You can configure name-value pairs of data, where the name is a property key and the value is a string value that you can use to set configuration properties for a cell. Defining a new custom property for a cell enables you to configure a setting beyond that which is available in the administrative console. The following is a list of custom properties that are already available to configure for a cell.
To specify a custom property for a cell:
Following is a list of custom properties provided with the product.
Specifies, in milliseconds, the length of time the node agent waits for cluster data to be gathered after a client submits the first request for that cluster. You do not need to specify a value for this property if the IBM_CLUSTER_ENABLE_PRELOAD custom property is set to true, because in that situation, the data is pre-loaded during the server startup process.
If the amount of time specified for this property is not sufficient for the amount of cluster data that needs to be gathered, NO_IMPLEMENT: No Cluster Data Available exceptions might still occur the first few times a client sends requests to a cluster. Specifying an appropriate length of time for this property or specifying a value of 0, which eliminates the timeouts completely prevents the NO_IMPLEMENT: No Cluster Data Available exceptions from occurring because the cluster data is gathered within the specified length of time.
Property | IBM_CLUSTER_CALLBACK_TIMEOUT |
Data type | integer |
Default | 180000, which is equivalent to 3 minutes |
Specifies the number of threads in the thread pool that are used to run the custom advisors.
There is one thread pool that is used to run all the custom advisors configured on a proxy server. If there are more custom advisors configured than there are threads in the pool, and the polling interval and other circumstances are such that more custom advisors should be running at the same time than there are threads in the pool, some custom advisors get queued up and run as soon as a thread becomes available.
Property | IBM_CLUSTER_CUSTOM_ADVISOR_THREAD_POOL_SIZE |
Data type | integer |
Range | 1 - 50 |
Default | 5 |
Specifies whether publishing of updates to the ClusterDescription is delayed. Enabling this custom property provides a performance improvement in large service integration bus (SIBus) topologies and configured destinations, which results in a reduction of messaging fngine start and stop times.
When this property is set to true, publishing of updates is delayed.
When this property is set to false, updates are published immediately.
Property | IBM_CLUSTER_ENABLE_CAR_DELAY_POSTING |
Data type | boolean |
Default | false |
Specifies whether the pre-load logic should run at server startup on the node agent. Without preload, a node agent only loads the data for a cluster after the node agent receives the first request for that cluster.
When this property is set to true, cluster data is loaded on the node agent at startup, and does not have to be created and propagated at runtime.
When this property is set to false, the cluster data is created and propagated the first time there is a request to a cluster, which sometimes causes NO_IMPLEMENT: No Cluster Data Available exceptions the first few times a client sends requests to a cluster.
Property | IBM_CLUSTER_ENABLE_PRELOAD |
Data type | boolean |
Default | false |
Specifies whether references to identities are deleted when there are no ClusterObservers registered for notifications on those identities. When an identity is deleted, all related posts on the BulletinBoard about that identity are cleared. Setting this property to true enables the references to identities that do not have any ClusterObservers registered on them to be deleted.
Typically there are many destinations defined in a Service Integration Bus (SIB) hierarchical scenario. If this property is set to false, and a product, such as the WebSphere Process Server is installed on top of WebSphere Application Server, the Workload Management (WLM) does not properly allow data stored on the Bulletin Board to be garbage collected. This situation can cause a slow memory leak if certain tasks, such as the installing and uninstalling applications, are repeated without restarting the process. If the process is restarted all posts associated with that server are automatically removed, thereby preventing the memory leaks.
Property | IBM_CLUSTER_PURGE_NOTIFICATIONS |
Data type | boolean |
Default | false |
Specify a value, in milliseconds, to indicate the amount of time the ripplestart function waits for processes to shut down before restarting them. If you attempt a ripplestart and the processes have not shutdown before the start operation begins, one or more of the processes will not restart.
Property | IBM_CLUSTER_RIPPLESTART_NOTIFICATION_TIMEOUT |
Data type | integer |
Default | 300000 milliseconds (5 minutes) |
Exception stack trace: javax.naming.NamingException: Error during resolve. Root exception is rg.omg.CORBA.NO_IMPLEMENT:Trace from server: server_name at host host_name
To enable the property at the server level, in the administrative console click Servers > Server Type > WebSphere application servers, and then under Server Infrastructure, click Administration > Custom Properties.
gotchaProperty | IBM_CLUSTER_WBI_SUPPORT |
Data type | boolean |
Default | false |
Specify a value of displayServerInFront to display the name of the cell, node, and server in front of the output for the ps -ef command. Use of this property is intended to help you identify the process ID of a server. The property has no impact on the server process.
Property | com.ibm.websphere.management.launcher.options |
Data type | string |
Default | None |
Specifies whether publishing of updates to the ActiveClusterSet is delayed. Enabling this custom property provides a performance improvement in large SIBus topologies and configured destinations, which results in a reduction of Messaging Engine Start and Stop times.
When this property is set to true, publishing of updates is delayed.
When this property is set to false, updates are published immediately.
Property | IBM_CLUSTER_ENABLE_ACS_DELAY_POSTING |
Data type | boolean |
Default | false |
Starting with Version 6.1.0.37, Workload management (WLM) uses a new procedure for compressing and decompressing data being sent between processes. This procedure reduces compression overhead in large WebSphere Process Server and service integration bus topologies, that have 2,000 or more destinations. This procedure also prevents compressed data from being lost if a series of bytes to be compressed could not actually be compressed because every byte is unique. This situation could potentially cause a loss of data because of how WLM previously handled the uncompressed data.
This optimized compression and decompression procedure is appropriate for most environments. However if this procedure causes problems in your environment, add the IBM_CLUSTER_USE_LEGACY_COMPRESSOR custom property to your cell settings, and set it to true. When you set this property to true, WLM handles data compression and decompression the same way that it did prior to Version 6.1.0.37 or later being installed.
If you add this custom property to your cell settings, you must synchronize the nodes, and restart all of the processes in the cell before this change goes into effect.
Property | IBM_CLUSTER_USE_LEGACY_COMPRESSOR |
Data type | boolean |
Default | false |
By default, the autoRestart attribute for the server monitoring policy is set to false because the Automatic Restart Management (ARM) is typically used to manage automatic restarts of the application servers. The autoRestart attribute for the server monitoring policy is defined in the server.xml file.
If set the autoRestart attribute to true because you do not want to use ARM to manage your server automatic restarts, you must also add the com.ibm.websphere.process.terminator.deletepid cell custom property to your cell configuration and set this custom property to true. If you only change the setting on the autoRestart attribute, the servers in the cell keep restarting in response to a stopImmediate command.
Property | com.ibm.websphere.process.terminator.deletepid |
Data type | boolean |
Default | false |