Graph Events
Each event carries properties of graph, which is source of event. If there is a event listener specified, task may use these properties. i.e. next graphs in the chain may use "event_file_name" placeholder which activated first graph in the chain. Graph properties, which are set specifically for each graph run (i.e. RUN_ID), are overridden by last graph.
For now, there are these types of graph events:
graph started
Event of this type is created, when graph is successfully executed. It means, that threads of graph nodes and watchdog are running.
graph finished OK
Event of this type is created, when all phases and nodes of graph
are finished with status FINISHED_OK.
graph error
Event of this type is created, when graph cannot be executed from any reason, or when any node of graph fails.
graph aborted
Since 1.2.1
Event of this type is created, when graph is explicitly aborted.
graph timeout
Event of this type is created, when graph runs longer then specified interval. Thus you have to specify "Graph timeout interval" attribute for each listener of graph timeout event. You can specify this interval in seconds or in minutes or in hours.
graph status unknown
Since 1.3.
Event of this type is created, when server starts and there is graph with undefined status in the executions history. Undefined status means, that server has been killed during graph run. Server automatically changes state of graph to "Not Available" and sends 'graph status unknown' event. Please note, that this works just for executions, which have persistent record in executions history. It is possible to execute transformation without persistent record in executions history, typically for better performance of fast running transformations (i.e. using Launch Services).