Using kubectl to Operate a Cluster

kubectl

kubectl is a command line tool for Kubernetes clusters. You can install kubectl on any node and run kubectl commands to operate your Kubernetes clusters.

For details about how to install kubectl, see Connecting to a Cluster Using kubectl. After connection, run the kubectl cluster-info command to view the cluster information. The following shows an example:

# kubectl cluster-info
Kubernetes master is running at https://*.*.*.*:5443
CoreDNS is running at https://*.*.*.*:5443/api/v1/namespaces/kube-system/services/coredns:dns/proxy

To further debug and diagnose cluster problems, use 'kubectl cluster-info dump'.

Run the kubectl get nodes command to view information about nodes in the cluster.

# kubectl get nodes
NAME            STATUS    ROLES     AGE       VERSION
192.168.0.153   Ready     <none>    7m        v1.15.6-r1-20.3.0.2.B001-15.30.2
192.168.0.207   Ready     <none>    7m        v1.15.6-r1-20.3.0.2.B001-15.30.2
192.168.0.221   Ready     <none>    7m        v1.15.6-r1-20.3.0.2.B001-15.30.2

Getting Started

get

The get command displays one or many resources of a cluster.

This command prints a table of the most important information about all resources, including cluster nodes, running pods, Deployments, and Services.

Examples:

To list all pods with detailed information:

kubectl get pod -o wide

To list pods in all namespaces:

kubectl get pod --all-namespaces

To list labels of pods in all namespaces:

kubectl get pod --show-labels

To list all namespaces of the node:

kubectl get namespace

To list a pod with a specified name in YAML output format:

kubectl get pod <podname> -o yaml

To list a pod with a specified name in JSON output format:

kubectl get pod <podname> -o json

kubectl get pod rc-nginx-2-btv4j -o=custom-columns=LABELS:.metadata.labels.app

create

The create command creates a cluster resource from a file or input.

If there is already a resource descriptor (a YAML or JSON file), you can create the resource from the file by running the following command:

kubectl create -f <filename>

expose

The expose command exposes a resource as a new Kubernetes service. Possible resources include a pod, Service, and Deployment.

kubectl expose deployment <deployname> --port=81 --type=NodePort --target-port=80 --name=<service-name>

run

To run a particular image in the cluster:

Examples:

kubectl run <deployname> --image=nginx:latest

To run a particular image using a specified command:

kubectl run <deployname> --image=busybox --command -- ping example.com

set

The set command configures object resources.

Example:

To update the container image of a Deployment to version 1.0 in rolling mode:

kubectl set image deployment/<deployname> <containername>=<containername>:1.0

edit

The edit command edits a resource from the default editor.

Examples:

To update a pod:

kubectl edit pod po-nginx-btv4j

The example command yields the same effect as the following command:

kubectl get pod po-nginx-btv4j -o yaml >> /tmp/nginx-tmp.yaml
vim /tmp/nginx-tmp.yaml
/*do some changes here */
kubectl replace -f /tmp/nginx-tmp.yaml

explain

The explain command views documents or reference documents.

Example:

To get documentation of pods:

kubectl explain pod

delete

The delete command deletes resources by resource name or label.

Example:

To delete a pod with minimal delay:

kubectl delete pod <podname> --now 

kubectl delete -f nginx.yaml
kubectl delete deployment <deployname>

Deployment Commands

rollout

The rollout command manages the rollout of a resource.

Examples:

To check the rollout status of a particular deployment:

kubectl rollout status deployment/<deployname>

To view the rollout history of a particular deployment:

kubectl rollout history deployment/<deployname>

To roll back to the previous deployment: (by default, a resource is rolled back to the previous version)

kubectl rollout undo deployment/test-nginx

scale

The scale command sets a new size for a resource by adjusting the number of resource replicas.

kubectl scale deployment <deployname> --replicas=<newnumber>

autoscale

The autoscale command automatically chooses and sets the number of pods. This command specifies the range for the number of pod replicas maintained by a replication controller. If there are too many pods, the replication controller terminates the extra pods. If there is too few, the replication controller starts more pods.

kubectl autoscale deployment <deployname> --min=<minnumber> --max=<maxnumber>

Cluster Management Commands

cordon, drain, uncordon*

If a node to be upgraded is running many pods or is already down, perform the following steps to prepare the node for maintenance:

1. Run the cordon command to mark a node as unschedulable. This means that new pods will not be scheduled onto the node.

   kubectl cordon <nodename>

   Note: In CCE, nodename indicates the private network IP address of a node.

