This is the multi-page printable view of this section. Click here to print.
For an introduction to how Kubernetes implements and controls API access, read Controlling Access to the Kubernetes API.
Reference documentation:
This page provides an overview of authenticating.
All Kubernetes clusters have two categories of users: service accounts managed by Kubernetes, and normal users.
It is assumed that a cluster-independent service manages normal users in the following ways:
In this regard, Kubernetes does not have objects which represent normal user accounts. Normal users cannot be added to a cluster through an API call.
Even though a normal user cannot be added via an API call, any user that presents a valid certificate signed by the cluster's certificate authority (CA) is considered authenticated. In this configuration, Kubernetes determines the username from the common name field in the 'subject' of the cert (e.g., "/CN=bob"). From there, the role based access control (RBAC) sub-system would determine whether the user is authorized to perform a specific operation on a resource. For more details, refer to the normal users topic in certificate request for more details about this.
In contrast, service accounts are users managed by the Kubernetes API. They are
bound to specific namespaces, and created automatically by the API server or
manually through API calls. Service accounts are tied to a set of credentials
stored as Secrets, which are mounted into pods allowing in-cluster processes
to talk to the Kubernetes API.
API requests are tied to either a normal user or a service account, or are treated
as anonymous requests. This means every process inside or outside the cluster, from
a human user typing kubectl on a workstation, to kubelets on nodes, to members
of the control plane, must authenticate when making requests to the API server,
or be treated as an anonymous user.
Kubernetes uses client certificates, bearer tokens, or an authenticating proxy to authenticate API requests through authentication plugins. As HTTP requests are made to the API server, plugins attempt to associate the following attributes with the request:
kube-admin or jane@example.com.system:masters or devops-team.All values are opaque to the authentication system and only hold significance when interpreted by an authorizer.
You can enable multiple authentication methods at once. You should usually use at least two methods:
When multiple authenticator modules are enabled, the first module to successfully authenticate the request short-circuits evaluation. The API server does not guarantee the order authenticators run in.
The system:authenticated group is included in the list of groups for all authenticated users.
Integrations with other authentication protocols (LDAP, SAML, Kerberos, alternate x509 schemes, etc) can be accomplished using an authenticating proxy or the authentication webhook.
Client certificate authentication is enabled by passing the --client-ca-file=SOMEFILE
option to API server. The referenced file must contain one or more certificate authorities
to use to validate client certificates presented to the API server. If a client certificate
is presented and verified, the common name of the subject is used as the user name for the
request. As of Kubernetes 1.4, client certificates can also indicate a user's group memberships
using the certificate's organization fields. To include multiple group memberships for a user,
include multiple organization fields in the certificate.
For example, using the openssl command line tool to generate a certificate signing request:
openssl req -new -key jbeda.pem -out jbeda-csr.pem -subj "/CN=jbeda/O=app1/O=app2"
This would create a CSR for the username "jbeda", belonging to two groups, "app1" and "app2".
See Managing Certificates for how to generate a client cert.
The API server reads bearer tokens from a file when given the --token-auth-file=SOMEFILE option on the command line. Currently, tokens last indefinitely, and the token list cannot be
changed without restarting the API server.
The token file is a csv file with a minimum of 3 columns: token, user name, user uid, followed by optional group names.
If you have more than one group the column must be double quoted e.g.
token,user,uid,"group1,group2,group3"
When using bearer token authentication from an http client, the API
server expects an Authorization header with a value of Bearer <token>. The bearer token must be a character sequence that can be
put in an HTTP header value using no more than the encoding and
quoting facilities of HTTP. For example: if the bearer token is
31ada4fd-adec-460c-809a-9e56ceb75269 then it would appear in an HTTP
header as shown below.
Authorization: Bearer 31ada4fd-adec-460c-809a-9e56ceb75269
Kubernetes v1.18 [stable]
To allow for streamlined bootstrapping for new clusters, Kubernetes includes a
dynamically-managed Bearer token type called a Bootstrap Token. These tokens
are stored as Secrets in the kube-system namespace, where they can be
dynamically managed and created. Controller Manager contains a TokenCleaner
controller that deletes bootstrap tokens as they expire.
The tokens are of the form [a-z0-9]{6}.[a-z0-9]{16}. The first component is a
Token ID and the second component is the Token Secret. You specify the token
in an HTTP header as follows:
Authorization: Bearer 781292.db7bc3a58fc5f07e
You must enable the Bootstrap Token Authenticator with the
--enable-bootstrap-token-auth flag on the API Server. You must enable
the TokenCleaner controller via the --controllers flag on the Controller
Manager. This is done with something like --controllers=*,tokencleaner.
kubeadm will do this for you if you are using it to bootstrap a cluster.
The authenticator authenticates as system:bootstrap:<Token ID>. It is
included in the system:bootstrappers group. The naming and groups are
intentionally limited to discourage users from using these tokens past
bootstrapping. The user names and group can be used (and are used by kubeadm)
to craft the appropriate authorization policies to support bootstrapping a
cluster.
Please see Bootstrap Tokens for in depth
documentation on the Bootstrap Token authenticator and controllers along with
how to manage these tokens with kubeadm.
A service account is an automatically enabled authenticator that uses signed bearer tokens to verify requests. The plugin takes two optional flags:
--service-account-key-file A file containing a PEM encoded key for signing bearer tokens.
If unspecified, the API server's TLS private key will be used.--service-account-lookup If enabled, tokens which are deleted from the API will be revoked.Service accounts are usually created automatically by the API server and
associated with pods running in the cluster through the ServiceAccount
Admission Controller. Bearer tokens are
mounted into pods at well-known locations, and allow in-cluster processes to
talk to the API server. Accounts may be explicitly associated with pods using the
serviceAccountName field of a PodSpec.
serviceAccountName is usually omitted because this is done automatically.
apiVersion: apps/v1 # this apiVersion is relevant as of Kubernetes 1.9
kind: Deployment
metadata:
name: nginx-deployment
namespace: default
spec:
replicas: 3
template:
metadata:
# ...
spec:
serviceAccountName: bob-the-bot
containers:
- name: nginx
image: nginx:1.14.2
Service account bearer tokens are perfectly valid to use outside the cluster and
can be used to create identities for long standing jobs that wish to talk to the
Kubernetes API. To manually create a service account, use the kubectl create serviceaccount (NAME) command. This creates a service account in the current
namespace and an associated secret.
kubectl create serviceaccount jenkins
serviceaccount "jenkins" created
Check an associated secret:
kubectl get serviceaccounts jenkins -o yaml
apiVersion: v1
kind: ServiceAccount
metadata:
# ...
secrets:
- name: jenkins-token-1yvwg
The created secret holds the public CA of the API server and a signed JSON Web Token (JWT).
kubectl get secret jenkins-token-1yvwg -o yaml
apiVersion: v1
data:
ca.crt: (APISERVER'S CA BASE64 ENCODED)
namespace: ZGVmYXVsdA==
token: (BEARER TOKEN BASE64 ENCODED)
kind: Secret
metadata:
# ...
type: kubernetes.io/service-account-token
The signed JWT can be used as a bearer token to authenticate as the given service account. See above for how the token is included in a request. Normally these secrets are mounted into pods for in-cluster access to the API server, but can be used from outside the cluster as well.
Service accounts authenticate with the username system:serviceaccount:(NAMESPACE):(SERVICEACCOUNT),
and are assigned to the groups system:serviceaccounts and system:serviceaccounts:(NAMESPACE).
WARNING: Because service account tokens are stored in secrets, any user with read access to those secrets can authenticate as the service account. Be cautious when granting permissions to service accounts and read capabilities for secrets.
OpenID Connect is a flavor of OAuth2 supported by some OAuth2 providers, notably Azure Active Directory, Salesforce, and Google. The protocol's main extension of OAuth2 is an additional field returned with the access token called an ID Token. This token is a JSON Web Token (JWT) with well known fields, such as a user's email, signed by the server.
