Run the Jobs

You have built the graph by executing the "tools" (emulated by cp) interactively under the control of the vw wrapper. By using that wrapper, you have established runtime tracing. With runtime tracing turned on, FlowTracer discovers the inputs and outputs at runtime as the program is run. It builds a dependency graph of the files related to the program. The flow holds a data set that defines the dependency graph, the jobs, and the current state of files. FlowTracer is now ready to execute the jobs in the flow, based on the dependency graph and the state of files in the system. It can schedule and deploy jobs that need to be run because they use an input file that has changed. This process is called "retracing".

  1. Go to the graph view.
  2. Point at the node bb.
  3. Right-click and hold to display the menu and select Run.
    A request to the FlowTracer server to bring the file bb up to date. This means re-executing the job vw cp aa bb.

    To accomplish this task, the server selects the fastest tasker in the network that can execute the job. When processing a run request, FlowTracer takes advantage of potential parallelism by sending multiple independent jobs to the available taskers. Depending on the number of taskers connected to your project, you may or may not see parallelism in action. This illustrates the process of bringing a particular file up-to-date. More commonly, you will want to bring the entire design up to date, or the entire set of nodes you are looking at.

  4. Select the set you are interested in; for example, choose the set "nodes" in the Systems folder.
  5. Select the Graph view in the Set Viewer.
  6. Click on the retrace icon to bring the current set up to date. If all elements in the set are already VALID (green) nothing needs to be done.
    Note: You can also run individual nodes using the pop-up menu, or the keyboard shortcut "r".