XKL DarkStar DRA, Руководство пользователя

Страница: 65 / 72

55
DarkStar User Guide
4: Troubleshooting
www.xkl.com
4.1.2.4 Automatic recovery from a warm restart
No operator action. The system will restart itself, using the configuration you have previously saved (see above) in system flash
memory.
•
Traffic is not affected, even though the system may not be responding to commands, and you may lose the console for a
while. The system’s management software (DXMOS) runs separately from the crossbar switch; failure of software or cor-
rupted flash memory on the management side does not affect traffic through the crossbar.
• You will not be able to make changes to the running configuration until DXMOS has reloaded.
•
The warm restart event is notified to the event log, and also via SNMP. An unexpected warm restart creates a system dump
file (
/dump/dump.exe
).
•
All warm restart events should be investigated, even though the system will restart itself and resume operations as quickly
as possible.
4.1.2.5 Automatic recovery from a cold boot or power recycle
No operator action.
• Traffic is disrupted since power is lost on all interfaces. Once the console has come back, you should verify (use s
how
interface
) that optical traffic has resumed. Check the report output for LOL and LOS markers that show the optical sig-
nal is lost. Check the receive laser power levels, if they are around -40dBm no light is reaching them.
•
No system dump file is available.
4.1.2.6 Manual crash recovery only if automatic recovery fails
You will need to intervene manually in the boot process if the system can’t restart due to a corrupt or invalid saved config file,
or corrupted system flash memory.
• Intervene from the console by entering
CTRL+C
before the boot process completes.
4.1.2.7 Diagnosis
You should be able to provide the following information to your technical support team to help them diagnose the cause of
the crash:
•
The crash notification from the console, if it is available.
• The system dump file
/dump/dump.exe.
The file must be uploaded to a TFTP server, then transferred to XKL via email or
FTP upload (the file is large), for analysis. The dump file cannot be debugged by the customer.
• The event log, from the event log server, leading up to the crash. This event log is persistent; the console event log buffer
clears when the system reloads.
• The output of a
show tech
command which itemizes the status of all system resources.
•
A review of any environmental issue (power, cooling, physical activity in the rack or nearby racks) at the crash location
before the crash.
4.2 Console Procedures
Most troubleshooting is done from remote consoles, using
show
commands to assess the status of the system or network,
and
configuration
(
config
) commands to revise configuration settings. Issues fall into two categories:
•
System Issues. Many system issues are due to errors or changes in management configuration settings. This includes oper-