To identify the user, the authenticator uses the id_token (not the access_token)
from the OAuth2 token response
as a bearer token. See above for how the token
is included in a request.
access_token, id_token and a refresh_tokenkubectl, use your id_token with the --token flag or add it directly to your kubeconfigkubectl sends your id_token in a header called Authorization to the API serverid_token hasn't expiredkubectlkubectl provides feedback to the userSince all of the data needed to validate who you are is in the id_token, Kubernetes doesn't need to
"phone home" to the identity provider. In a model where every request is stateless this provides a very scalable solution for authentication. It does offer a few challenges:
id_token can't be revoked, it's like a certificate so it should be short-lived (only a few minutes) so it can be very annoying to have to get a new token every few minutes.kubectl proxy command or a reverse proxy that injects the id_token.To enable the plugin, configure the following flags on the API server:
| Parameter | Description | Example | Required |
|---|---|---|---|
--oidc-issuer-url |
URL of the provider which allows the API server to discover public signing keys. Only URLs which use the https:// scheme are accepted. This is typically the provider's discovery URL without a path, for example "https://accounts.google.com" or "https://login.salesforce.com". This URL should point to the level below .well-known/openid-configuration |
If the discovery URL is https://accounts.google.com/.well-known/openid-configuration, the value should be https://accounts.google.com |
Yes |
--oidc-client-id |
A client id that all tokens must be issued for. | kubernetes | Yes |
--oidc-username-claim |
JWT claim to use as the user name. By default sub, which is expected to be a unique identifier of the end user. Admins can choose other claims, such as email or name, depending on their provider. However, claims other than email will be prefixed with the issuer URL to prevent naming clashes with other plugins. |
sub | No |
--oidc-username-prefix |
Prefix prepended to username claims to prevent clashes with existing names (such as system: users). For example, the value oidc: will create usernames like oidc:jane.doe. If this flag isn't provided and --oidc-username-claim is a value other than email the prefix defaults to ( Issuer URL )# where ( Issuer URL ) is the value of --oidc-issuer-url. The value - can be used to disable all prefixing. |
oidc: |
No |
--oidc-groups-claim |
JWT claim to use as the user's group. If the claim is present it must be an array of strings. | groups | No |
--oidc-groups-prefix |
Prefix prepended to group claims to prevent clashes with existing names (such as system: groups). For example, the value oidc: will create group names like oidc:engineering and oidc:infra. |
oidc: |
No |
--oidc-required-claim |
A key=value pair that describes a required claim in the ID Token. If set, the claim is verified to be present in the ID Token with a matching value. Repeat this flag to specify multiple claims. | claim=value |
No |
--oidc-ca-file |
The path to the certificate for the CA that signed your identity provider's web certificate. Defaults to the host's root CAs. | /etc/kubernetes/ssl/kc-ca.pem |
No |
Importantly, the API server is not an OAuth2 client, rather it can only be
configured to trust a single issuer. This allows the use of public providers,
such as Google, without trusting credentials issued to third parties. Admins who
wish to utilize multiple OAuth clients should explore providers which support the
azp (authorized party) claim, a mechanism for allowing one client to issue
tokens on behalf of another.
Kubernetes does not provide an OpenID Connect Identity Provider. You can use an existing public OpenID Connect Identity Provider (such as Google, or others). Or, you can run your own Identity Provider, such as dex, Keycloak, CloudFoundry UAA, or Tremolo Security's OpenUnison.
For an identity provider to work with Kubernetes it must:
A note about requirement #3 above, requiring a CA signed certificate. If you deploy your own identity provider (as opposed to one of the cloud providers like Google or Microsoft) you MUST have your identity provider's web server certificate signed by a certificate with the CA flag set to TRUE, even if it is self signed. This is due to GoLang's TLS client implementation being very strict to the standards around certificate validation. If you don't have a CA handy, you can use this script from the Dex team to create a simple CA and a signed certificate and key pair.
Or you can use this similar script that generates SHA256 certs with a longer life and larger key size.
Setup instructions for specific systems:
The first option is to use the kubectl oidc authenticator, which sets the id_token as a bearer token for all requests and refreshes the token once it expires. After you've logged into your provider, use kubectl to add your id_token, refresh_token, client_id, and client_secret to configure the plugin.
Providers that don't return an id_token as part of their refresh token response aren't supported by this plugin and should use "Option 2" below.
kubectl config set-credentials USER_NAME \
--auth-provider=oidc \
--auth-provider-arg=idp-issuer-url=( issuer url ) \
--auth-provider-arg=client-id=( your client id ) \
--auth-provider-arg=client-secret=( your client secret ) \
--auth-provider-arg=refresh-token=( your refresh token ) \
--auth-provider-arg=idp-certificate-authority=( path to your ca certificate ) \
--auth-provider-arg=id-token=( your id_token )
As an example, running the below command after authenticating to your identity provider:
kubectl config set-credentials mmosley \
--auth-provider=oidc \
--auth-provider-arg=idp-issuer-url=https://oidcidp.tremolo.lan:8443/auth/idp/OidcIdP \
--auth-provider-arg=client-id=kubernetes \
--auth-provider-arg=client-secret=1db158f6-177d-4d9c-8a8b-d36869918ec5 \
--auth-provider-arg=refresh-token=q1bKLFOyUiosTfawzA93TzZIDzH2TNa2SMm0zEiPKTUwME6BkEo6Sql5yUWVBSWpKUGphaWpxSVAfekBOZbBhaEW+VlFUeVRGcluyVF5JT4+haZmPsluFoFu5XkpXk5BXqHega4GAXlF+ma+vmYpFcHe5eZR+slBFpZKtQA= \
--auth-provider-arg=idp-certificate-authority=/root/ca.pem \
--auth-provider-arg=id-token=eyJraWQiOiJDTj1vaWRjaWRwLnRyZW1vbG8ubGFuLCBPVT1EZW1vLCBPPVRybWVvbG8gU2VjdXJpdHksIEw9QXJsaW5ndG9uLCBTVD1WaXJnaW5pYSwgQz1VUy1DTj1rdWJlLWNhLTEyMDIxNDc5MjEwMzYwNzMyMTUyIiwiYWxnIjoiUlMyNTYifQ.eyJpc3MiOiJodHRwczovL29pZGNpZHAudHJlbW9sby5sYW46ODQ0My9hdXRoL2lkcC9PaWRjSWRQIiwiYXVkIjoia3ViZXJuZXRlcyIsImV4cCI6MTQ4MzU0OTUxMSwianRpIjoiMm96US15TXdFcHV4WDlHZUhQdy1hZyIsImlhdCI6MTQ4MzU0OTQ1MSwibmJmIjoxNDgzNTQ5MzMxLCJzdWIiOiI0YWViMzdiYS1iNjQ1LTQ4ZmQtYWIzMC0xYTAxZWU0MWUyMTgifQ.w6p4J_6qQ1HzTG9nrEOrubxIMb9K5hzcMPxc9IxPx2K4xO9l-oFiUw93daH3m5pluP6K7eOE6txBuRVfEcpJSwlelsOsW8gb8VJcnzMS9EnZpeA0tW_p-mnkFc3VcfyXuhe5R3G7aa5d8uHv70yJ9Y3-UhjiN9EhpMdfPAoEB9fYKKkJRzF7utTTIPGrSaSU6d2pcpfYKaxIwePzEkT4DfcQthoZdy9ucNvvLoi1DIC-UocFD8HLs8LYKEqSxQvOcvnThbObJ9af71EwmuE21fO5KzMW20KtAeget1gnldOosPtz1G5EwvaQ401-RPQzPGMVBld0_zMCAwZttJ4knw
Which would produce the below configuration:
users:
- name: mmosley
user:
auth-provider:
config:
client-id: kubernetes
client-secret: 1db158f6-177d-4d9c-8a8b-d36869918ec5
id-token: eyJraWQiOiJDTj1vaWRjaWRwLnRyZW1vbG8ubGFuLCBPVT1EZW1vLCBPPVRybWVvbG8gU2VjdXJpdHksIEw9QXJsaW5ndG9uLCBTVD1WaXJnaW5pYSwgQz1VUy1DTj1rdWJlLWNhLTEyMDIxNDc5MjEwMzYwNzMyMTUyIiwiYWxnIjoiUlMyNTYifQ.eyJpc3MiOiJodHRwczovL29pZGNpZHAudHJlbW9sby5sYW46ODQ0My9hdXRoL2lkcC9PaWRjSWRQIiwiYXVkIjoia3ViZXJuZXRlcyIsImV4cCI6MTQ4MzU0OTUxMSwianRpIjoiMm96US15TXdFcHV4WDlHZUhQdy1hZyIsImlhdCI6MTQ4MzU0OTQ1MSwibmJmIjoxNDgzNTQ5MzMxLCJzdWIiOiI0YWViMzdiYS1iNjQ1LTQ4ZmQtYWIzMC0xYTAxZWU0MWUyMTgifQ.w6p4J_6qQ1HzTG9nrEOrubxIMb9K5hzcMPxc9IxPx2K4xO9l-oFiUw93daH3m5pluP6K7eOE6txBuRVfEcpJSwlelsOsW8gb8VJcnzMS9EnZpeA0tW_p-mnkFc3VcfyXuhe5R3G7aa5d8uHv70yJ9Y3-UhjiN9EhpMdfPAoEB9fYKKkJRzF7utTTIPGrSaSU6d2pcpfYKaxIwePzEkT4DfcQthoZdy9ucNvvLoi1DIC-UocFD8HLs8LYKEqSxQvOcvnThbObJ9af71EwmuE21fO5KzMW20KtAeget1gnldOosPtz1G5EwvaQ401-RPQzPGMVBld0_zMCAwZttJ4knw
idp-certificate-authority: /root/ca.pem
idp-issuer-url: https://oidcidp.tremolo.lan:8443/auth/idp/OidcIdP
refresh-token: q1bKLFOyUiosTfawzA93TzZIDzH2TNa2SMm0zEiPKTUwME6BkEo6Sql5yUWVBSWpKUGphaWpxSVAfekBOZbBhaEW+VlFUeVRGcluyVF5JT4+haZmPsluFoFu5XkpXk5BXq
name: oidc
Once your id_token expires, kubectl will attempt to refresh your id_token using your refresh_token and client_secret storing the new values for the refresh_token and id_token in your .kube/config.
