Troubleshooting the cluster

 

Troubleshooting the cluster

 

+

Search Tips   |   Advanced Search

 

  1. Overview
  2. "Error 503: Failed to load target servlet [portal]" received when attempting to access portal
  3. WebSphere Portal does not start due to missing class file: com/ibm/wps/services/puma/AccessBean
  4. addNode command fails with "doMergeConfig failed" error
  5. Performance in cluster degraded due to insufficient threads in default thread pool
  6. Global settings cannot be set in a cluster by using the XML configuration interface

 

Overview

For problems with the operation of IBM WebSphere Portal, we can review a list of problems addressed by interim fixes and service packs at the WebSphere Portal support site.

We can also review WebSphere Application Server support site for known issues, fixes, and workarounds.

Determine node of failure and inspect portal logs, as usual

If undetermined, start from the top. Eliminate the variables. Isolate the problem.

Note that Dmgr is only involved during configuration and administration processes. Not used at all for runtime requests. All requests go straight to the nodes themselves.

  1. For problems in initial installation and configuration process, check:

    • <portal_server_root>/log/wpsinstalllog.txt
    • <portal_server_root>/log/configTrace.log

  2. Check the HTTP Server plugin log to verify there are no problems with the plugin and request routing

  3. Stop Portal cluster members to isolate the problem to a specific cluster member, or modify plugin to direct traffic to one member. Does it happen on all cluster members, or just one?

  4. Check the cluster member SystemOut.log & SystemErr.log. Default location of logs for secondary nodes is

    <profile_root>/logs/<server_name>

  5. Use DMgr Admin Console to validate configuration

  6. For portlet deployment and synchronization errors, also check nodeagent logs and Deployment Manager logs

    • Default location for nodeagent logs is <profile_root>/logs/nodeagent
    • Default location for DMgr logs is <profile_root>/logs/dmgr

 

Problem: "Error 503: Failed to load target servlet [portal]" received when attempting to access portal

After federating a WebSphere Portal node and then attempting to access the portal, you might receive...

Error 503: Failed to load target servlet [portal]

...in the browser. The WebSphere Portal log file might have...

2004.09.15 11:18:08.428 E com.ibm.wps.engine.Servlet init EJPFD0016E: Initialization of service failed. - StackTrace follows...

Solution:

When this error occurs, it could indicate that you have not updated the deployment manager configuration for the new WebSphere Portal node. To perform this update...

  1. Ensure that the CellName property in wpconfig.properties is set.

    The cell name can be identified by examing the directory...

    was_config_root/cells/cell_name

    ...on the node

  2. Update the deployment manager configuration...

    The task...

    post-portal-node-federation-configuration

    ...requires accurate database information in the wpconfig_dbdomain.properties file.

    Windows and UNIX:

    cd portal_server_root/config
    ./WPSconfig.sh post-portal-node-federation-configuration

    i5/OS:

    cd portal_server_root/config directory:
    WPSconfig.sh post-portal-node-federation-configuration

 

WebSphere Portal does not start due to missing class file: com/ibm/wps/services/puma/AccessBean

When attempting to access a WebSphere Portal cluster for the first time through an external Web server, you might receive an Error 503 response. In addition, the following messages are generated in the WebSphere Portal log file:

com.ibm.hrl.pse.portlets.WebScannerManager getSecretKeyFromCredentialVault java.lang.NoClassDefFoundError: com/ibm/wps/services/puma/AccessBean 

Solution:

This problem can occur when the PortalAdminId property values specified for each node do not match and you enabled security with an LDAP registry.

In this situation, the value of PortalAdminID is the fully-qualified DN of the WebSphere Portal administrator. Because the DN is case sensitive, differences in case between PortalAdminID values on the various nodes can cause a problem.

For example, the error can be generated if you have two nodes in the cluster with the following values for PortalAdminId in the wpconfig.properties files:

  • Node 1: uid=wpsadmin,ou=People,dc=raleigh,dc=ibm,dc=com
  • Node 2: uid=wpsadmin,ou=people,dc=raleigh,dc=ibm,dc=com

To correct this problem, ensure that you have used the proper case for the value of the PortalAdminId property and that the property values match on each node in the cluster.

 

addNode command fails with "doMergeConfig failed" error

When attempting to federate a WebSphere Portal node, the addNode command might fail with the following error:

    ADMU0040E: Exception on MBean invocation...
    ...
    com.ibm.websphere.management.exception.AdminException: doMergeConfig failed 

This error is caused in turn by a "Too many open files" error.

Solution:

Increase the number of files that we can open at the same time by using the ulimit command.

 

Performance in cluster degraded due to insufficient threads in default thread pool

To confirm that the thread pool capacity is contributing to any performance issues, verify whether the Tivoli Performance Viewer provided with WebSphere Application Server shows the Percent Maxed metric to remain consistently in the double digits.

Solution:

Increase the maximum number of threads allocated for the default thread pool.

 

Global settings cannot be set in a cluster by using the XML configuration interface

Problem: The XML configuration interface fails with the following error when you try to set global settings in a clustered environment.

<status element="[global-settings]" result="failed">    
    <message id="EJPXA0043E">com.ibm.wps.command.xml.XmlCommandException: 
                 EJPXA0043E: An error occurred while creating or updating the resource. 
                 [global-settings]</message>     
    <message id="EJPEB0002E">com.ibm.wps.command.CommandException: 
                 EJPEB0002E: An exception occurred.</message>      
    <message>java.security.PrivilegedActionException: java.lang.NullPointerException</message>
</status> 

Cause:

The global settings are stored in the WAS. For changes to the global settings a WAS API call is invoked. During this call the following exception is thrown in the clustered environment:

Caused by: java.lang.NullPointerException     
     at  at com.ibm.wps.services.ServiceHelper$4.run(Unknown Source)     
     at  at com.ibm.ws.security.auth.distContextManagerImpl.runAs(distContextManagerImpl.java:2782)
    ... 43 more 

Solution:

To configure these settings, configure them directly in the administrative console, or use the normal configuration task update-properties.

 

Parent Topic

Search the product documentation for a solution