Difference between revisions of "UM:Event Processing"

Line 8: Line 8:
[[File:Event_flow.png|thumb|500px|none|alt=Event flow|Event flow inside the monitoring system]]
[[File:Event_flow.png|thumb|500px|none|alt=Event flow|Event flow inside the monitoring system]]


As you can see on the flowchart, events can come from various sources: polling processes (status, configuration, discovery, and data collection), SNMP traps, and directly from external applications via client library. All incoming events go to single event queue for processing. A special process, called ''Event Processor'', takes events from the queue one by one and matches them against Event Processing Policy. As a result, alarms may be generated and actions may be executed. If event has ''write lo log'' attribute set, it is written to NetXMS event log at the end of processing.
As you can see on the flowchart, events can come from various sources: polling processes (status, configuration, discovery, and data collection), SNMP traps, and directly from external applications via client library. All incoming events go to single event queue for processing. A special process, called ''Event Processor'', takes events from the queue one by one and matches them against Event Processing Policy. As a result, alarms may be generated and actions may be executed. If the event has ''write to log'' attribute set, it is written to NetXMS event log at the end of processing.


Although it may seem that processing all events one by one may become a bottleneck in the system, this should not be the case. Event processor is highly optimized, and all potentially long operations (like action execution) are performed by separate processes.
Although it may seem that processing all events one by one may become a bottleneck in the system, this should not be the case. Event processor is highly optimized, and all potentially long operations (like action execution) are performed by separate processes.
AutoPatrol
191

edits