--token OptionThe kubectl command lets you pass in a token using the --token option. Copy and paste the id_token into this option:
kubectl --token=eyJhbGciOiJSUzI1NiJ9.eyJpc3MiOiJodHRwczovL21sYi50cmVtb2xvLmxhbjo4MDQzL2F1dGgvaWRwL29pZGMiLCJhdWQiOiJrdWJlcm5ldGVzIiwiZXhwIjoxNDc0NTk2NjY5LCJqdGkiOiI2RDUzNXoxUEpFNjJOR3QxaWVyYm9RIiwiaWF0IjoxNDc0NTk2MzY5LCJuYmYiOjE0NzQ1OTYyNDksInN1YiI6Im13aW5kdSIsInVzZXJfcm9sZSI6WyJ1c2VycyIsIm5ldy1uYW1lc3BhY2Utdmlld2VyIl0sImVtYWlsIjoibXdpbmR1QG5vbW9yZWplZGkuY29tIn0.f2As579n9VNoaKzoF-dOQGmXkFKf1FMyNV0-va_B63jn-_n9LGSCca_6IVMP8pO-Zb4KvRqGyTP0r3HkHxYy5c81AnIh8ijarruczl-TK_yF5akjSTHFZD-0gRzlevBDiH8Q79NAr-ky0P4iIXS8lY9Vnjch5MF74Zx0c3alKJHJUnnpjIACByfF2SCaYzbWFMUNat-K1PaUk5-ujMBG7yYnr95xD-63n8CO8teGUAAEMx6zRjzfhnhbzX-ajwZLGwGUBT4WqjMs70-6a7_8gZmLZb2az1cZynkFRj2BaCkVT3A2RrjeEwZEtGXlMqKJ1_I2ulrOVsYx01_yD35-rw get nodes
Webhook authentication is a hook for verifying bearer tokens.
--authentication-token-webhook-config-file a configuration file describing how to access the remote webhook service.--authentication-token-webhook-cache-ttl how long to cache authentication decisions. Defaults to two minutes.--authentication-token-webhook-version determines whether to use authentication.k8s.io/v1beta1 or authentication.k8s.io/v1
TokenReview objects to send/receive information from the webhook. Defaults to v1beta1.The configuration file uses the kubeconfig
file format. Within the file, clusters refers to the remote service and
users refers to the API server webhook. An example would be:
# Kubernetes API version
apiVersion: v1
# kind of the API object
kind: Config
# clusters refers to the remote service.
clusters:
- name: name-of-remote-authn-service
cluster:
certificate-authority: /path/to/ca.pem # CA for verifying the remote service.
server: https://authn.example.com/authenticate # URL of remote service to query. 'https' recommended for production.
# users refers to the API server's webhook configuration.
users:
- name: name-of-api-server
user:
client-certificate: /path/to/cert.pem # cert for the webhook plugin to use
client-key: /path/to/key.pem # key matching the cert
# kubeconfig files require a context. Provide one for the API server.
current-context: webhook
contexts:
- context:
cluster: name-of-remote-authn-service
user: name-of-api-server
name: webhook
When a client attempts to authenticate with the API server using a bearer token as discussed above,
the authentication webhook POSTs a JSON-serialized TokenReview object containing the token to the remote service.
Note that webhook API objects are subject to the same versioning compatibility rules as other Kubernetes API objects.
Implementers should check the apiVersion field of the request to ensure correct deserialization,
and must respond with a TokenReview object of the same version as the request.
authentication.k8s.io/v1beta1 token reviews for backwards compatibility.
To opt into receiving authentication.k8s.io/v1 token reviews, the API server must be started with --authentication-token-webhook-version=v1.
{
"apiVersion": "authentication.k8s.io/v1",
"kind": "TokenReview",
"spec": {
# Opaque bearer token sent to the API server
"token": "014fbff9a07c...",
# Optional list of the audience identifiers for the server the token was presented to.
# Audience-aware token authenticators (for example, OIDC token authenticators)
# should verify the token was intended for at least one of the audiences in this list,
# and return the intersection of this list and the valid audiences for the token in the response status.
# This ensures the token is valid to authenticate to the server it was presented to.
# If no audiences are provided, the token should be validated to authenticate to the Kubernetes API server.
"audiences": ["https://myserver.example.com", "https://myserver.internal.example.com"]
}
}
{
"apiVersion": "authentication.k8s.io/v1beta1",
"kind": "TokenReview",
"spec": {
# Opaque bearer token sent to the API server
"token": "014fbff9a07c...",
# Optional list of the audience identifiers for the server the token was presented to.
# Audience-aware token authenticators (for example, OIDC token authenticators)
# should verify the token was intended for at least one of the audiences in this list,
# and return the intersection of this list and the valid audiences for the token in the response status.
# This ensures the token is valid to authenticate to the server it was presented to.
# If no audiences are provided, the token should be validated to authenticate to the Kubernetes API server.
"audiences": ["https://myserver.example.com", "https://myserver.internal.example.com"]
}
}
The remote service is expected to fill the status field of the request to indicate the success of the login.
The response body's spec field is ignored and may be omitted.
The remote service must return a response using the same TokenReview API version that it received.
A successful validation of the bearer token would return:
{
"apiVersion": "authentication.k8s.io/v1",
"kind": "TokenReview",
"status": {
"authenticated": true,
"user": {
# Required
"username": "janedoe@example.com",
# Optional
"uid": "42",
# Optional group memberships
"groups": ["developers", "qa"],
# Optional additional information provided by the authenticator.
# This should not contain confidential data, as it can be recorded in logs
# or API objects, and is made available to admission webhooks.
"extra": {
"extrafield1": [
"extravalue1",
"extravalue2"
]
}
},
# Optional list audience-aware token authenticators can return,
# containing the audiences from the `spec.audiences` list for which the provided token was valid.
# If this is omitted, the token is considered to be valid to authenticate to the Kubernetes API server.
"audiences": ["https://myserver.example.com"]
}
}
{
"apiVersion": "authentication.k8s.io/v1beta1",
"kind": "TokenReview",
"status": {
"authenticated": true,
"user": {
# Required
"username": "janedoe@example.com",
# Optional
"uid": "42",
# Optional group memberships
"groups": ["developers", "qa"],
# Optional additional information provided by the authenticator.
# This should not contain confidential data, as it can be recorded in logs
# or API objects, and is made available to admission webhooks.
"extra": {
"extrafield1": [
"extravalue1",
"extravalue2"
]
}
},
# Optional list audience-aware token authenticators can return,
# containing the audiences from the `spec.audiences` list for which the provided token was valid.
# If this is omitted, the token is considered to be valid to authenticate to the Kubernetes API server.
"audiences": ["https://myserver.example.com"]
}
}
An unsuccessful request would return:
{
"apiVersion": "authentication.k8s.io/v1",
"kind": "TokenReview",
"status": {
"authenticated": false,
# Optionally include details about why authentication failed.
# If no error is provided, the API will return a generic Unauthorized message.
# The error field is ignored when authenticated=true.
"error": "Credentials are expired"
}
}
{
"apiVersion": "authentication.k8s.io/v1beta1",
"kind": "TokenReview",
"status": {
"authenticated": false,
# Optionally include details about why authentication failed.
# If no error is provided, the API will return a generic Unauthorized message.
# The error field is ignored when authenticated=true.
"error": "Credentials are expired"
}
}
The API server can be configured to identify users from request header values, such as X-Remote-User.
It is designed for use in combination with an authenticating proxy, which sets the request header value.
