kubectl
is a CLI tool to control a cluster Kubernetes.
$ docker run --rm wernight/kubectl --help
Note: Entrypoint is set to kubectl so do not type wernight/kubectl kubectl
.
For example to access a local Kubernetes cluster you may run:
$ docker run --rm --net=host --user $UID \
-v ~/.kube:/config/.kube \
wernight/kubectl cluster-info
-net=host
: (optional) allows to connect to a local Kubernetes cluster.--user $UID
: (optional) by default runs as random UID2342
, this allows to access your existing~/.kube
if you have one. As you can note, you can runkubectl
as any UID/GID.-v XXX:/config
: (optional) allows to store its configuration and possibly access existing configuration. Note that/config
will always be the directory containing.kube
(it's the forcedHOME
directory). Can be read-only. Alternatively you can mount a Kubernetes service account for example:-v /var/run/secrets/kubernetes.io/serviceaccount:/var/run/secrets/kubernetes.io/serviceaccount:ro
.
Here we use the service-account, so this should work from within a Pod on your cluster as long as you've docker installed (and may be DOCKER_HOST
set up properly):
$ docker run \
-v /var/run/secrets/kubernetes.io/serviceaccount/:/var/run/secrets/kubernetes.io/serviceaccount/:ro \
wernight/kubectl \
-s https://kubernetes \
--token="$(cat /var/run/secrets/kubernetes.io/serviceaccount/token)" \
--certificate-authority=/var/run/secrets/kubernetes.io/serviceaccount/ca.crt \
cluster-info
Note: Alternatively to using kube-dns, you can use environment variables set within Kubernetes containers: https://$KUBERNETES_SERVICE_HOST:$KUBERNETES_PORT_443_TCP_PORT
.
You may setup an alias to run this is if you were running kubectl
directly.
Here is a function POSIX-compatible that work for most shells:
kubectl () {
docker run --rm -it --user $UID:$GID \
-v /var/run/secrets/kubernetes.io/serviceaccount:/var/run/secrets/kubernetes.io/serviceaccount:ro \
-w /code -v "$PWD":/code:ro \
wernight/kubectl "$@"
}
It's mostly meant to be used during continuous integration or as part of an automated build/deployment:
- So that your machine (e.g. build server) doesn't need
kubectl
to be installed; only Docker. - To avoid
kubectl config use-context
and similar to affect your build and other projects' builds.
Suggestions are welcome on our GitHub issue tracker.