Often used commands
Common Naming Convention
Follow an APPNAME-RESOURCE-TYPE convention consistent with Helm when naming resources. For example:
$RESOURCE (e.g. app-configurator)
$APPNAME-$RESOURCE (e.g. myhiway-app-configurator)
$APPNAME-$RESOURCE-$TYPE (e.g. myhiway-app-configurator-secret)
APPNAME=name-of-application (e.g. myhiway)
RESOURCE=name-of-kube-resource (e.g. app-configurator)
TYPE=cm, configmap
TYPE=cronjob
TYPE=daemonset
TYPE=deployment
TYPE=endpoint
TYPE=event
TYPE=ing, ingress
TYPE=namespace
TYPE=node
TYPE=pv, persistentvolume
TYPE=pvc, persistentvolumeclaim
TYPE=pod
TYPE=pdb, poddisruptionbudget
TYPE=rs, replicaset
TYPE=sa, serviceaccount
TYPE=secretproviderclass
TYPE=secret
TYPE=svc, service
TYPE=statefulset
TYPE=sc, storageclass
Many TYPE names have shortcuts and can be used without the plural 's' where shown. And the RESOURCE variables in the commands below can represent different strings depending on what specificity the context needs.
CONTAINER=name-of-container-inside-of-pod
CONDITION=evicted, error, or crash
FILENAME=name-of-file
STRING=search-string
Often used flags
-c, --container
-n, --namespace,
-A, --all-namespaces
-l (show label, requires a resource)
--show-labels (show labels for all resource)
-o, --output (there are many output shapes => wide, json, jsonpath, yaml, go-template, etc.)
-w, --watch
-it, --stdin --tty
See the watch command on MacOS for a broader watch utility , e.g.
watch 'kubectl get all
get
Display one or many resources.
kubectl get $TYPE
kubectl get $TYPE -w
kubectl get $TYPE -o wide
kubectl get $TYPE -o json
Get all resources.
kubectl get all
kubectl get cm,daemonset,deployment,job,ing,pvc,pdb,pod,rs,secret,statefulset,svc
NAMESPACE=the-wheel
kubectl api-resources --verbs=list --namespaced -o name | xargs -n1 kubectl get --show-kind --ignore-not-found "$@" -n $NAMESPACE
kubectl api-resources --verbs=list --namespaced -o name | xargs -n1 kubectl get --show-kind --ignore-not-found -nl -n $NAMESPACE | grep $APPNAME
Get all resources - by filter.
kubectl get configmap,daemonset,deployment,job,ing,pvc,pdb,pod,replicaset,secret,statefulset,svc | grep solr
Get a resource.
RESOURCE=thewheel-app-configurator-67cd4dd5f-j9s24
TYPE=pod
kubectl get $TYPE $RESOURCE
+ kubectl get pod thewheel-app-configurator-67cd4dd5f-j9s24
RESOURCE=thewheel-app-configurator-secret
TYPE=secret
kubectl get $TYPE $RESOURCE
+ kubectl get secret thewheel-app-configurator-secret
$TYPE $RESOURCE can also be expressed as $TYPE/$RESOURCE, e.g. kubectl get pod foo => kubectl get pod/foo
Get a resource - filter by regex.
RESOURCE=thewheel-app-configurator
TYPE=secret
kubectl get $TYPE $(kubectl get $TYPE | grep $RESOURCE | awk '{print $1}')
+ kubectl get secret thewheel-app-configurator-secret
Get a resource - filter by label selector.
kubectl get $TYPE --show-labels
kubectl get $TYPE --show-labels | grep $RESOURCE
kubectl get pod -l app=$RESOURCE
kubectl get pod -l app=$RESOURCE --no-headers -o name
kubectl get pod -l app=$RESOURCE --no-headers -o name | awk -F '/' '{print $2}'
Get events - sort by time desc.