--requestheader-username-headers Required, case-insensitive. Header names to check, in order, for the user identity. The first header containing a value is used as the username.--requestheader-group-headers 1.6+. Optional, case-insensitive. "X-Remote-Group" is suggested. Header names to check, in order, for the user's groups. All values in all specified headers are used as group names.--requestheader-extra-headers-prefix 1.6+. Optional, case-insensitive. "X-Remote-Extra-" is suggested. Header prefixes to look for to determine extra information about the user (typically used by the configured authorization plugin). Any headers beginning with any of the specified prefixes have the prefix removed. The remainder of the header name is lowercased and percent-decoded and becomes the extra key, and the header value is the extra value.For example, with this configuration:
--requestheader-username-headers=X-Remote-User
--requestheader-group-headers=X-Remote-Group
--requestheader-extra-headers-prefix=X-Remote-Extra-
this request:
GET / HTTP/1.1
X-Remote-User: fido
X-Remote-Group: dogs
X-Remote-Group: dachshunds
X-Remote-Extra-Acme.com%2Fproject: some-project
X-Remote-Extra-Scopes: openid
X-Remote-Extra-Scopes: profile
would result in this user info:
name: fido
groups:
- dogs
- dachshunds
extra:
acme.com/project:
- some-project
scopes:
- openid
- profile
In order to prevent header spoofing, the authenticating proxy is required to present a valid client certificate to the API server for validation against the specified CA before the request headers are checked. WARNING: do not reuse a CA that is used in a different context unless you understand the risks and the mechanisms to protect the CA's usage.
--requestheader-client-ca-file Required. PEM-encoded certificate bundle. A valid client certificate must be presented and validated against the certificate authorities in the specified file before the request headers are checked for user names.--requestheader-allowed-names Optional. List of Common Name values (CNs). If set, a valid client certificate with a CN in the specified list must be presented before the request headers are checked for user names. If empty, any CN is allowed.When enabled, requests that are not rejected by other configured authentication methods are
treated as anonymous requests, and given a username of system:anonymous and a group of
system:unauthenticated.
For example, on a server with token authentication configured, and anonymous access enabled,
a request providing an invalid bearer token would receive a 401 Unauthorized error.
A request providing no bearer token would be treated as an anonymous request.
In 1.5.1-1.5.x, anonymous access is disabled by default, and can be enabled by
passing the --anonymous-auth=true option to the API server.
In 1.6+, anonymous access is enabled by default if an authorization mode other than AlwaysAllow
is used, and can be disabled by passing the --anonymous-auth=false option to the API server.
Starting in 1.6, the ABAC and RBAC authorizers require explicit authorization of the
system:anonymous user or the system:unauthenticated group, so legacy policy rules
that grant access to the * user or * group do not include anonymous users.
A user can act as another user through impersonation headers. These let requests manually override the user info a request authenticates as. For example, an admin could use this feature to debug an authorization policy by temporarily impersonating another user and seeing if a request was denied.
Impersonation requests first authenticate as the requesting user, then switch to the impersonated user info.
The following HTTP headers can be used to performing an impersonation request:
Impersonate-User: The username to act as.Impersonate-Group: A group name to act as. Can be provided multiple times to set multiple groups. Optional. Requires "Impersonate-User".Impersonate-Extra-( extra name ): A dynamic header used to associate extra fields with the user. Optional. Requires "Impersonate-User". In order to be preserved consistently, ( extra name ) must be lower-case, and any characters which aren't legal in HTTP header labels MUST be utf8 and percent-encoded.Impersonate-Uid: A unique identifier that represents the user being impersonated. Optional. Requires "Impersonate-User". Kubernetes does not impose any format requirements on this string.( extra name ) could only contain characters which were legal in HTTP header labels.
Impersonate-Uid is only available in versions 1.22.0 and higher.
An example of the impersonation headers used when impersonating a user with groups:
Impersonate-User: jane.doe@example.com
Impersonate-Group: developers
Impersonate-Group: admins
An example of the impersonation headers used when impersonating a user with a UID and extra fields:
Impersonate-User: jane.doe@example.com
Impersonate-Extra-dn: cn=jane,ou=engineers,dc=example,dc=com
Impersonate-Extra-acme.com%2Fproject: some-project
Impersonate-Extra-scopes: view
Impersonate-Extra-scopes: development
Impersonate-Uid: 06f6ce97-e2c5-4ab8-7ba5-7654dd08d52b
When using kubectl set the --as flag to configure the Impersonate-User
header, set the --as-group flag to configure the Impersonate-Group header.
kubectl drain mynode
Error from server (Forbidden): User "clark" cannot get nodes at the cluster scope. (get nodes mynode)
Set the --as and --as-group flag:
kubectl drain mynode --as=superman --as-group=system:masters
node/mynode cordoned
node/mynode drained
kubectl cannot impersonate extra fields or UIDs.
To impersonate a user, group, user identifier (UID) or extra fields, the impersonating user must have the ability to perform the "impersonate" verb on the kind of attribute being impersonated ("user", "group", "uid", etc.). For clusters that enable the RBAC authorization plugin, the following ClusterRole encompasses the rules needed to set user and group impersonation headers:
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
name: impersonator
rules:
- apiGroups: [""]
resources: ["users", "groups", "serviceaccounts"]
verbs: ["impersonate"]
For impersonation, extra fields and impersonated UIDs are both under the "authentication.k8s.io" apiGroup.
Extra fields are evaluated as sub-resources of the resource "userextras". To
allow a user to use impersonation headers for the extra field "scopes" and
for UIDs, a user should be granted the following role:
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
name: scopes-and-uid-impersonator
rules:
# Can set "Impersonate-Extra-scopes" header and the "Impersonate-Uid" header.
- apiGroups: ["authentication.k8s.io"]
resources: ["userextras/scopes", "uids"]
verbs: ["impersonate"]
The values of impersonation headers can also be restricted by limiting the set
of resourceNames a resource can take.
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
name: limited-impersonator
rules:
# Can impersonate the user "jane.doe@example.com"
- apiGroups: [""]
resources: ["users"]
verbs: ["impersonate"]
resourceNames: ["jane.doe@example.com"]
# Can impersonate the groups "developers" and "admins"
- apiGroups: [""]
resources: ["groups"]
verbs: ["impersonate"]
resourceNames: ["developers","admins"]
# Can impersonate the extras field "scopes" with the values "view" and "development"
- apiGroups: ["authentication.k8s.io"]
resources: ["userextras/scopes"]
verbs: ["impersonate"]
resourceNames: ["view", "development"]
# Can impersonate the uid "06f6ce97-e2c5-4ab8-7ba5-7654dd08d52b"
- apiGroups: ["authentication.k8s.io"]
resources: ["uids"]
verbs: ["impersonate"]
resourceNames: ["06f6ce97-e2c5-4ab8-7ba5-7654dd08d52b"]
ClusterRole and a ClusterRoleBinding,
not a Role and RoleBinding.
Kubernetes v1.22 [stable]
k8s.io/client-go and tools using it such as kubectl and kubelet are able to execute an
external command to receive user credentials.
This feature is intended for client side integrations with authentication protocols not natively
supported by k8s.io/client-go (LDAP, Kerberos, OAuth2, SAML, etc.). The plugin implements the
protocol specific logic, then returns opaque credentials to use. Almost all credential plugin
use cases require a server side component with support for the webhook token authenticator
to interpret the credential format produced by the client plugin.
In a hypothetical use case, an organization would run an external service that exchanges LDAP credentials for user specific, signed tokens. The service would also be capable of responding to webhook token authenticator requests to validate the tokens. Users would be required to install a credential plugin on their workstation.
To authenticate against the API:
kubectl command.TokenReview to the external service.Credential plugins are configured through kubectl config files as part of the user fields.
apiVersion: v1
kind: Config
users:
- name: my-user
user:
exec:
# Command to execute. Required.
command: "example-client-go-exec-plugin"
# API version to use when decoding the ExecCredentials resource. Required.
#
# The API version returned by the plugin MUST match the version listed here.
#
# To integrate with tools that support multiple versions (such as client.authentication.k8s.io/v1beta1),
# set an environment variable, pass an argument to the tool that indicates which version the exec plugin expects,
# or read the version from the ExecCredential object in the KUBERNETES_EXEC_INFO environment variable.
apiVersion: "client.authentication.k8s.io/v1"
# Environment variables to set when executing the plugin. Optional.
env:
- name: "FOO"
value: "bar"
# Arguments to pass when executing the plugin. Optional.
args:
- "arg1"
- "arg2"
# Text shown to the user when the executable doesn't seem to be present. Optional.
installHint: |
example-client-go-exec-plugin is required to authenticate
to the current cluster. It can be installed:
On macOS: brew install example-client-go-exec-plugin
On Ubuntu: apt-get install example-client-go-exec-plugin
On Fedora: dnf install example-client-go-exec-plugin
...
