory.
The following is an example of hpvmcollect output on the
VM Host:
 |
# hpvmcollect -P compass1
HPVM host crash/log collection tool version 0.8
Gathering info for post-mortem analysis of guest 'test' on host
Collecting I/O configuration info ................................... OK
Collecting filesystem info .......................................... OK
Collecting system info .............................................. OK
Collecting lan info ................................................. OK
Running lanshow ..................................................... NO
Collecting installed sw info ........................................ OK
Collecting command logs ............................................. OK
Collecting messages from vmm ........................................ OK
Collecting lv info .................................................. N/A
Collecting vgdisplay info ........................................... OK
Collecting vxprint info ............................................. OK
Collecting disk info ................................................ N/A
Collecting passthru disk info ....................................... N/A
Collecting file backing store info .................................. N/A
Copying guest's log file ............................................ OK
Copying guest's tombstone file ...................................... N/A
Copying guest's console log file .................................... OK
Copying hpvm configuration .......................................... OK
Copying hpvm control script ......................................... OK
Copying guest's config file ......................................... OK
Getting status of the guest ......................................... OK
Getting detailed status of the guest ................................ OK
Getting guest's entitlement ......................................... OK
Copying guest's config file change log .............................. OK
Copying guest VM crash image ........................................ OK
Copying host vmunix image ........................................... OK
Copying host hpvmmkimage image ...................................... N/A
Copying VMM image ................................................... OK
Copying hpvmdvr image ............................................... OK
Copying hpvmntdvr image ............................................. OK
Copying NVRAM image ................................................. OK
Collecting IPMI logs ................................................ OK
Collecting crash dump ............................................... NO
Running crashinfo ................................................... NO
Collecting tombstone ................................................ NO
Collecting system message buffer .................................... OK
Collecting system syslogs ........................................... OK
Collecting measureware logs .......................................... OK
Finished with the collection
Tar archiving and compressing ....................................... TGZ
Remote copying the archive ......................................... NO
The collection is
"/tmp/sornson/hpvmcollect/hpvmcollect_archive/test_Sep.28.06_095249EDT.tar.gz" |
|
If the command results in an error message like the following, you are
out of disk space in the current directory or in the directory you specified
with the -d option:
msgcnt 10 vxfs: mesg 001: vx_nospace - /dev/vg00/lvol5 file system full(1 block
extent)
Tar: end of tape
Tar: to continue, enter device/file name when ready or null string to quit. |
Use a file system with enough free space for the archive, especially
when you use the -c option.
When you use the hpvmcollect command on the guest,
do not specify the guest name. By default, the guest name is used as an archive
directory name. You can use the -d option
to specify the archive name. The following is an example of the hpvmcollect when
it is run on the guest compass1:
|
compass1# hpvmcollect -c
HPVM guest crash/log collection tool version 0.8
Gathering info for post-mortem analysis on guest (hostname 'compass1')
Collecting I/O configuration info ................................... OK
Collecting filesystem info .......................................... OK
Collecting system info .............................................. OK
Collecting lan info ................................................. OK
Running lanshow ..................................................... NO
Collecting installed sw info ........................................ OK
Collecting crash dump 1 ............................................ OK
Running crashinfo ................................................... NO
Collecting tombstone ................................................ N/A
Collecting system message buffer .................................... OK
Collecting system syslogs ........................................... OK
Collecting measureware log .......................................... N/A
Finished with the collection
Tar archiving and compressing ....................................... TAR
Remote copying the archive ......................................... NO
The collection is
"//hpvmcollect_archive/compass1_Sep.29.05_122453PST.tar" |
|
Additional data collected by the hpvmcollect command
includes log files (guest, Integrity VM, and VM Host) as well as VM Host system
information, including output from the ioscan, lanscan,
and swlist commands. The hpvmcollect command
also collects information about devices used by the guest. Output from the crashinfo and lanshowcommands
are included, if available.
The hpvmcollect command records device information
in the following files:
config/
host.diskinfo
host.fsinfo
host.ioscan
host.laninfo
host.sysinfo |
Printable version