2. Run the drain command to smoothly migrate the running pods from the node to another node.

   kubectl drain <nodename> --ignore-daemonsets --ignore-emptydir

   ignore-emptydir ignores the pods that use emptyDirs.

3. Perform maintenance operations on the node, such as resetting the node.
4. After node maintenance is completed, run the uncordon command to mark the node as schedulable.

   kubectl uncordon <nodename>

cluster-info

To display the add-ons running in the cluster:

kubectl cluster-info

To dump current cluster information to stdout:

kubectl cluster-info dump

top*

The top command displays resource (CPU/memory/storage) usage. This command requires Heapster to be correctly configured and working on the server.

taint*

The taint command updates the taints on one or more nodes.

certificate*

The certificate command modifies the certificate resources.

Fault Diagnosis and Debugging Commands

describe

The describe command is similar to the get command. The difference is that the describe command shows details of a specific resource or group of resources, whereas the get command lists one or more resources in a cluster. The describe command does not support the -o flag. For resources of the same type, resource details are printed out in the same format.

logs

The logs command prints logs for a container in a pod or specified resource to stdout. To display logs in the tail -f mode, run this command with the -f flag.

kubectl logs -f <podname>

exec

The kubectl exec command is similar to the Docker exec command and executes a command in a container. If there are multiple containers in a pod, use the -c flag to choose a container.

kubectl exec -it <podname> -- bash
kubectl exec -it <podname> -c <containername> -- bash

port-forward*

The port-forward command forwards one or more local ports to a pod.

Example:

To listen to local port 5000 and forward requests to port 6000 in a pod created in <my-deployment>:

kubectl port-forward deploy/my-deployment 5000:6000

cp

To copy files or directories and paste them to a container:

kubectl cp /tmp/foo <podname>:/tmp/bar -c <containername>

The local files in /tmp/foo are copied and pasted to the /tmp/bar directory of a specific container in a remote pod.

auth*

The auth command inspects authorization.

attach*

The attach command is similar to the logs -f command and attaches to a process that is already running inside an existing container. To exit, run the ctrl-c command. If a pod contains multiple containers, to view the output of a specific container, use the -c flag and containername following podname to specify a container.

kubectl attach <podname> -c <containername>

Advanced Commands

replace

The replace command updates or replaces an existing resource by attributes including the number of replicas, labels, image versions, and ports. You can directly modify the original YAML file and then run the replace command.

kubectl replace -f <filename>

apply*

The apply command provides a more strict control on resource updating than patch and edit commands. The apply command applies a configuration to a resource and maintains a set of configuration files in source control. Whenever there is an update, the configuration file is pushed to the server, and then the kubectl apply command applies the latest configuration to the resource. The Kubernetes compares the new configuration file with the original one and updates only the changed configuration instead of the whole file. The configuration that is not contained in the -f flag will remain unchanged. Unlike the replace command which deletes the resource and creates a new one, the apply command directly updates the original resource. Similar to the git operation, the apply command adds an annotation to the resource to mark the current apply.

kubectl apply -f <filename>

patch

If you want to modify attributes of a running container without first deleting the container or using the replace command, the patch command is to the rescue. The patch command updates field(s) of a resource using strategic merge patch, a JSON merge patch, or a JSON patch. For example, to change a pod label from app=nginx1 to app=nginx2 while the pod is running, use the following command:

kubectl patch pod <podname> -p '{"metadata":{"labels":{"app":"nginx2"}}}'

convert*

The convert command converts configuration files between different API versions.

Configuration Commands

label

The label command update labels on a resource.

kubectl label pods my-pod new-label=newlabel

annotate

The annotate command update annotations on a resource.

kubectl annotate pods my-pod icon-url=http://*****

completion

The completion command provides autocompletion for shell.

Other Commands

api-versions

The api-versions command prints the supported API versions.

kubectl api-versions

api-resources

The api-resources command prints the supported API resources.

kubectl api-resources

config*

The config command modifies kubeconfig files. An example use case of this command is to configure authentication information in API calls.

help

The help command gets all command references.

version

The version command prints the client and server version information for the current context.

kubectl version