# Whether or not to provide cluster information, which could potentially contain
# very large CA data, to this exec plugin as a part of the KUBERNETES_EXEC_INFO
# environment variable.
provideClusterInfo: true
# The contract between the exec plugin and the standard input I/O stream. If the
# contract cannot be satisfied, this plugin will not be run and an error will be
# returned. Valid values are "Never" (this exec plugin never uses standard input),
# "IfAvailable" (this exec plugin wants to use standard input if it is available),
# or "Always" (this exec plugin requires standard input to function). Required.
interactiveMode: Never
clusters:
- name: my-cluster
cluster:
server: "https://172.17.4.100:6443"
certificate-authority: "/etc/kubernetes/ca.pem"
extensions:
- name: client.authentication.k8s.io/exec # reserved extension name for per cluster exec config
extension:
arbitrary: config
this: can be provided via the KUBERNETES_EXEC_INFO environment variable upon setting provideClusterInfo
you: ["can", "put", "anything", "here"]
contexts:
- name: my-cluster
context:
cluster: my-cluster
user: my-user
current-context: my-cluster
apiVersion: v1
kind: Config
users:
- name: my-user
user:
exec:
# Command to execute. Required.
command: "example-client-go-exec-plugin"
# API version to use when decoding the ExecCredentials resource. Required.
#
# The API version returned by the plugin MUST match the version listed here.
#
# To integrate with tools that support multiple versions (such as client.authentication.k8s.io/v1),
# set an environment variable, pass an argument to the tool that indicates which version the exec plugin expects,
# or read the version from the ExecCredential object in the KUBERNETES_EXEC_INFO environment variable.
apiVersion: "client.authentication.k8s.io/v1beta1"
# Environment variables to set when executing the plugin. Optional.
env:
- name: "FOO"
value: "bar"
# Arguments to pass when executing the plugin. Optional.
args:
- "arg1"
- "arg2"
# Text shown to the user when the executable doesn't seem to be present. Optional.
installHint: |
example-client-go-exec-plugin is required to authenticate
to the current cluster. It can be installed:
On macOS: brew install example-client-go-exec-plugin
On Ubuntu: apt-get install example-client-go-exec-plugin
On Fedora: dnf install example-client-go-exec-plugin
...
# Whether or not to provide cluster information, which could potentially contain
# very large CA data, to this exec plugin as a part of the KUBERNETES_EXEC_INFO
# environment variable.
provideClusterInfo: true
# The contract between the exec plugin and the standard input I/O stream. If the
# contract cannot be satisfied, this plugin will not be run and an error will be
# returned. Valid values are "Never" (this exec plugin never uses standard input),
# "IfAvailable" (this exec plugin wants to use standard input if it is available),
# or "Always" (this exec plugin requires standard input to function). Optional.
# Defaults to "IfAvailable".
interactiveMode: Never
clusters:
- name: my-cluster
cluster:
server: "https://172.17.4.100:6443"
certificate-authority: "/etc/kubernetes/ca.pem"
extensions:
- name: client.authentication.k8s.io/exec # reserved extension name for per cluster exec config
extension:
arbitrary: config
this: can be provided via the KUBERNETES_EXEC_INFO environment variable upon setting provideClusterInfo
you: ["can", "put", "anything", "here"]
contexts:
- name: my-cluster
context:
cluster: my-cluster
user: my-user
current-context: my-cluster
Relative command paths are interpreted as relative to the directory of the config file. If
KUBECONFIG is set to /home/jane/kubeconfig and the exec command is ./bin/example-client-go-exec-plugin,
the binary /home/jane/bin/example-client-go-exec-plugin is executed.
- name: my-user
user:
exec:
# Path relative to the directory of the kubeconfig
command: "./bin/example-client-go-exec-plugin"
apiVersion: "client.authentication.k8s.io/v1"
interactiveMode: Never
The executed command prints an ExecCredential object to stdout. k8s.io/client-go
authenticates against the Kubernetes API using the returned credentials in the status.
The executed command is passed an ExecCredential object as input via the KUBERNETES_EXEC_INFO
environment variable. This input contains helpful information like the expected API version
of the returned ExecCredential object and whether or not the plugin can use stdin to interact
with the user.
When run from an interactive session (i.e., a terminal), stdin can be exposed directly
to the plugin. Plugins should use the spec.interactive field of the input
ExecCredential object from the KUBERNETES_EXEC_INFO environment variable in order to
determine if stdin has been provided. A plugin's stdin requirements (i.e., whether
stdin is optional, strictly required, or never used in order for the plugin
to run successfully) is declared via the user.exec.interactiveMode field in the
kubeconfig (see table
below for valid values). The user.exec.interactiveMode field is optional in client.authentication.k8s.io/v1beta1
and required in client.authentication.k8s.io/v1.
interactiveMode Value |
Meaning |
|---|---|
Never |
This exec plugin never needs to use standard input, and therefore the exec plugin will be run regardless of whether standard input is available for user input. |
IfAvailable |
This exec plugin would like to use standard input if it is available, but can still operate if standard input is not available. Therefore, the exec plugin will be run regardless of whether stdin is available for user input. If standard input is available for user input, then it will be provided to this exec plugin. |
Always |
This exec plugin requires standard input in order to run, and therefore the exec plugin will only be run if standard input is available for user input. If standard input is not available for user input, then the exec plugin will not be run and an error will be returned by the exec plugin runner. |
To use bearer token credentials, the plugin returns a token in the status of the
ExecCredential
{
"apiVersion": "client.authentication.k8s.io/v1",
"kind": "ExecCredential",
"status": {
"token": "my-bearer-token"
}
}
{
"apiVersion": "client.authentication.k8s.io/v1beta1",
"kind": "ExecCredential",
"status": {
"token": "my-bearer-token"
}
}
Alternatively, a PEM-encoded client certificate and key can be returned to use TLS client auth.
If the plugin returns a different certificate and key on a subsequent call, k8s.io/client-go
will close existing connections with the server to force a new TLS handshake.
If specified, clientKeyData and clientCertificateData must both must be present.
clientCertificateData may contain additional intermediate certificates to send to the server.
{
"apiVersion": "client.authentication.k8s.io/v1",
"kind": "ExecCredential",
"status": {
"clientCertificateData": "-----BEGIN CERTIFICATE-----\n...\n-----END CERTIFICATE-----",
"clientKeyData": "-----BEGIN RSA PRIVATE KEY-----\n...\n-----END RSA PRIVATE KEY-----"
}
}
{
"apiVersion": "client.authentication.k8s.io/v1beta1",
"kind": "ExecCredential",
"status": {
"clientCertificateData": "-----BEGIN CERTIFICATE-----\n...\n-----END CERTIFICATE-----",
"clientKeyData": "-----BEGIN RSA PRIVATE KEY-----\n...\n-----END RSA PRIVATE KEY-----"
}
}
Optionally, the response can include the expiry of the credential formatted as a RFC 3339 timestamp.
Presence or absence of an expiry has the following impact:
{
"apiVersion": "client.authentication.k8s.io/v1",
"kind": "ExecCredential",
"status": {
"token": "my-bearer-token",
"expirationTimestamp": "2018-03-05T17:30:20-08:00"
}
}
{
"apiVersion": "client.authentication.k8s.io/v1beta1",
"kind": "ExecCredential",
"status": {
"token": "my-bearer-token",
"expirationTimestamp": "2018-03-05T17:30:20-08:00"
}
}
To enable the exec plugin to obtain cluster-specific information, set provideClusterInfo on the user.exec
field in the kubeconfig.
The plugin will then be supplied this cluster-specific information in the KUBERNETES_EXEC_INFO environment variable.
Information from this environment variable can be used to perform cluster-specific
credential acquisition logic.
The following ExecCredential manifest describes a cluster information sample.
{
"apiVersion": "client.authentication.k8s.io/v1",
"kind": "ExecCredential",
"spec": {
"cluster": {
"server": "https://172.17.4.100:6443",
"certificate-authority-data": "LS0t...",
"config": {
"arbitrary": "config",
"this": "can be provided via the KUBERNETES_EXEC_INFO environment variable upon setting provideClusterInfo",
"you": ["can", "put", "anything", "here"]
}
},
"interactive": true
}
}
{
"apiVersion": "client.authentication.k8s.io/v1beta1",
"kind": "ExecCredential",
"spec": {
"cluster": {
"server": "https://172.17.4.100:6443",
"certificate-authority-data": "LS0t...",
"config": {
"arbitrary": "config",
"this": "can be provided via the KUBERNETES_EXEC_INFO environment variable upon setting provideClusterInfo",
"you": ["can", "put", "anything", "here"]
}
},
"interactive": true
}
}
Kubernetes v1.18 [stable]
Bootstrap tokens are a simple bearer token that is meant to be used when
creating new clusters or joining new nodes to an existing cluster. It was built
to support kubeadm, but can be used in other contexts
for users that wish to start clusters without kubeadm. It is also built to
work, via RBAC policy, with the
Kubelet TLS Bootstrapping system.
Bootstrap Tokens are defined with a specific type
(bootstrap.kubernetes.io/token) of secrets that lives in the kube-system
namespace. These Secrets are then read by the Bootstrap Authenticator in the
API Server. Expired tokens are removed with the TokenCleaner controller in the
Controller Manager. The tokens are also used to create a signature for a
specific ConfigMap used in a "discovery" process through a BootstrapSigner
controller.
