I wrote this article was originally written for SigNoz.
Kubectl Top command can be used to retrieve snapshots of resource utilization of pods or nodes in your Kubernetes cluster. Resource utilization is an important thing to monitor for Kubernetes cluster owners. In order to monitor resource utilization, you can keep track of things like CPU, memory, and storage.
In this article, we will see how to use
Kubectl Top command to get and read metrics about pods and nodes. We will also breakdown the output to understand what it means.
But before we get down to learn about Kubectl Top command, let’s have a brief overview of a few concepts in Kubernetes.
kubectl is the Kubernetes command-line tool, and it allows you to run commands against your Kubernetes cluster.
kubectl lets you interact with your Kubernetes cluster for day-to-day management. For example,
kubectl get nodes lets you retrieve details about nodes running in your cluster or namespace.
Under the hood,
kubectl interacts with the API server. The API server is responsible for all communication between Kubernetes components, including both the internal components of your Kubernetes cluster and external components.
kubectl sends out
POST commands to the API server endpoint in order to execute its commands.
Kubernetes is meant to orchestrate the management of containers. Pods are the first level of abstraction that it provides over containers.
Pods are groups of one or more containers with shared resources like storage and networks. They are used as units of replication in cases where applications need to be scaled up or down.
Nodes are where Kubernetes pods run. Nodes can be virtual machines, bare metal servers in a data center, or instances in a private or public cloud. Kubernetes uses nodes to handle on-demand scaling of resources.
A single node can have multiple pods separated by namespaces.
As mentioned before,
kubectl - pronounced (Kube: c-t-l) - is a CLI for running commands that can help you interact with a Kubernetes interface or resources in a k8 cluster. These resources include pods, deploy, replica set, etc.
A kubectl top is a command used to list all the running nodes and pods along with their resource utilization. It provides you a snapshot of resource utilization metrics like CPU, memory, and storage on each running node.
Each node in Kubernetes comes with cAdvisor, which is an open-source agent that monitors resource usage about containers.
kubectl command gets resource utilization metrics from cAdvisor via the metrics-server.
To obtain these metrics, you need to run the
kubectl top command which shows the CPU, memory, and network utilization for the containers, pods, or nodes. For the
kubectl top command to work, you need to have metrics API installed. You can find instructions to install metrics API here.
Now that you have a brief understanding of the concepts, let’s see how the
kubectl top command operates.
kubectl top node command lists metrics of the current node which would look like this:
$ kubectl top node NAME CPU(cores) CPU% MEMORY(bytes) MEMORY% kind-control-plane 338m 4% 1662Mi 10%
The output from
kubectl top node gives you information about CPU(cores), CPU%, memory, and memory%. Let’s see what these terms mean:
338m means 338 millicpu.
1000m is equal to 1 CPU, hence 338m means 33.8% of 1 CPU.
It is displayed only for nodes, and it stands for the total CPU usage percentage of that node.
Memory being used by that node
It is also displayed only for nodes, and it stands for total memory usage percentage of that node.
kubectl top pod command displays the metrics about pods from the
default namespace which looks like this:
NAME CPU(Cores) MEMORY(Bytes) nginx 3m 1Mi nginx-653c7b42sd-4g5ce 3m 1Mi nginx-653c7b42sd-7c9ae 3m 1Mi
Mi under memory stands for mebibytes.
kubectl top pod command with
--all-namespaces lists down pods from all namespaces in your k8s cluster. For example, below is a snapshot from SigNoz k8s cluster.
$ kubectl top pod --all-namespaces NAMESPACE NAME CPU(cores) MEMORY(bytes) kube-system coredns-558bd4d5db-k7mfl 8m 11Mi kube-system coredns-558bd4d5db-qwrrk 8m 12Mi kube-system etcd-kind-control-plane 32m 45Mi kube-system kindnet-trm65 1m 7Mi kube-system kube-apiserver-kind-control-plane 105m 408Mi kube-system kube-controller-manager-kind-control-plane 22m 62Mi kube-system kube-proxy-8n86t 2m 17Mi kube-system kube-scheduler-kind-control-plane 5m 28Mi kube-system metrics-server-57bfd75b9-bhrwl 5m 14Mi local-path-storage local-path-provisioner-547f784dff-tjnqb 3m 9Mi platform chi-signoz-cluster-0-0-0 43m 149Mi platform clickhouse-operator-8cff468-hggdm 1m 24Mi platform my-release-signoz-alertmanager-0 2m 14Mi platform my-release-signoz-frontend-f8587978f-7wj8f 1m 6Mi platform my-release-signoz-otel-collector-cbf578f44-69twr 4m 52Mi platform my-release-signoz-otel-collector-metrics-5dcb767c77-5bgpt 4m 38Mi platform my-release-signoz-query-service-0 3m 57Mi platform my-release-zookeeper-0 5m 90Mi
You can also use the
--namespace flag to get information about pods from a particular namespace. For example, in the below snapshot, we can see details about pods from the
$ kubectl top pod --namespace platform NAME CPU(cores) MEMORY(bytes) chi-signoz-cluster-0-0-0 44m 165Mi clickhouse-operator-8cff468-hggdm 1m 24Mi my-release-signoz-alertmanager-0 2m 14Mi my-release-signoz-frontend-f8587978f-7wj8f 1m 6Mi my-release-signoz-otel-collector-cbf578f44-69twr 5m 54Mi my-release-signoz-otel-collector-metrics-5dcb767c77-5bgpt 5m 38Mi my-release-signoz-query-service-0 3m 57Mi my-release-zookeeper-0 4m 91Mi
Resource utilization metrics are key to understanding the health of your Kubernetes cluster. From the article, you learned how to get resource utilization snapshots using the
kubectl top command.
kubectl top command gives you basic metrics about resource utilization, it is very convenient to inspect your nodes and pods at any time. For example, if you see that there is a sudden spike in your resource utilization, you can check which pod is consuming the most resources.
Kubernetes provides us with a smarter way to manage our resources for scaling cloud-native applications on demand. But you also need to monitor your Kubernetes resources effectively. If you want to dive deeper into Kubernetes monitoring, you can check out SigNoz.
SigNoz is a full-stack open-source APM tool that can help you monitor your Kubernetes cluster. It uses OpenTelemetry to collect metrics from your K8s cluster for monitoring. OpenTelemetry is becoming the world standard for instrumentation of cloud-native applications, and it is backed by CNCF foundation, the same foundation under which Kubernetes graduated.
If you wish to learn more about how to monitor your Kubernetes cluster with OpenTelemetry and SigNoz, follow this blog:
If you wish to learn more about SigNoz, follow this blog: