Job process workflow

Understanding how Spotfire Statistics Services processes jobs can help you troubleshoot problems or just understand how the results are derived. The basic steps of the typical job process and the expected results you can expect are described in this topic.

  1. A client uploads the data necessary to complete a job to the server using a WebDAV service and submits a job request to a Manager node using a web request.
  2. Execution is asynchronous, and the Manager node stores the job in the job database, places it in a queue for execution, and marks the job as waiting.
  3. The job ID and the status are returned to the client.

    Using the job ID, a client can manage the job and view the status.

  4. When the job is submitted to a Worker node, the status is changed to running.
  5. When the job completes, the client is notified and, if it completed successfully, the client can retrieve the results.

If an error occurred, the status is set to either failed or done with error.

  • Done with error means that an error occurred while parsing and executing the job inside the engine.
  • Failed means that an unexpected internal error occurred while processing the job.)
Note: The client APIs provide for synchronous execution. The server achieves this by blocking the call until the job completion notification is received from the server. Synchronous jobs have a higher priority over asynchronous jobs.

When you deploy Spotfire Statistics Services as part of your Spotfire deployment, Spotfire Analyst or Spotfire Web Player is the client sending the job request to the cluster. The Manager node passes it on to an available Worker node, and then returns the result when the job is complete.