שתף באמצעות

Charts and Reports: HPC Cluster Manager

  • מאמר

In HPC Cluster Manager, in Charts and Reports, you can collect and display information about your HPC cluster. This information can help you understand usage, throughput, turnaround time, and availability of your cluster over time. You can view charts and modify the way they display, and you can print and save reports.

The Monitoring Charts display current and recent data regarding node health, job throughput, and cluster utilization. The node performance counter data is stored in the HCPManagement database. By default, this data is kept for three days. The data retention period for the node performance data is defined by the MinuteCounterRetention cluster property. You can configure this property by using the Set-HpcClusterProperty cmdlet. For example, to retain performance counter data for one week, run HPC PowerShell as an administrator and type the following cmdlet:

Set-HpcClusterProperty –MinuteCounterRetention 7

Warning

The default SQL database that is installed with HPC Pack 2008 R2 has a 4 GB limitation, and the default database that is installed with HPC Pack 2012 has a 10 GB limitation. Ensure that your database has the necessary capacity for longer data retention periods.

The Reports display historical data about the cluster such as cluster utilization, node availability, and job statistics. You can filter job reports by specific users, projects, services, or templates. The historical, aggregated data for the reports is stored in the HPCReporting database. Data for the current day is not available in these reports. For more information, see The HPCReporting database and data extensibility below.

In this section

The HPCReporting database and data extensibility

The information that is available in the charts and reports section is aggregated data that is stored in the HPCReporting database. The database can also store raw data about jobs that is available to support custom reporting when data extensibility is enabled (it is enabled by default). For example, you can create custom charge back reports that correspond to the charging methods that are used by your organization. For information about using the raw data for custom reporting, see Reporting Extensibility Step-by-Step Guide.

The following table describes the settings that control the data extensibility and retention periods for the raw data. These settings do not affect the aggregated data that is used for the built-in reports.

Setting Description
DataExtensibilityEnabled Specifies whether the cluster stores information for custom reporting about jobs, nodes, and the allocation of jobs to nodes.

True indicates that the cluster stores information for custom reporting about jobs, nodes, and the allocation of jobs to nodes. False indicates that the cluster does not store this information. The default value is True.
DataExtensibilityTtl Specifies the number of days that the HPCReporting database should store all of the information about jobs and nodes except for the allocation of jobs to nodes. This parameter has a default value of 365.
AllocationHistoryTtl Specifies the number of days that the HPCReporting database should store information about the allocation of jobs to nodes. This parameter has a default value of 5.
ReportingDBSize Contains the current size of the HPCReporting database. This value is a string that includes the units of measurement for the size. This parameter is read-only.

To view this property, the computer that is running HPC PowerShell must be able to access the HPCReporting database. For more information about enabling remote database access, see deploying a cluster with remote databases step-by-step guide.

You can view the values of the settings with the Get-HPCClusterProperty cmdlet and set the values with the Set-HpcClusterProperty cmdlet. For example, to disable data extensibility, run HPC PowerShell as an administrator and type the following cmdlet:

Set-HpcClusterProperty –DataExtensibilityEnabled $false

Additional considerations