Skip to main content
GPU Cloud provides monitoring capabilities to help you view the operating status of computing, cluster, storage, and network resources. You can go to OBSERVABILITY > Monitor and filter monitoring data by resource type, region, resource name, and time range to assess resource loads, identify performance anomalies, or help troubleshoot task issues. This topic helps you understand the resource types supported by Monitor and the basic usage workflow. For details about metrics, see the monitoring documentation for the corresponding resource type.

​
Supported Resource Types

Monitor supports viewing data by resource type:

​
Usage workflow

When viewing monitoring data, you typically follow this path:
  1. Go to OBSERVABILITY > Monitor.
  2. Select a resource type.
  3. Select a region and specific resources. When viewing the Cluster Overview, you can also filter by namespace or Pod.
  4. Select a time range.
  5. View the metrics charts.
  6. To retrieve the latest data, click Refresh.

​
Usage Recommendations

  • To determine whether an individual CES instance is busy, view the corresponding GPU Instance or CPU Instance monitoring data.
  • To determine the overall load of a CKS cluster, view the Cluster Overview monitoring data.
  • To view dedicated metrics for CES, a GPFS file system, or a VPC, select the corresponding resource type. When viewing GPFS file system monitoring data, select Xstor under Resource Type.
  • If a chart displays No data, first verify that the resource and time range are correct.
  • If metrics are abnormal, continue troubleshooting by reviewing logs and alert history.