Start script might display error messages like Could not start Grouping service or Could not start Platform Agent.
You might see these error messages when you run the es-guiinst command on the Sun Fire system controller or when the Sun Management Center components are started on a heavily loaded machine.
In addition, if a particular component process does not start within a specific amount of time, the start script might display the same error message, even if the component was successfully started.
This problem has been observed for both the Grouping service and the Platform Agent components.
The start script does the following actions:
Attempts to start the component process
Sleeps n seconds
Checks the process status and reports success or failure
A heavily loaded machine can delay the results of the status check, which results in an erroneous status check failure. The error occurs because the results were not received within n seconds.
Workaround: Complete the following workaround:
Verify whether the following processes are running:
Agent
Topology service
Trap-handler service
Configuration service
Event-handler service
Metadata service
Platform Agent
Instances of Platform Agent
Type the following command:
/usr/bin/ps -ef | grep esd |
(On Solaris 10) The command would be /usr/bin/ps -fz global | grep esd on a global zone and /usr/bin/ps -fe | grep esd on a non-global zone.
To verify whether the Grouping service is running, type the following command:
/usr/ucb/ps —wwxa | grep SunMCService |
To verify whether the Java server is running, use the following command:
/usr/ucb/ps —wwxa | grep java | grep ServerMain |
To verify whether the web server is running, type the following command:
/usr/ucb/ps —wwxa | grep java | grep tomcat |
To verify whether the Hardware Service is running, type the following command:
/usr/ucb/ps —wwxa | grep java | grep HWDS |
If a service is not running, reduce the machine load and restart the Sun Management Center processes.
Assume that the command-line interface (CLI) batch mode is run on system2, where back-end processes are created.
CLI batch mode does not work in the following scenarios:
The user uninstalls and reinstalls the server on system1 and tries to connect to system1 by using the batch mode from system2.
A temporary network disconnection has occurred between system1 and system2. The user is trying to connect to system1 by using the batch mode from system2.
CLI batch mode does not work because uninstalling the server does not kill the back-end process.
Workaround: Follow these steps to kill the back-end process:
View the running back-end processes:
# cat /var/opt/SUNWsymon/cli/process-file
The process-file has the following format for each process:
<username>:<hostname>:<cpid>:<jpid>, where cpid is the C process id and jpid is the Java process id. Note the IDs of the processes that you want to kill.
Kill the back-end process:
# kill -9 <cpid> <jpid>
If you attempt to uninstall Sun Management Center on Windows using Start -> Settings -> Control Panel -> Add/Remove Programs, the SunMC.exe is not removed from the %WINDIR% directory. This is because the Windows platform does not allow deletion of an executing program and displays “Access Denied. Source file in use” message.
Workaround: Uninstall Sun Management Center on Windows by using es-uninst or es-guiuninst.
The maximum length of the security seed is 8 characters in GUI setup and 16 characters in CLI setup.
If the Sun Management Center agent is intended to be used as a stand-alone SNMPV3 agent, then it is recommended to configure the agent layer using CLI setup.
Workaround: Use CLI setup for Sun Management Center if the security seed of more than 8 characters length is desired.