Overview

About Analytics Monitoring

CAF deploys configured analytics to multiple runtimes.

The runtimes include:
  • Predix (Analytics Framework, also known as PAF (Predix Analytics Framework))
  • Smart Signal
  • Spark (Predix Insights)
Deployments are of two types:
  • Single Analytic
  • Orchestration (group of analytics executed in a sequence one at time)
You can schedule a runtime deployment with one of the following options:
  • Run Once
  • Recurrent

You can access the Predix runtime deployment details in the Deployments page. Whereas, to view the SmartSignal runtime deployment details, you must access the SmartSignal Maintenance application. Similarly, you can view the Spark runtime deployment details in the Predix Insights runtime application.

About Predix (PAF) Runtime Deployment Monitoring

PAF deployments are submitted to a runtime with Run Once or Recurrent scheduling options.

Deployments are of two types:
  • Single Analytic
  • Orchestration (group of analytics executed in a sequence one at time)

If one of the steps in an Orchestration fails to run, the remaining steps will be terminated. For example, an orchestration with analytics, A1, A2, and A3, A1 runs first, followed by A2, and then A3. If any of the analytics fails to run, the Predix runtime terminates the orchestration deployment, and the remaining analytics in the orchestration will not run.

The execution status of a deployment appears in the SUCCESS RATIO and LAST RUN columns in the Deployments page. The following table describes the execution status that appear in the SUCCESS RATIO and LAST RUN columns based on the scheduling options.

Runtime StateSUCCESS RATIOLAST RUN
Run OnceNone
  • Success
  • Failed

Success implies that data of all assets has been processed successfully, and Failed implies that one or more asset jobs failed.

RecurrentStatus of the last three runs.

For example, if one of the last three runs failed, 2/3 Runs appear.

You can view the details of the three runs by selecting the link.

Success rate for the last run completed or the latest run in progress.

For example, if the column contains 100% Assets, it means that data of all assets has been processed successfully.

You can access the asset-level results by selecting the link.

The following table describes the execution status indicator.
Execution Status IndicatorDescription
Indicates that the data of all the assets has been processed successfully.
Indicates that one or more runs failed.
Indicates that the deployment execution is in progress.
The following are some of the reasons a deployment execution fails in PAF:
  • Runtime failed to schedule the job
  • Analytic instance is down
  • Failed to connect to the Data source
  • Failed to create an alert

You can drill down to the deployment execution status for a given run to download the results of assets corresponding to that run.

The results corresponding to a successful execution of an asset job contains timeseries data, asset attributes, and alert. The results corresponding to a failed execution of an asset job contains debug information.

An orchestration deployment execution result contains the following set of files per step:
  • Input
  • Output
  • Analytic template log
  • Error file when something fails.