Unable to manage applications

 

You are trying to manage applications, for example starting, stopping, installing an application, and are having problems.

You get these problems while installing the application, or after installing the application or after a server restart.



àData to collect:

The following logs and output can be helpful in determining why you can’t mange WebSphere Application Server services

Deployment manager logs
Node agent logs
Application server logs

àWhat to look for:

When error occurs when you are trying to manage an application, following are things you can do

Verify that there are no resource or classpath errors
Verify that the application was installed correctly
Verify that the configuration repository is not corrupted

Verify that there are no resource or classpath errors

The application server log files are the first place to look as any problems with the application itself, such as classpath issues or resource issues, are reported there. The error message from the administrative console tells you which node and server to check, as shown in Figure 4-7 on page 142. Should you find such an error, work with your application developers to resolve the application coding or resource problem.

Verify that the application was installed correctly

If the application fails to start on a server after it has been installed, the installation might not have completed successfully. The installation might have appeared to be successful, but when you try to start the application, it fails with a
message in the administrative console as shown in Figure 4-7 on page 142.

———————————————

Figure 4-7 Message from administrative console when application fails to start
This message also appears in the deployment manager SystemOut.log, as
shown in Example 4-17.
———————————————

Example 4-17 Message in deployment manager log when application fails to start
[7/12/05 9:44:07:729 EDT] 000001ea MBeanHelper E Could not invoke an operation on object:
WebSphere:platform=dynamicproxy,cell=kll6571Cell01,version=6.0.1.2,name=ApplicationManager,mbea
nIdentifier=ApplicationManager,type=ApplicationManager,node=kll6571Node01,process=server01
because of an mbean exception: com.ibm.ws.exception.ConfigurationWarning: Application HelloApp
not installed
———————————————————–

The application server SystemOut.log shows a similar message, as shown in
Example 4-18.

——————————————————————————

Example 4-18 Message in application server log when application fails to start
[7/12/05 9:44:07:379 EDT] 00000039 ApplicationMg W WSVR0215W: Starting application, HelloApp,failed. The application is not installed.
—————————————————————–

In this example, the application was not installed on this node because automatic synchronization was disabled for the node. The administrator did not perform manual synchronization when saving the application installation changes to the master configuration repository. You can resolve this problem by synchronizing the node.

àVerify that the configuration repository is not corrupted

Corruptions in the configuration repository can also cause enterprise applications to not start. For example, a node’s serverindex.xml file contains a listing of all servers and ports in a cell with which the node might need to communicate. It also lists applications that are deployed on the servers in those nodes.

If this file becomes corrupt and the list of applications is lost, then the applications will not be started on the server in this node. When the server process starts, it thinks that the application is not installed and does not try and start it. If you try and start the application from the administrative console, you see the message that is shown in Figure 4-7 on page 142. The application server
logs tell you that the application is not installed, as shown in Example 4-18 on page 142.

You can resolve this problem by restoring the serverindex.xml file from a backup or forcing the deployment manager to transfer the uncorrupted copy of the file from the master repository. You would do this by updating the master copy of the file but without changing it so that it appears to have been updated to the deployment manager.