Configure Service Accounts for Pods

Kubernetes offers two distinct ways for clients that run within your cluster, or that otherwise have a relationship to your cluster's control plane to authenticate to the API server.

A service account provides an identity for processes that run in a Pod, and maps to a ServiceAccount object. When you authenticate to the API server, you identify yourself as a particular user. Kubernetes recognises the concept of a user, however, Kubernetes itself does not have a User API.

This task guide is about ServiceAccounts, which do exist in the Kubernetes API. The guide shows you some ways to configure ServiceAccounts for Pods.

Before you begin

You need to have a Kubernetes cluster, and the kubectl command-line tool must be configured to communicate with your cluster. It is recommended to run this tutorial on a cluster with at least two nodes that are not acting as control plane hosts. If you do not already have a cluster, you can create one by using minikube or you can use one of these Kubernetes playgrounds:

Use the default service account to access the API server

When Pods contact the API server, Pods authenticate as a particular ServiceAccount (for example, default). There is always at least one ServiceAccount in each namespace.

Every Kubernetes namespace contains at least one ServiceAccount: the default ServiceAccount for that namespace, named default. If you do not specify a ServiceAccount when you create a Pod, Kubernetes automatically assigns the ServiceAccount named default in that namespace.

You can fetch the details for a Pod you have created. For example:

kubectl get pods/<podname> -o yaml

In the output, you see a field spec.serviceAccountName. Kubernetes automatically sets that value if you don't specify it when you create a Pod.

An application running inside a Pod can access the Kubernetes API using automatically mounted service account credentials. See accessing the Cluster to learn more.

When a Pod authenticates as a ServiceAccount, its level of access depends on the authorization plugin and policy in use.

Opt out of API credential automounting

If you don't want the kubelet to automatically mount a ServiceAccount's API credentials, you can opt out of the default behavior. You can opt out of automounting API credentials on /var/run/secrets/kubernetes.io/serviceaccount/token for a service account by setting automountServiceAccountToken: false on the ServiceAccount:

For example:

apiVersion: v1
kind: ServiceAccount
metadata:
  name: build-robot
automountServiceAccountToken: false
...

You can also opt out of automounting API credentials for a particular Pod:

apiVersion: v1
kind: Pod
metadata:
  name: my-pod
spec:
  serviceAccountName: build-robot
  automountServiceAccountToken: false
  ...

If both the ServiceAccount and the Pod's .spec specify a value for automountServiceAccountToken, the Pod spec takes precedence.

Use more than one ServiceAccount

Every namespace has at least one ServiceAccount: the default ServiceAccount resource, called default. You can list all ServiceAccount resources in your current namespace with:

kubectl get serviceaccounts

The output is similar to this:

NAME      SECRETS    AGE
default   1          1d

You can create additional ServiceAccount objects like this:

kubectl apply -f - <<EOF
apiVersion: v1
kind: ServiceAccount
metadata:
  name: build-robot
EOF

The name of a ServiceAccount object must be a valid DNS subdomain name.

If you get a complete dump of the service account object, like this:

kubectl get serviceaccounts/build-robot -o yaml

The output is similar to this:

apiVersion: v1
kind: ServiceAccount
metadata:
  creationTimestamp: 2019-06-16T00:12:34Z
  name: build-robot
  namespace: default
  resourceVersion: "272500"
  uid: 721ab723-13bc-11e5-aec2-42010af0021e

You can use authorization plugins to set permissions on service accounts.

To use a non-default service account, set the spec.serviceAccountName field of a Pod to the name of the ServiceAccount you wish to use.

You can only set the serviceAccountName field when creating a Pod, or in a template for a new Pod. You cannot update the .spec.serviceAccountName field of a Pod that already exists.

Cleanup

If you tried creating build-robot ServiceAccount from the example above, you can clean it up by running:

kubectl delete serviceaccount/build-robot

Manually create an API token for a ServiceAccount

Suppose you have an existing service account named "build-robot" as mentioned earlier.

You can get a time-limited API token for that ServiceAccount using kubectl:

kubectl create token build-robot

The output from that command is a token that you can use to authenticate as that ServiceAccount. You can request a specific token duration using the --duration command line argument to kubectl create token (the actual duration of the issued token might be shorter, or could even be longer).

Manually create a long-lived API token for a ServiceAccount

If you want to obtain an API token for a ServiceAccount, you create a new Secret with a special annotation, kubernetes.io/service-account.name.

kubectl apply -f - <<EOF
apiVersion: v1
kind: Secret
metadata:
  name: build-robot-secret
  annotations:
    kubernetes.io/service-account.name: build-robot
type: kubernetes.io/service-account-token
EOF

If you view the Secret using:

kubectl get secret/build-robot-secret -o yaml

you can see that the Secret now contains an API token for the "build-robot" ServiceAccount.

Because of the annotation you set, the control plane automatically generates a token for that ServiceAccounts, and stores them into the associated Secret. The control plane also cleans up tokens for deleted ServiceAccounts.

kubectl describe secrets/build-robot-secret

The output is similar to this:

Name:           build-robot-secret
Namespace:      default
Labels:         <none>
Annotations:    kubernetes.io/service-account.name: build-robot
                kubernetes.io/service-account.uid: da68f9c6-9d26-11e7-b84e-002dc52800da

Type:   kubernetes.io/service-account-token

Data
====
ca.crt:         1338 bytes
namespace:      7 bytes
token:          ...

