The Streams Job Graph in Cloud Pak for Data allows you to visually monitor running Streams applications. You can observe metrics, view data flow rates, download operator logs, view live data, and more.

To open the job graph:

• From the IBM Cloud Pak for Data web user interface navigation menu, click My instances.
• Click the Jobs tab.
• From the options menu for the job that you want to manage, choose View graph.

Job Graph Highlights

– Monitor tuple flow rate to see how quickly (or slowly) data is being processed:
Select the operator whose metrics you want to observe, right click, Watch > Watch (upstream/downstream)

– Easily observe congestion and other flow rate issues: Streams in the graph are colour coded, as shown below:

– View live data streams:

To view a sample of the data flowing through a stream, select the stream, right-click and click View Data > Create new view.
A pane showing the data on the Stream will be visible below the application graph.
You can also use this pane to see a list of any views created at compile time.

– Pin system and custom metrics:
If your application computes custom metrics, those are also easily accessible. For example, the application below keeps track of the current number of sensors that are reporting data and the average temperature reported by all sensors.

– Pinpoint where errors have occurred:

In addition to the above tasks, you can also

• Enable tracing: Select the operator, right-click, and choose Set trace. Select a trace level and click Set trace.
• Download job and operator logs:
  • Download logs for the whole job: My instances > Jobs > find the job, and choose " Download logs" from the menu.
  • Download logs for a single operator: Click the log button on an operator to download its logs.
    
• Download a sample of streaming data as a CSV file: From the view pane, click Export.
• Restart processing elements and view nested graphs.

Try it

To try out the new graph in your Streams instance, go to My instances > Jobs, select the job you want to view, and click View graph.