Viewing CPU, memory, disk and network charts for a VM

Overview

The Metrics tab provides real-time and historical time-series performance telemetry for your virtual machines. You can inspect CPU utilisation, system memory allocation, disk storage consumption, and cumulative network input/output rates directly from the dashboard.

Telemetric data is collected by the lightweight CloudPe monitoring agent installed within the guest operating system and streamed securely to regional metric collectors. The console queries regional collectors to render performance charts across selectable evaluation periods.

Before you start

• Install and enrol the CloudPe monitoring agent inside the guest VM. See Installing the monitoring agent.
• You need the vms:read permission to view VM performance metrics and agent health status.
• Fleet-wide aggregation across multiple instances is accessible from Monitoring in the sidebar. See Fleet health and metric dashboards.
• Virtual machines without an enrolled monitoring agent appear with an empty state on the Metrics tab offering an agent installation command.

Steps

View performance charts for a VM

1. In the sidebar, open Instances under COMPUTE.
2. Select a virtual machine to open its detail view.
3. Switch to the Metrics tab.
4. If the agent is not yet registered, select Generate Install Command under Agent Installation to copy the enrolment script.
5. Once the agent is actively streaming, review the four core performance visualisations:
   • CPU Usage: Average processor load across assigned virtual cores.
   • Memory Usage: Guest memory consumption and cache utilisation.
   • Disk Usage: Root and attached filesystem capacity utilisation.
   • Network I/O (cumulative): Cumulative inbound and outbound network byte counters.
6. Use the period selector buttons (1h, 6h, 24h, 7d, 30d) to adjust the time window.
7. On GPU instances, additional telemetry cards display GPU core utilization, video memory usage, and GPU processor temperature.
8. If an agent has stopped reporting, select Re-enrol agent to retrieve an updated enrolment key and re-link the instance.

API

Retrieve metric time series and agent diagnostics directly using the VM endpoints.

Method and path	Permission
GET /api/v1/vms/{vm_id}/metrics	vms:read
GET /api/v1/vms/{vm_id}/metrics/latest	vms:read
GET /api/v1/vms/{vm_id}/metrics/gpu	vms:read
GET /api/v1/vms/{vm_id}/agent/health-metrics	vms:read

Query metric time-series data for an instance across a specified period:

curl "https://app.cloudpe.com/api/v1/vms/<vm_id>/metrics?period=1h" \
  -H "Authorization: Bearer <API_KEY>"

Fetch the latest snapshot reading:

curl "https://app.cloudpe.com/api/v1/vms/<vm_id>/metrics/latest" \
  -H "Authorization: Bearer <API_KEY>"

Fetch GPU-specific telemetry:

curl "https://app.cloudpe.com/api/v1/vms/<vm_id>/metrics/gpu" \
  -H "Authorization: Bearer <API_KEY>"

The response includes a status field indicating query resolution: ok when readings are available, no_data when the agent has not reported during the window, or monitoring_unavailable when regional collection services are unreachable.

Limits

• Metric telemetry is retained in tiered rollups: short windows query raw samples, while multi-day periods are computed from downsampled averages.
• Downsampled long-term rollup tables omit disk I/O and network rate series; extended windows on these metrics may render with empty intervals.
• Cumulative network and disk I/O metrics represent lifetime monotonic counters rather than instantaneous rates.
• Threshold alerts and auto-scaling evaluators reference standard CPU and memory utilization series. See Alert rules and notification channels.

Troubleshooting

Message	What it means	What to do
Region metrics collector is not configured	The regional metric collection service has not been activated in the specified region.	Contact support to verify monitoring service availability for your cloud region.
Agent not found	No agent registration record exists for the specified virtual machine.	Navigate to the Metrics tab and select Generate Install Command to install the agent.
Instance not found	The target instance UUID does not exist or is outside your current project scope.	Verify that you have selected the correct project in the console switcher.

FAQ

Why does a running instance show no metric data? A newly launched VM requires the CloudPe monitoring agent to be installed and enrolled before performance metrics can be collected.

Why do long-range charts show gaps in disk and network metrics? Historical metric retention downsamples long-range queries into statistical averages. Disk and network lifetime counters are preserved primarily in raw short-term resolution.

Are performance metrics collected when an instance is shelved? No. Shelved instances release host compute capacity and the operating system is halted, so no agent metrics are generated.

Related

• Installing the monitoring agent
• Fleet health and metric dashboards
• Alert rules and notification channels