Updating resources for an application server
Properly updating resources ensures that all transactional work completes while the original versions of the resources are still available. If resources are not properly updated, data builds up in the transaction partner log. Eventually high CPU usage is observed in the controller.
Before you begin
Before you start to update a resource, verify that all of the transactions being handled by that resource have completed..
About this task
In severe cases, when high CPU usage occurs in the controller because resources are improperly updated, the partner logs becomes full, and the application server becomes unusable. When the partner logs become full, the following error message appears in the servant log:
BBOO0220E: WTRN0000E: An internal error occurred in method logData in class com.ibm.ws.Transaction.JTA.PartnerLog
If before updating a resource, you ensure that there is no work pending that involves this resource, data will not build up in the transaction partner log. However, data accumulates in the transaction partner logs if there is a change in resource, there is a change in the configuration for a resource, or a resource is deleted, before all of the transactional work that the resource is handling completes. This situation occurs because after the resource update occurs, the old version of the resource is no longer available for recovery when the server is restarted.
If the recovery process does not completes when you restart the server, periodically, the product attempts to recover those transactions. To determine if there are transactions that have pending resolutions, look for the following message in the controller log:
BBOT0009I: TRANSACTION SERVICE RESTART UR STATUS COUNTS FOR SERVER {0}: IN-BACKOUT={1}, IN-DOUBT={2}, IN-COMMIT={3}
Procedure
Results
The application server is now using the changed resources with no impact to the transaction partner logs.