Bootstrap Tokens take the form of abcdef.0123456789abcdef. More formally,
they must match the regular expression [a-z0-9]{6}\.[a-z0-9]{16}.
The first part of the token is the "Token ID" and is considered public information. It is used when referring to a token without leaking the secret part used for authentication. The second part is the "Token Secret" and should only be shared with trusted parties.
The Bootstrap Token authenticator can be enabled using the following flag on the API server:
--enable-bootstrap-token-auth
When enabled, bootstrapping tokens can be used as bearer token credentials to authenticate requests against the API server.
Authorization: Bearer 07401b.f395accd246ae52d
Tokens authenticate as the username system:bootstrap:<token id> and are members
of the group system:bootstrappers. Additional groups may be specified in the
token's Secret.
Expired tokens can be deleted automatically by enabling the tokencleaner
controller on the controller manager.
--controllers=*,tokencleaner
Each valid token is backed by a secret in the kube-system namespace. You can
find the full design doc
here.
Here is what the secret looks like.
apiVersion: v1
kind: Secret
metadata:
# Name MUST be of form "bootstrap-token-<token id>"
name: bootstrap-token-07401b
namespace: kube-system
# Type MUST be 'bootstrap.kubernetes.io/token'
type: bootstrap.kubernetes.io/token
stringData:
# Human readable description. Optional.
description: "The default bootstrap token generated by 'kubeadm init'."
# Token ID and secret. Required.
token-id: 07401b
token-secret: f395accd246ae52d
# Expiration. Optional.
expiration: 2017-03-10T03:22:11Z
# Allowed usages.
usage-bootstrap-authentication: "true"
usage-bootstrap-signing: "true"
# Extra groups to authenticate the token as. Must start with "system:bootstrappers:"
auth-extra-groups: system:bootstrappers:worker,system:bootstrappers:ingress
The type of the secret must be bootstrap.kubernetes.io/token and the name must
be bootstrap-token-<token id>. It must also exist in the kube-system
namespace.
The usage-bootstrap-* members indicate what this secret is intended to be used
for. A value must be set to true to be enabled.
usage-bootstrap-authentication indicates that the token can be used to
authenticate to the API server as a bearer token.usage-bootstrap-signing indicates that the token may be used to sign the
cluster-info ConfigMap as described below.The expiration field controls the expiry of the token. Expired tokens are
rejected when used for authentication and ignored during ConfigMap signing.
The expiry value is encoded as an absolute UTC time using RFC3339. Enable the
tokencleaner controller to automatically delete expired tokens.
You can use the kubeadm tool to manage tokens on a running cluster. See the
kubeadm token docs for details.
In addition to authentication, the tokens can be used to sign a ConfigMap. This is used early in a cluster bootstrap process before the client trusts the API server. The signed ConfigMap can be authenticated by the shared token.
Enable ConfigMap signing by enabling the bootstrapsigner controller on the
Controller Manager.
--controllers=*,bootstrapsigner
The ConfigMap that is signed is cluster-info in the kube-public namespace.
The typical flow is that a client reads this ConfigMap while unauthenticated and
ignoring TLS errors. It then validates the payload of the ConfigMap by looking
at a signature embedded in the ConfigMap.
The ConfigMap may look like this:
apiVersion: v1
kind: ConfigMap
metadata:
name: cluster-info
namespace: kube-public
data:
jws-kubeconfig-07401b: eyJhbGciOiJIUzI1NiIsImtpZCI6IjA3NDAxYiJ9..tYEfbo6zDNo40MQE07aZcQX2m3EB2rO3NuXtxVMYm9U
kubeconfig: |
apiVersion: v1
clusters:
- cluster:
certificate-authority-data: <really long certificate data>
server: https://10.138.0.2:6443
name: ""
contexts: []
current-context: ""
kind: Config
preferences: {}
users: []
The kubeconfig member of the ConfigMap is a config file with only the cluster
information filled out. The key thing being communicated here is the
certificate-authority-data. This may be expanded in the future.
The signature is a JWS signature using the "detached" mode. To validate the
signature, the user should encode the kubeconfig payload according to JWS
rules (base64 encoded while discarding any trailing =). That encoded payload
is then used to form a whole JWS by inserting it between the 2 dots. You can
verify the JWS using the HS256 scheme (HMAC-SHA256) with the full token (e.g.
07401b.f395accd246ae52d) as the shared secret. Users must verify that HS256
is used.
Consult the kubeadm implementation details section for more information.
Kubernetes v1.19 [stable]
The Certificates API enables automation of X.509 credential provisioning by providing a programmatic interface for clients of the Kubernetes API to request and obtain X.509 certificates from a Certificate Authority (CA).
A CertificateSigningRequest (CSR) resource is used to request that a certificate be signed by a denoted signer, after which the request may be approved or denied before finally being signed.
The CertificateSigningRequest resource type allows a client to ask for an X.509 certificate
be issued, based on a signing request.
The CertificateSigningRequest object includes a PEM-encoded PKCS#10 signing request in
the spec.request field. The CertificateSigningRequest denotes the signer (the
recipient that the request is being made to) using the spec.signerName field.
Note that spec.signerName is a required key after API version certificates.k8s.io/v1.
In Kubernetes v1.22 and later, clients may optionally set the spec.expirationSeconds
field to request a particular lifetime for the issued certificate. The minimum valid
value for this field is 600, i.e. ten minutes.
Once created, a CertificateSigningRequest must be approved before it can be signed.
Depending on the signer selected, a CertificateSigningRequest may be automatically approved
by a controller.
Otherwise, a CertificateSigningRequest must be manually approved either via the REST API (or client-go)
or by running kubectl certificate approve. Likewise, a CertificateSigningRequest may also be denied,
which tells the configured signer that it must not sign the request.
For certificates that have been approved, the next step is signing. The relevant signing controller
first validates that the signing conditions are met and then creates a certificate.
The signing controller then updates the CertificateSigningRequest, storing the new certificate into
the status.certificate field of the existing CertificateSigningRequest object. The
status.certificate field is either empty or contains a X.509 certificate, encoded in PEM format.
The CertificateSigningRequest status.certificate field is empty until the signer does this.
Once the status.certificate field has been populated, the request has been completed and clients can now
fetch the signed certificate PEM data from the CertificateSigningRequest resource.
The signers can instead deny certificate signing if the approval conditions are not met.
In order to reduce the number of old CertificateSigningRequest resources left in a cluster, a garbage collection controller runs periodically. The garbage collection removes CertificateSigningRequests that have not changed state for some duration:
Custom signerNames can also be specified. All signers should provide information about how they work so that clients can predict what will happen to their CSRs. This includes:
spec.expirationSeconds field, etc
and the behavior when the signer-determined expiration is different from the CSR spec.expirationSeconds field.Commonly, the status.certificate field contains a single PEM-encoded X.509
certificate once the CSR is approved and the certificate is issued. Some
signers store multiple certificates into the status.certificate field. In
that case, the documentation for the signer should specify the meaning of
additional certificates; for example, this might be the certificate plus
intermediates to be presented during TLS handshakes.
The PKCS#10 signing request format does not have a standard mechanism to specify a
certificate expiration or lifetime. The expiration or lifetime therefore has to be set
through the spec.expirationSeconds field of the CSR object. The built-in signers
use the ClusterSigningDuration configuration option, which defaults to 1 year,
(the --cluster-signing-duration command-line flag of the kube-controller-manager)
as the default when no spec.expirationSeconds is specified. When spec.expirationSeconds
is specified, the minimum of spec.expirationSeconds and ClusterSigningDuration is
used.
spec.expirationSeconds field was added in Kubernetes v1.22. Earlier versions of Kubernetes do not honor this field.
Kubernetes API servers prior to v1.22 will silently drop this field when the object is created.
Kubernetes provides built-in signers that each have a well-known signerName:
kubernetes.io/kube-apiserver-client: signs certificates that will be honored as client certificates by the API server.
Never auto-approved by kube-controller-manager.
CertificateSubjectRestriction admission plugin is enabled by default to restrict system:masters,
but it is often not the only cluster-admin subject in a cluster.["client auth"]. Must not include key usages beyond ["digital signature", "key encipherment", "client auth"].--cluster-signing-duration option or, if specified, the spec.expirationSeconds field of the CSR object.kubernetes.io/kube-apiserver-client-kubelet: signs client certificates that will be honored as client certificates by the
API server.
May be auto-approved by kube-controller-manager.
["system:nodes"], common name starts with "system:node:".["key encipherment", "digital signature", "client auth"].--cluster-signing-duration option or, if specified, the spec.expirationSeconds field of the CSR object.kubernetes.io/kubelet-serving: signs serving certificates that are honored as a valid kubelet serving certificate
by the API server, but has no other guarantees.