kubectl get events --sort-by=.metadata.creationTimestamp
Get secrets - base64 decoded.
kubectl get secret $RESOURCE -o json | jq '.data | map_values(@base64d)'
Get secrets - base64 decoded as file.
kubectl get secret $RESOURCE -o json | '.data | map_values(@base64d)' > ~/filename.json
Get secrets - base64 encoded.
kubectl get secret $APPNAME-$RESOURCE-$TYPE -o go-template='{{.data.bootstrapPassword|base64decode}}{{"\n"}}'
Get value from resource manifest by following json path.
kubectl get $TYPE $APPNAME-$RESOURCE-$TYPE -o jsonpath={.spec.template.spec.containers[0].name}
Get all containers inside a Pod.
kubectl get pod <podname> -o jsonpath='{.spec.containers[*].name}*' $RESOURCE $CONTAINER
delete
Delete resources by filenames, stdin, resources and names, or by resources and label selector.
kubectl delete $TYPE $RESOURCE
kubectl delete $TYPE $(kubectl get $TYPE | grep $STRING | awk '{print $1}')
Forcefully kill a pod named (be careful with this!).
kubectl delete pod $RESOURCE --force --grace-period 0
Delete all evicted, erroring, or crashed pods.
kubectl get pod -n default | grep prod | grep $CONDITION | awk '{print $1}' | xargs kubectl delete pod -n default
describe
Show details of a specific resource or group of resources.
kubectl describe $TYPE $APPNAME-$RESOURCE-$TYPE
kubectl describe $TYPE $(kubectl get $TYPE | grep $STRING | awk '{print $1}')
cp
Copy files and directories to and from containers.
PODNAME=(e.g. thewheel-app-configurator-67cd4dd5f-j9s24)
DESTINATION_PATH=(e.g. /var/www/wp-content-mount)
SOURCE_PATH=(e.g. ./data/wp-content/plugins)
kubectl cp $SOURCE_PATH $PODNAME:$DESTINATION_PATH
exec
Open console shell in single container pod.
kubectl exec -it $RESOURCE -- /bin/sh
Containers can have a variety of shells, e.g. bash, sh, bin/bash, bin/sh, dash
Open console shell in one of multiple containers in pod.
kubectl exec -it $RESOURCE -c $CONTAINER -- /bin/bash
kubectl exec -it $(kubectl get $TYPE | grep $STRING | awk '{print $1}') -- /bin/sh
Run commands directly in a container.
kubectl exec -it $RESOURCE -- mysql -uwordpress -phard2find --database=blog < ./data/blog.sql
kubectl exec -it $RESOURCE -- chown -R 33:33 /var/www/wp-content-mount
logs
Print the logs for a container in a pod.
kubectl logs $RESOURCE
Follow log.
kubectl logs -f $RESOURCE
kubectl logs -f $(kubectl get pod | grep $STRING | awk '{print $1}')
Follow log in a specific container.
kubectl logs -f -l "$STRING" -c $RESOURCE
Find specific text or results in log.
kubectl logs -f $(kubectl get pod | grep $STRING | awk '{print $1}') | grep -i 'we are authenticated'
kubectl logs -f $(kubectl get pods | grep transformer | awk '{print $1}') | grep 'ingester = ' | awk '{print $4 $5 $6}' | sort -u
port-forward
Forward one or more local ports to a pod port.
kubectl port-forward $RESOURCE 8080:80
kubectl port-forward pod/$RESOURCE 8080:80
kubectl port-forward pod/$(kubectl get pods | grep $STRING| awk '{print $1}') 8080:80
kubectl port-forward service/$RESOURCE 8080:80
Preserve existing environment variables.
sudo -E kubectl port-forward service/$RESOURCE 8080:80
Port-forward local port to server.
