Monitoring OSA Servers

OSA servers push events related to their activity and status to IBM® Sterling Control Center Monitor and there are two services in IBM Sterling Control Center Monitor utilized to handle processing of OSA server data, the EventProcessorService and the UnprocessedEventService. Performance of these services is controlled by a combination of:

Configuration file data in EventProcessorService.xml and
Engine property values

OSA event processing may be performed synchronously or asynchronously depending on the version of IBM Sterling Control Center Monitor you’re running, and asynchronous OSA event processing is controlled by various engine property values.

Originally, only synchronous processing was supported, and it is currently the default – it’s not clear that asynchronous processing performs better than synchronous processing, and the configuration parameters stored in the EventProcessorService.xml data, which is kept in the CC_FILES database table, controlled all aspects of OSA server processing. The parameters in EventProcessorService.xml that affect processing are:

hbCheckFrequency
historyCheckFrequency
eventsToProcessAtOnce

The following table includes the default values for these configuration parameters and explains how they are used:

Parameter	Details	Default
hbCheckFrequency	A value in milliseconds that controls the frequency at which the EventProcessorService checks to see which OSA servers are late in sending in Heartbeat events, which is the indicator that they are down. The larger this value is, the longer the delay could be for IBM Sterling Control Center Monitor to ascertain that a down OSA server is down.	10000
historyCheckFrequency	A value in milliseconds that controls the frequency at which the EventProcessorService checks for new OSA events to process from servers monitored by this EP. Note when the number of OSA events processed matches eventsToProcessAtOnce no wait is performed between checks for new OSA events to process. This is referred to as catchup mode. This property is only used when OSA event processing is done synchronously.	10000
eventsToProcessAtOnce	Specifies the limit on OSA records to be retrieved by the UnprocessedEventService from the CC_UNPROCESSED_EVENT table. This property is only used when OSA event processing is done synchronously.	1000

Parameter

Details

Default

hbCheckFrequency

A value in milliseconds that controls the frequency at which the EventProcessorService checks to see which OSA servers are late in sending in Heartbeat events, which is the indicator that they are down.

The larger this value is, the longer the delay could be for IBM Sterling Control Center Monitor to ascertain that a down OSA server is down.

10000

historyCheckFrequency

A value in milliseconds that controls the frequency at which the EventProcessorService checks for new OSA events to process from servers monitored by this EP.

Note when the number of OSA events processed matches eventsToProcessAtOnce no wait is performed between checks for new OSA events to process. This is referred to as catchup mode.

This property is only used when OSA event processing is done synchronously.

10000

eventsToProcessAtOnce

Specifies the limit on OSA records to be retrieved by the UnprocessedEventService from the CC_UNPROCESSED_EVENT table.

This property is only used when OSA event processing is done synchronously.

1000

Since OSA servers were introduced to ICC, additional engine properties may now be used to also affect, and in some cases override, the performance of OSA server monitoring. These engine properties are:

MULTI_THREAD_OSA_EVENT_PROCESSING
OSA_EVENTS_TO_GET_AT_ONCE_COUNT
OSA_EVENTS_TO_PROCESS_QUEUE_SIZE
TIME_BETWEEN_PROCESSING_QUEUED_OSA_EVENTS
TIME_BETWEEN_GETTING_UNPROCESSED_OSA_EVENTS

The following table includes the default values for these engine properties and explains how they’re used:

Parameter	Details	Default
MULTI_THREAD_OSA_EVENT_PROCESSING	Controls whether OSA event processing is done in a synchronous fashion, where by OSA records received are retrieved from the database, then processed, and then deleted from the database, or in an asynchronous fashion, where one thread –handleUnprocessedEvents, is used to process OSA records previously retrieved from the database by another thread –queueUnprocessedEvents, which retrieves OSA records from the database and queues them for processing by handleUnprocessedEvents.	FALSE
OSA_EVENTS_TO_GET_AT_ONCE_COUNT	Sets the limit used in queries initiated by the queueUnprocessedEvents thread to retrieve unprocessed OSA events. This property is only used when OSA event processing is done asynchronously.	500
OSA_EVENTS_TO_PROCESS_QUEUE_SIZE	Sets the limit on the queue used to contain unprocessed OSA events retrieved by queueUnprocessedEvents thread, which are subsequently processed by the handleUnprocessedEvents thread. Always make the value for this property be some multiple, greater than one, of the OSA_EVENTS_TO_GET_AT_ONCE_COUNT value. This property is only used when OSA event processing is done asynchronously.	5000
TIME_BETWEEN_PROCESSING_QUEUED_OSA_EVENTS	A value in milliseconds that dictates how long the handleUnprocessedEvents thread waits after processing the previous batch of OSA events. Note when the batch size of OSA events previously processed matches the OSA_EVENTS_TO_PROCESS_QUEUE_SIZE size, no wait is performed. This is referred to as catchup mode. This property is only used when OSA event processing is done asynchronously.	5000
TIME_BETWEEN_GETTING_UNPROCESSED_OSA_EVENTS	A value in milliseconds that dictates how long the queueUnprocessedEvents thread waits after retrieving and queuing unprocessed OSA events. Note when the number of OSA events previously retrieved matches the OSA_EVENTS_TO_GET_AT_ONCE_COUNT value no wait is performed. This is referred to as catchup mode. This property is only used when OSA event processing is done asynchronously.	5000