Jaa

View Performance Charts

  • Artikkeli

In HPC Cluster Manager, you can view one or more graphs of the performance metrics for a single node over a configurable span of time.

To view and configure performance charts

  1. In Node Management, in the Navigation Pane, click Nodes.

  2. In the List or Heat Map view, select a node.

  3. Right-click your selection, then click View Performance Charts. In the Performance Charts dialog box, the CPU Usage chart is displayed by default.

  4. Optionally, modify the displayed time span by specifying the number of minutes in the X axis (minutes) text box.

    Note

    You can configure the displayed time span up to 50,000 minutes. However, node performance counter data is only stored in the HCPManagement database for three days by default. For information about modifying the data retention period, see Charts and Reports: HPC Cluster Manager.

  5. To display additional charts: In the Metrics drop-down list, select the performance metric you want to chart, then click Add.

  6. You can close individual charts by clicking the close icon (x).

The display options that you configure in the Performance Charts dialog box are saved, so that the metrics and time spans that you configure are displayed the next time that you view performance charts.

Additional considerations

  • You can view a set of aggregated metrics for all cluster nodes by Using the Monitoring Charts.

  • To add custom performance counters for your cluster, you can create an XML file that describes the metric to add, then import the new metric using the HPC PowerShell cmdlet Import-HpcMetric. You can also modify an existing metric (for example, you can change the sampling frequency) by overwriting the existing settings. For more information, see Customize Metrics Collection.

  • For information about using HPC Cluster Manager, see Overview of HPC Cluster Manager.

Additional references