ssh -L 6443:localhost:6443 fusiondev1
Manage cronjobs/jobs
Create job from cronjob
kubectl create job --from=cronjob/$RESOURCE manual-$RESOURCE-job
kubectl delete job $(kubectl get job | grep manual | awk '{print $1}') && kubectl create job --from=cronjob/$RESOURCE manual-$RESOURCE-job
Show status of all jobs
kubectl get job -o json | jq -r '.items[] | .metadata.name + ":" + (.status.conditions[] | select(.status == "True") .type + ":" + .status)' | grep $STRING
CRUD type operations
apply
Apply a configuration to a resource by filename or stdin - performs a diff and only applies changes if the resource already exists.
kubectl apply -f $FILENAME
create
Create a resource from a file or from stdin.
kubectl create -f $FILENAME
patch
Update field(s) of a resource using strategic merge patch.
kubectl patch $TYPE $RESOURCE
replace
Replace a resource by filename or stdin.
rollout
Manage the rollout of a resource.
kubectl rollout restart $TYPE $RESOURCE
Cluster/node maintenance
NODE=server-name
Show what pods are on what nodes?
for pod in $(kubectl get pods | grep thewheel | awk '{print $1}'); do echo $pod; kubectl describe pod $pod | grep 'Node:'; done
for pod in $(kubectl get pods -l app=$RESOURCE --no-headers -o name | awk -F '/' '{print $2}'); do echo -n "Checking $pod .... "; kubectl exec "$pod" -- bash -c "ps aux | grep -v grep | grep enable-ssl-passthrough=true" > /dev/null 2>&1 && echo "Good" || echo "Bad"; done
config
kubectl config use-context k3d-k3s-default
kubectl config set-context --current --namespace=team-a
cluster info
kubectl cluster-info
top
Get node resource consumption.
kubectl top node
cordon
Mark node as unschedulable.
kubectl cordon ...
drain
Drain node in preparation for maintenance.
kubectl drain $NODE
uncordon
Mark node as schedulable.
kubectl uncordon ...
proxy
Proxy to a node.
See kubectl proxy
Cluster/node info
api-resources
Print the supported API resources on the server.
api-versions
Print the supported API versions on the server, in the form of "group/version".
completion
Output shell completion code for the specified shell (bash or zsh).
diff
Diff live version against would-be applied version.
explain
Documentation of resources.
top
Display Resource (CPU/Memory/Storage) usage.
version
Print the client and server version information.
Kubectl Aliases
Shorten kubectl commands by creating a file with a starter set of aliases that leverage bash-completion.
On a Mac, install bash-completion with
brew install bash-completion, depending on your version of bash.
vim ~/.kubectl_aliases
# KUBECTL - ALIASES
alias k='kubectl'
alias kd='kubectl describe '
alias kg='kubectl get '
alias kdl='kubectl delete '
alias kex='kubectl exec -it '
alias klo='kubectl logs -f '
alias kgoy='kubectl get -o yaml '
alias cro='cronjob'
alias dep='deployment'
alias sec='secret'
alias spc='secretproviderclass'
# KUBECTL - EXPAND COMMAND ON USE
[ -f ~/.kubectl_aliases ] && source ~/.kubectl_aliases
function kubectl() { echo "+ kubectl $@">&2; command kubectl $@; }
# KUBECTL - ENABLE AUTOCOMPLETION
[[ -r "/usr/local/etc/profile.d/bash_completion.sh" ]] && . "/usr/local/etc/profile.d/bash_completion.sh"
source <(kubectl completion bash)
Spaces at the ends are important, and some shortcuts are not needed because they already exist
Examples:
kg pod
+ kubectl get pod
kdl pod $RESOURCE
+ kubectl delete pod $RESOURCE
kgoy pod $RESOURCE
+ kubectl get -o yaml pod $RESOURCE
kex $(kg pod | grep $RESOURCE | '{print $1}') -- /bin/sh
+ kubectl exec -it $(kubectl get cronjob | grep $RESOURCE | awk '{print $1}') -- /bin/sh