Maintaining Alert Engine Processes 
The component-based message alerting mechanism uses dedicated local stores for the following purposes:
“Alert event” stores for keeping alert events incoming from message processing components before the Alert Engine evaluates them against the alert rules.
“Consumer” stores for keeping the alerts created for each consumer.
For both types of stores, the system takes special care to prevent the stores in use from growing uncontrollably. The system performs checks and measures on the discrete number of alert events or alerts sent to a particular store. For the purpose of describing the mechanism of the checks, within the current document the term “chunk” refers to a discrete, fixed-length sequence of alert events or alerts.
Alert Event Store on AS Java Systems
The message processing component running on AS Java has two parameters for managing an alert event store: chunk length and high volume threshold interval (in milliseconds). When you need to view or change their default values, you can do this using the corresponding parameters of the Adapter Framework Core (AF Core) service listed below.
The message processing component measures the time interval in which it sends two consecutive chunks to an alert event store. When this interval is shorter than the threshold interval, the message processing component switches the process of sending alert events to emergency mode. In this mode the message processing component behaves as follows:
It does not put new alert events into the store.
When there is new chunk available, the system writes only a general emergency message into the log. This message contains information that on the particular node more than “n” alert events in time “t” happened.
If the time interval between two consecutive chunks is greater than the threshold interval (or the threshold is reached before the end of the chunk), the message processing component switches to normal mode and starts putting all alert events into the store.
The local maintenance of the chunk statistics is sufficient for a cluster environment, since each store instance resides in its own cluster node, together with its statistical information.
Consumer Store on AS Java Systems
The Alert Engine running on AS Java has three parameters for managing consumer stores: chunk length, high volume threshold interval (in milliseconds), and maximum acceptable alert age (in milliseconds). When you need to view or change their default values, you can do this using the corresponding parameters of the Alert service listed below.
If there are many alert rules, a high volume situation can occur in the consumer store. For example, this can happen when the intervals at which the consumer reads alerts from the consumer store are long, and as a result the number of alerts stored in the consumer store grows to high levels.
The Alert Engine uses the following approach for maintaining a consumer store: at the end of each chunk, it determines the age of first entry and compares it to the maximum acceptable alert age. If this entry is older, the alert engine proceeds in the following way:
It deletes the whole consumer store.
It adds a specific “alert deletion” alert to the store. This entry contains information about the number of alert rules deleted as a result of the maintenance of the consumer store.
If another removal happens before the consumer reads the alerts from the store, the system deletes all alerts again. Note that the system keeps the “alert deletion” alert in the store, and adds the sum of alerts deleted previously and just now to this alert.
Consumer Store on AS ABAP Systems
The Alert Engine running on AS ABAP has one parameter for managing the consumer store: high volume threshold. This is an internal parameter and its value (1,000,000) cannot be modified.
The Alert Engine checks if the number of all alerts in all consumer stores is bigger than the high volume threshold, and when this happens it does the following:
It deletes all alerts in all consumer stores.
It adds a specific “alert deletion” alert to the store. This entry contains information about the number of alert rules deleted as a result of the maintenance of the consumer store.
If the store grows again beyond the threshold, the system deletes again all alerts. Note that the system keeps the “alert deletion” alert in the store, and adds the sum of alerts deleted previously and just now to this alert.
Services and Parameters for Maintaining Stores on AS Java
Store |
Store Content |
Service |
Parameters |
|---|---|---|---|
Alert Event |
Alert events |
XPI Service: AF Core |
alerting.events.chunkLength alerting.events.highVolumeThresholdInterval |
Consumer |
Alerts |
XPI Service: Alert |
housekeeping.alertsChunkLength housekeeping.alertsAcceptableAgeLimit[millisec] housekeeping.chunksHighVolumeThresholdInterval[millisec] |
For more information about other properties of the XPI Service: AF Core service, see Properties for XPI Service: AF Core.
Prerequisites
The user account you use is allowed to view and modify system details in SAP NetWeaver Administrator. For more information, see SAP NetWeaver Administrator.
Procedure
Access the Java services management tool. For more information, see Starting and Stopping Java Services.
Choose one of the services for the alert event or consumer store listed above.
In case the service properties are not visible, choose
More Actions 
View Service Properties 
.In the Extended Details table, locate the service parameters listed above.
You can use the Show Details button to obtain additional information about each parameter.
If necessary, modify and save the parameters.
Result
The changes you make to the service parameters are immediately applied to the system, there is no need to restart the system or a particular service.