Never auto-approved by kube-controller-manager.
["system:nodes"], common name starts with "system:node:".["key encipherment", "digital signature", "server auth"].--cluster-signing-duration option or, if specified, the spec.expirationSeconds field of the CSR object.kubernetes.io/legacy-unknown: has no guarantees for trust at all. Some third-party distributions of Kubernetes
may honor client certificates signed by it. The stable CertificateSigningRequest API (version certificates.k8s.io/v1 and later)
does not allow to set the signerName as kubernetes.io/legacy-unknown.
Never auto-approved by kube-controller-manager.
--cluster-signing-duration option or, if specified, the spec.expirationSeconds field of the CSR object.spec.expirationSeconds field was added in Kubernetes v1.22. Earlier versions of Kubernetes do not honor this field.
Kubernetes API servers prior to v1.22 will silently drop this field when the object is created.
Distribution of trust happens out of band for these signers. Any trust outside of those described above are strictly
coincidental. For instance, some distributions may honor kubernetes.io/legacy-unknown as client certificates for the
kube-apiserver, but this is not a standard.
None of these usages are related to ServiceAccount token secrets .data[ca.crt] in any way. That CA bundle is only
guaranteed to verify a connection to the API server using the default service (kubernetes.default.svc).
To allow creating a CertificateSigningRequest and retrieving any CertificateSigningRequest:
create, get, list, watch, group: certificates.k8s.io, resource: certificatesigningrequestsFor example:
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
name: csr-creator
rules:
- apiGroups:
- certificates.k8s.io
resources:
- certificatesigningrequests
verbs:
- create
- get
- list
- watch
To allow approving a CertificateSigningRequest:
get, list, watch, group: certificates.k8s.io, resource: certificatesigningrequestsupdate, group: certificates.k8s.io, resource: certificatesigningrequests/approvalapprove, group: certificates.k8s.io, resource: signers, resourceName: <signerNameDomain>/<signerNamePath> or <signerNameDomain>/*For example:
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
name: csr-approver
rules:
- apiGroups:
- certificates.k8s.io
resources:
- certificatesigningrequests
verbs:
- get
- list
- watch
- apiGroups:
- certificates.k8s.io
resources:
- certificatesigningrequests/approval
verbs:
- update
- apiGroups:
- certificates.k8s.io
resources:
- signers
resourceNames:
- example.com/my-signer-name # example.com/* can be used to authorize for all signers in the 'example.com' domain
verbs:
- approve
To allow signing a CertificateSigningRequest:
get, list, watch, group: certificates.k8s.io, resource: certificatesigningrequestsupdate, group: certificates.k8s.io, resource: certificatesigningrequests/statussign, group: certificates.k8s.io, resource: signers, resourceName: <signerNameDomain>/<signerNamePath> or <signerNameDomain>/*apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
name: csr-signer
rules:
- apiGroups:
- certificates.k8s.io
resources:
- certificatesigningrequests
verbs:
- get
- list
- watch
- apiGroups:
- certificates.k8s.io
resources:
- certificatesigningrequests/status
verbs:
- update
- apiGroups:
- certificates.k8s.io
resources:
- signers
resourceNames:
- example.com/my-signer-name # example.com/* can be used to authorize for all signers in the 'example.com' domain
verbs:
- sign
A few steps are required in order to get a normal user to be able to authenticate and invoke an API. First, this user must have a certificate issued by the Kubernetes cluster, and then present that certificate to the Kubernetes API.
The following scripts show how to generate PKI private key and CSR. It is important to set CN and O attribute of the CSR. CN is the name of the user and O is the group that this user will belong to. You can refer to RBAC for standard groups.
openssl genrsa -out myuser.key 2048
openssl req -new -key myuser.key -out myuser.csr
Create a CertificateSigningRequest and submit it to a Kubernetes Cluster via kubectl. Below is a script to generate the CertificateSigningRequest.
cat <<EOF | kubectl apply -f -
apiVersion: certificates.k8s.io/v1
kind: CertificateSigningRequest
metadata:
name: myuser
spec:
request: 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
signerName: kubernetes.io/kube-apiserver-client
expirationSeconds: 86400 # one day
usages:
- client auth
EOF
Some points to note:
usages has to be 'client auth'expirationSeconds could be made longer (i.e. 864000 for ten days) or shorter (i.e. 3600 for one hour)request is the base64 encoded value of the CSR file content.
You can get the content using this command: cat myuser.csr | base64 | tr -d "\n"Use kubectl to create a CSR and approve it.
Get the list of CSRs:
kubectl get csr
Approve the CSR:
kubectl certificate approve myuser
Retrieve the certificate from the CSR:
kubectl get csr/myuser -o yaml
The certificate value is in Base64-encoded format under status.certificate.
Export the issued certificate from the CertificateSigningRequest.
kubectl get csr myuser -o jsonpath='{.status.certificate}'| base64 -d > myuser.crt
With the certificate created it is time to define the Role and RoleBinding for this user to access Kubernetes cluster resources.
This is a sample command to create a Role for this new user:
kubectl create role developer --verb=create --verb=get --verb=list --verb=update --verb=delete --resource=pods
This is a sample command to create a RoleBinding for this new user:
kubectl create rolebinding developer-binding-myuser --role=developer --user=myuser
The last step is to add this user into the kubeconfig file.
First, you need to add new credentials:
kubectl config set-credentials myuser --client-key=myuser.key --client-certificate=myuser.crt --embed-certs=true
Then, you need to add the context:
kubectl config set-context myuser --cluster=kubernetes --user=myuser
To test it, change the context to myuser:
kubectl config use-context myuser
The kube-controller-manager ships with a built-in approver for certificates with
a signerName of kubernetes.io/kube-apiserver-client-kubelet that delegates various
permissions on CSRs for node credentials to authorization.
The kube-controller-manager POSTs SubjectAccessReview resources to the API server
in order to check authorization for certificate approval.
kubectlA Kubernetes administrator (with appropriate permissions) can manually approve
(or deny) CertificateSigningRequests by using the kubectl certificate approve and kubectl certificate deny commands.
To approve a CSR with kubectl:
kubectl certificate approve <certificate-signing-request-name>
Likewise, to deny a CSR:
kubectl certificate deny <certificate-signing-request-name>
Users of the REST API can approve CSRs by submitting an UPDATE request to the approval
subresource of the CSR to be approved. For example, you could write an
operator that watches for a particular
kind of CSR and then sends an UPDATE to approve them.
When you make an approval or rejection request, set either the Approved or Denied
status condition based on the state you determine:
For Approved CSRs:
apiVersion: certificates.k8s.io/v1
kind: CertificateSigningRequest
...
status:
conditions:
- lastUpdateTime: "2020-02-08T11:37:35Z"
lastTransitionTime: "2020-02-08T11:37:35Z"
message: Approved by my custom approver controller
reason: ApprovedByMyPolicy # You can set this to any string
type: Approved
For Denied CSRs:
apiVersion: certificates.k8s.io/v1
kind: CertificateSigningRequest
...
status:
conditions:
- lastUpdateTime: "2020-02-08T11:37:35Z"
lastTransitionTime: "2020-02-08T11:37:35Z"
message: Denied by my custom approver controller
reason: DeniedByMyPolicy # You can set this to any string
type: Denied
It's usual to set status.conditions.reason to a machine-friendly reason
code using TitleCase; this is a convention but you can set it to anything
you like. If you want to add a note for human consumption, use the
status.conditions.message field.
The Kubernetes control plane implements each of the Kubernetes signers, as part of the kube-controller-manager.
spec.expirationSeconds field was added in Kubernetes v1.22. Earlier versions of Kubernetes do not honor this field.
Kubernetes API servers prior to v1.22 will silently drop this field when the object is created.
Users of the REST API can sign CSRs by submitting an UPDATE request to the status
subresource of the CSR to be signed.
As part of this request, the status.certificate field should be set to contain the
signed certificate. This field contains one or more PEM-encoded certificates.
All PEM blocks must have the "CERTIFICATE" label, contain no headers, and the encoded data must be a BER-encoded ASN.1 Certificate structure as described in section 4 of RFC5280.
Example certificate content:
-----BEGIN CERTIFICATE-----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-----END CERTIFICATE-----
Non-PEM content may appear before or after the CERTIFICATE PEM blocks and is unvalidated, to allow for explanatory text as described in section 5.2 of RFC7468.
When encoded in JSON or YAML, this field is base-64 encoded. A CertificateSigningRequest containing the example certificate above would look like this:
apiVersion: certificates.k8s.io/v1
kind: CertificateSigningRequest
...
status:
certificate: "LS0tLS1CRUdJTiBDRVJUSUZJQ0FURS0tLS0tCk1JS..."
This page provides an overview of Admission Controllers.