When you delete a ServiceAccount that has an associated Secret, the Kubernetes control plane automatically cleans up the long-lived token from that Secret.

Add ImagePullSecrets to a service account

First, create an imagePullSecret. Next, verify it has been created. For example:

  • Create an imagePullSecret, as described in Specifying ImagePullSecrets on a Pod.

    kubectl create secret docker-registry myregistrykey --docker-server=DUMMY_SERVER \
            --docker-username=DUMMY_USERNAME --docker-password=DUMMY_DOCKER_PASSWORD \
            --docker-email=DUMMY_DOCKER_EMAIL
    
  • Verify it has been created.

    kubectl get secrets myregistrykey
    

    The output is similar to this:

    NAME             TYPE                              DATA    AGE
    myregistrykey    kubernetes.io/.dockerconfigjson   1       1d
    

Add image pull secret to service account

Next, modify the default service account for the namespace to use this Secret as an imagePullSecret.

kubectl patch serviceaccount default -p '{"imagePullSecrets": [{"name": "myregistrykey"}]}'

You can achieve the same outcome by editing the object manually:

kubectl edit serviceaccount/default

The output of the sa.yaml file is similar to this:

Your selected text editor will open with a configuration looking something like this:

apiVersion: v1
kind: ServiceAccount
metadata:
  creationTimestamp: 2021-07-07T22:02:39Z
  name: default
  namespace: default
  resourceVersion: "243024"
  uid: 052fb0f4-3d50-11e5-b066-42010af0d7b6

Using your editor, delete the line with key resourceVersion, add lines for imagePullSecrets: and save it. Leave the uid value set the same as you found it.

After you made those changes, the edited ServiceAccount looks something like this:

apiVersion: v1
kind: ServiceAccount
metadata:
  creationTimestamp: 2021-07-07T22:02:39Z
  name: default
  namespace: default
  uid: 052fb0f4-3d50-11e5-b066-42010af0d7b6
imagePullSecrets:
  - name: myregistrykey

Verify that imagePullSecrets are set for new Pods

Now, when a new Pod is created in the current namespace and using the default ServiceAccount, the new Pod has its spec.imagePullSecrets field set automatically:

kubectl run nginx --image=nginx --restart=Never
kubectl get pod nginx -o=jsonpath='{.spec.imagePullSecrets[0].name}{"\n"}'

The output is:

myregistrykey

ServiceAccount token volume projection

FEATURE STATE: Kubernetes v1.20 [stable]

The kubelet can also project a ServiceAccount token into a Pod. You can specify desired properties of the token, such as the audience and the validity duration. These properties are not configurable on the default ServiceAccount token. The token will also become invalid against the API when either the Pod or the ServiceAccount is deleted.

You can configure this behavior for the spec of a Pod using a projected volume type called ServiceAccountToken.

Launch a Pod using service account token projection

To provide a Pod with a token with an audience of vault and a validity duration of two hours, you could define a Pod manifest that is similar to:

apiVersion: v1
kind: Pod
metadata:
  name: nginx
spec:
  containers:
  - image: nginx
    name: nginx
    volumeMounts:
    - mountPath: /var/run/secrets/tokens
      name: vault-token
  serviceAccountName: build-robot
  volumes:
  - name: vault-token
    projected:
      sources:
      - serviceAccountToken:
          path: vault-token
          expirationSeconds: 7200
          audience: vault

Create the Pod:

kubectl create -f https://k8s.io/examples/pods/pod-projected-svc-token.yaml

The kubelet will: request and store the token on behalf of the Pod; make the token available to the Pod at a configurable file path; and refresh the token as it approaches expiration. The kubelet proactively requests rotation for the token if it is older than 80% of its total time-to-live (TTL), or if the token is older than 24 hours.

The application is responsible for reloading the token when it rotates. It's often good enough for the application to load the token on a schedule (for example: once every 5 minutes), without tracking the actual expiry time.

Service account issuer discovery

FEATURE STATE: Kubernetes v1.21 [stable]

If you have enabled token projection for ServiceAccounts in your cluster, then you can also make use of the discovery feature. Kubernetes provides a way for clients to federate as an identity provider, so that one or more external systems can act as a relying party.

When enabled, the Kubernetes API server publishes an OpenID Provider Configuration document via HTTP. The configuration document is published at /.well-known/openid-configuration. The OpenID Provider Configuration is sometimes referred to as the discovery document. The Kubernetes API server publishes the related JSON Web Key Set (JWKS), also via HTTP, at /openid/v1/jwks.

Clusters that use RBAC include a default ClusterRole called system:service-account-issuer-discovery. A default ClusterRoleBinding assigns this role to the system:serviceaccounts group, which all ServiceAccounts implicitly belong to. This allows pods running on the cluster to access the service account discovery document via their mounted service account token. Administrators may, additionally, choose to bind the role to system:authenticated or system:unauthenticated depending on their security requirements and which external systems they intend to federate with.

The JWKS response contains public keys that a relying party can use to validate the Kubernetes service account tokens. Relying parties first query for the OpenID Provider Configuration, and use the jwks_uri field in the response to find the JWKS.

In many cases, Kubernetes API servers are not available on the public internet, but public endpoints that serve cached responses from the API server can be made available by users or by service providers. In these cases, it is possible to override the jwks_uri in the OpenID Provider Configuration so that it points to the public endpoint, rather than the API server's address, by passing the --service-account-jwks-uri flag to the API server. Like the issuer URL, the JWKS URI is required to use the https scheme.

What's next

See also: