Falcon Bridge

Falcon Atlas Bridge

Falcon Model

The default falcon modelling is available in org.apache.atlas.falcon.model.FalconDataModelGenerator. It defines the following types:

falcon_cluster(ClassType) - super types [Infrastructure] - attributes [timestamp, colo, owner, tags]
falcon_feed(ClassType) - super types [DataSet] - attributes [timestamp, stored-in, owner, groups, tags]
falcon_feed_creation(ClassType) - super types [Process] - attributes [timestamp, stored-in, owner]
falcon_feed_replication(ClassType) - super types [Process] - attributes [timestamp, owner]
falcon_process(ClassType) - super types [Process] - attributes [timestamp, runs-on, owner, tags, pipelines, workflow-properties]

One falcon_process entity is created for every cluster that the falcon process is defined for.

The entities are created and de-duped using unique qualifiedName attribute. They provide namespace and can be used for querying/lineage as well. The unique attributes are:

• falcon_process - <process name>@<cluster name>
• falcon_cluster - <cluster name>
• falcon_feed - <feed name>@<cluster name>
• falcon_feed_creation - <feed name>
• falcon_feed_replication - <feed name>

Falcon Hook

Falcon supports listeners on falcon entity submission. This is used to add entities in Atlas using the model defined in org.apache.atlas.falcon.model.FalconDataModelGenerator. The hook submits the request to a thread pool executor to avoid blocking the command execution. The thread submits the entities as message to the notification server and atlas server reads these messages and registers the entities.

• Add 'org.apache.atlas.falcon.service.AtlasService' to application.services in <falcon-conf>/startup.properties
• Link falcon hook jars in falcon classpath - 'ln -s <atlas-home>/hook/falcon/* <falcon-home>/server/webapp/falcon/WEB-INF/lib/'
• In <falcon_conf>/falcon-env.sh, set an environment variable as follows:

     export FALCON_SERVER_OPTS="$FALCON_SERVER_OPTS -Datlas.conf=<atlas-conf>"
     

The following properties in <atlas-conf>/atlas-application.properties control the thread pool and notification details:

• atlas.hook.falcon.synchronous - boolean, true to run the hook synchronously. default false
• atlas.hook.falcon.numRetries - number of retries for notification failure. default 3
• atlas.hook.falcon.minThreads - core number of threads. default 5
• atlas.hook.falcon.maxThreads - maximum number of threads. default 5
• atlas.hook.falcon.keepAliveTime - keep alive time in msecs. default 10
• atlas.hook.falcon.queueSize - queue size for the threadpool. default 10000

Refer Configuration for notification related configurations

Limitations

• In falcon cluster entity, cluster name used should be uniform across components like hive, falcon, sqoop etc. If used with ambari, ambari cluster name should be used for cluster entity