Enable kubectl completion (needs the bash-completion package):
source <(kubectl completion bash)
Dry-run, outputs Service (--expose) and a Deployment in yaml:
kubectl run --image=apache \
--port=80 \
--replicas=3 \
--restart='Always' \
--expose \
--requests='cpu=100m,memory=256Mi' \
--limits='cpu=200m,memory=512Mi' \
--labels=app=apache,version=1 \
--dry-run=true \
-o yaml
In a running container run date:
kubectl exec POD -- bash -c "date"
kubectl exec POD -- date
kubectl exec POD date
Remove label this from a pod:
kubectl label pod POD this-
Add label that=thing to a pod:
kubectl label pod POD that=thing
Select pods based on selector across all namespaces:
kubectl get pods --all-namespaces --selector this=label
kubectl get pods --all-namespaces -l this=label
Create a single Pod without a Deployment (--restart=Never) and a ReplicaSet:
kubectl run nginx --image=nginx --restart=Never
Create a single Pod with a Deployment (and a ReplicaSet):
kubectl run nginx --image=nginx --replicas=1
How the --restart flag behaves with kubectl run:
--restart=Never Creates a single Pod without a Deployment or a ReplicaSet. You can achive this by creating a single Pod manifest and apply it.
--restart=OnFailure Creates a Pod and a Job.
--restart=Always Default, creates a Deployment and a ReplicaSet object.
Copy file to/from a Pod:
kubectl cp POD:/path/to/file.txt ./file.txt
kubectl $HOME/file.txt POD:/path/to/file.txt
Patch a Deployment with a new image:
kubectl patch deployment nginx -p '{"spec":{"template":{"spec":{"containers":[{ "name":"nginx", "image":"nginx:1.13.1"}]}}}}'
Sort Kubernetes events on creation timestamp:
kubectl get events --sort-by=.metadata.creationTimestamp
Output the number of Pods foo and skip Pods bar that are running per node:
kubectl get pods -o wide | grep foo | grep Running | grep -v bar | awk '{print $7}' | sort | uniq -c | sort -nr | wc -l