This topic describes how to use virtual nodes to obtain performance metrics supported by VCI (such as CPU, memory, and network throughput metrics).

Metrics

VCI performance metrics (such as CPU, memory, and network throughput metrics) are provided based on Volcano Engine VMP. For more information about metrics, see Basic VKE metrics and Common cAdvisor metrics.

Metric querying

You can query VCI metrics by using the following command:

The variable descriptions are as follows:

<kubeconfig_path>: The address of the KubeConfig file for the cluster where the VCI is located, for example, .kube/config.

<virtualnode_name>: The name of the target virtual node to which the VCI is scheduled, for example, vci-node1-cn-beijing-a.

<endpoint>: The endpoint corresponding to the metric, for example, /metrics/resource. For a list of endpoints, see the Endpoint column in the tables below.

A complete command example is as follows:

Supported metrics

CPU

Metric name	Metric type	Endpoint	Unit	Metric meaning
container_cpu_load_average_10s	gauge	/metrics/cadvisor	N/A	Container CPU load average over the last 10 seconds.
container_cpu_usage_seconds_total	Counter	/metrics/resource	Seconds	Total CPU time consumed.

Memory

Metric name	Metric type	Endpoint	Unit	Metric meaning
container_memory_working_set_bytes	Gauge	/metrics/resource	Bytes	Physical memory usage of the container.

Network

Metric name	Metric type	Endpoint	Unit	Metric meaning
container_network_receive_bytes_total	Counter	/metrics/cadvisor	Bytes	Total data received.
container_network_receive_packets_total	Counter	/metrics/cadvisor	Packet	Total number of packets received.
container_network_transmit_bytes_total	Counter	/metrics/cadvisor	Bytes	Total amount of data transmitted.
container_network_transmit_packets_total	Counter	/metrics/cadvisor	Packet	Total number of packets transmitted.
container_network_receive_errors_total	Counter	/metrics/cadvisor	Count	Total number of errors encountered while receiving.
container_network_transmit_errors_total	Counter	/metrics/cadvisor	Count	Total number of errors encountered while transmitting.
container_network_receive_packets_dropped_total	Counter	/metrics/cadvisor	Packet	Total number of packets dropped while receiving.
container_network_transmit_packets_dropped_total	Counter	/metrics/cadvisor	Packet	Total number of packets dropped while transmitting.

PVs

Metric name	Metric type	Endpoint	Unit	Metric meaning
kubelet_volume_stats_available_bytes	Gauge	/metrics	Bytes	Available capacity in the volume.
kubelet_volume_stats_capacity_bytes	Gauge	/metrics	Bytes	Total capacity of the volume.
kubelet_volume_stats_inodes	Gauge	/metrics	Count	Maximum number of inodes (index nodes) in the volume.
kubelet_volume_stats_inodes_free	Gauge	/metrics	Count	Number of free inodes in the volume.
kubelet_volume_stats_inodes_used	Gauge	/metrics	Count	Number of used inodes in the volume.
kubelet_volume_stats_used_bytes	Gauge	/metrics	Bytes	Used capacity in the volume.

Pods

Metric name	Metric type	Endpoint	Unit	Metric meaning
pod_cpu_usage_seconds_total	Counter	/metrics/resource	Seconds	Total CPU time consumed by the pod in seconds.
pod_memory_working_set_bytes	Gauge	/metrics/resource	Bytes	Physical memory usage of the pod in bytes.

Volcengine Kubernetes Engine