An admission controller is a piece of code that intercepts requests to the
Kubernetes API server prior to persistence of the object, but after the request
is authenticated and authorized. The controllers consist of the
list below, are compiled into the
kube-apiserver binary, and may only be configured by the cluster
administrator. In that list, there are two special controllers:
MutatingAdmissionWebhook and ValidatingAdmissionWebhook. These execute the
mutating and validating (respectively)
admission control webhooks
which are configured in the API.
Admission controllers may be "validating", "mutating", or both. Mutating controllers may modify related objects to the requests they admit; validating controllers may not.
Admission controllers limit requests to create, delete, modify objects or connect to proxy. They do not limit requests to read objects.
The admission control process proceeds in two phases. In the first phase, mutating admission controllers are run. In the second phase, validating admission controllers are run. Note again that some of the controllers are both.
If any of the controllers in either phase reject the request, the entire request is rejected immediately and an error is returned to the end-user.
Finally, in addition to sometimes mutating the object in question, admission controllers may sometimes have side effects, that is, mutate related resources as part of request processing. Incrementing quota usage is the canonical example of why this is necessary. Any such side-effect needs a corresponding reclamation or reconciliation process, as a given admission controller does not know for sure that a given request will pass all of the other admission controllers.
Many advanced features in Kubernetes require an admission controller to be enabled in order to properly support the feature. As a result, a Kubernetes API server that is not properly configured with the right set of admission controllers is an incomplete server and will not support all the features you expect.
The Kubernetes API server flag enable-admission-plugins takes a comma-delimited list of admission control plugins to invoke prior to modifying objects in the cluster.
For example, the following command line enables the NamespaceLifecycle and the LimitRanger
admission control plugins:
kube-apiserver --enable-admission-plugins=NamespaceLifecycle,LimitRanger ...
The Kubernetes API server flag disable-admission-plugins takes a comma-delimited list of admission control plugins to be disabled, even if they are in the list of plugins enabled by default.
kube-apiserver --disable-admission-plugins=PodNodeSelector,AlwaysDeny ...
To see which admission plugins are enabled:
kube-apiserver -h | grep enable-admission-plugins
In the current version, the default ones are:
CertificateApproval, CertificateSigning, CertificateSubjectRestriction, DefaultIngressClass, DefaultStorageClass, DefaultTolerationSeconds, LimitRanger, MutatingAdmissionWebhook, NamespaceLifecycle, PersistentVolumeClaimResize, PodSecurity, Priority, ResourceQuota, RuntimeClass, ServiceAccount, StorageObjectInUseProtection, TaintNodesByCondition, ValidatingAdmissionWebhook
Kubernetes v1.13 [deprecated]
This admission controller allows all pods into the cluster. It is deprecated because its behavior is the same as if there were no admission controller at all.
Kubernetes v1.13 [deprecated]
Rejects all requests. AlwaysDeny is DEPRECATED as it has no real meaning.
This admission controller modifies every new Pod to force the image pull policy to Always. This is useful in a multitenant cluster so that users can be assured that their private images can only be used by those who have the credentials to pull them. Without this admission controller, once an image has been pulled to a node, any pod from any user can use it by knowing the image's name (assuming the Pod is scheduled onto the right node), without any authorization check against the image. When this admission controller is enabled, images are always pulled prior to starting containers, which means valid credentials are required.
This admission controller observes requests to 'approve' CertificateSigningRequest resources and performs additional
authorization checks to ensure the approving user has permission to approve certificate requests with the
spec.signerName requested on the CertificateSigningRequest resource.
See Certificate Signing Requests for more information on the permissions required to perform different actions on CertificateSigningRequest resources.
This admission controller observes updates to the status.certificate field of CertificateSigningRequest resources
and performs an additional authorization checks to ensure the signing user has permission to sign certificate
requests with the spec.signerName requested on the CertificateSigningRequest resource.
See Certificate Signing Requests for more information on the permissions required to perform different actions on CertificateSigningRequest resources.
This admission controller observes creation of CertificateSigningRequest resources that have a spec.signerName
of kubernetes.io/kube-apiserver-client. It rejects any request that specifies a 'group' (or 'organization attribute')
of system:masters.
This admission controller observes creation of Ingress objects that do not request any specific
ingress class and automatically adds a default ingress class to them. This way, users that do not
request any special ingress class do not need to care about them at all and they will get the
default one.
This admission controller does not do anything when no default ingress class is configured. When more than one ingress
class is marked as default, it rejects any creation of Ingress with an error and an administrator
must revisit their IngressClass objects and mark only one as default (with the annotation
"ingressclass.kubernetes.io/is-default-class"). This admission controller ignores any Ingress
updates; it acts only on creation.
See the ingress documentation for more about ingress classes and how to mark one as default.
This admission controller observes creation of PersistentVolumeClaim objects that do not request any specific storage class
and automatically adds a default storage class to them.
This way, users that do not request any special storage class do not need to care about them at all and they
will get the default one.
This admission controller does not do anything when no default storage class is configured. When more than one storage
class is marked as default, it rejects any creation of PersistentVolumeClaim with an error and an administrator
must revisit their StorageClass objects and mark only one as default.
This admission controller ignores any PersistentVolumeClaim updates; it acts only on creation.
See persistent volume documentation about persistent volume claims and storage classes and how to mark a storage class as default.
This admission controller sets the default forgiveness toleration for pods to tolerate
the taints notready:NoExecute and unreachable:NoExecute based on the k8s-apiserver input parameters
default-not-ready-toleration-seconds and default-unreachable-toleration-seconds if the pods don't already
have toleration for taints node.kubernetes.io/not-ready:NoExecute or
node.kubernetes.io/unreachable:NoExecute.
The default value for default-not-ready-toleration-seconds and default-unreachable-toleration-seconds is 5 minutes.
Kubernetes v1.13 [deprecated]
This admission controller will deny exec and attach commands to pods that run with escalated privileges that allow host access. This includes pods that run as privileged, have access to the host IPC namespace, and have access to the host PID namespace.
The DenyEscalatingExec admission plugin is deprecated.
Use of a policy-based admission plugin (like PodSecurityPolicy or a custom admission plugin) which can be targeted at specific users or Namespaces and also protects against creation of overly privileged Pods is recommended instead.
Kubernetes v1.13 [deprecated]
This admission controller will intercept all requests to exec a command in a pod if that pod has a privileged container.
This functionality has been merged into DenyEscalatingExec. The DenyExecOnPrivileged admission plugin is deprecated.
Use of a policy-based admission plugin (like PodSecurityPolicy or a custom admission plugin) which can be targeted at specific users or Namespaces and also protects against creation of overly privileged Pods is recommended instead.
This admission controller rejects all net-new usage of the Service field externalIPs. This
feature is very powerful (allows network traffic interception) and not well
controlled by policy. When enabled, users of the cluster may not create new
Services which use externalIPs and may not add new values to externalIPs on
existing Service objects. Existing uses of externalIPs are not affected,
and users may remove values from externalIPs on existing Service objects.
Most users do not need this feature at all, and cluster admins should consider disabling it. Clusters that do need to use this feature should consider using some custom policy to manage usage of it.
Kubernetes v1.13 [alpha]
This admission controller mitigates the problem where the API server gets flooded by event requests. The cluster admin can specify event rate limits by:
EventRateLimit admission controller;EventRateLimit configuration file from the file provided to the API
server's command line flag --admission-control-config-file:apiVersion: apiserver.config.k8s.io/v1
kind: AdmissionConfiguration
plugins:
- name: EventRateLimit
path: eventconfig.yaml
...
There are four types of limits that can be specified in the configuration:
Server: All event requests received by the API server share a single bucket.Namespace: Each namespace has a dedicated bucket.User: Each user is allocated a bucket.SourceAndObject: A bucket is assigned by each combination of source and
involved object of the event.Below is a sample eventconfig.yaml for such a configuration:
apiVersion: eventratelimit.admission.k8s.io/v1alpha1
kind: Configuration
limits:
- type: Namespace
qps: 50
burst: 100
cacheSize: 2000
- type: User
qps: 10
burst: 50
See the EventRateLimit Config API (v1alpha1) for more details.
This plug-in facilitates creation of dedicated nodes with extended resources. If operators want to create dedicated nodes with extended resources (like GPUs, FPGAs etc.), they are expected to taint the node with the extended resource name as the key. This admission controller, if enabled, automatically adds tolerations for such taints to pods requesting extended resources, so users don't have to manually add these tolerations.
The ImagePolicyWebhook admission controller allows a backend webhook to make admission decisions.
ImagePolicyWebhook uses a configuration file to set options for the behavior of the backend. This file may be json or yaml and has the following format:
imagePolicy:
kubeConfigFile: /path/to/kubeconfig/for/backend
# time in s to cache approval
allowTTL: 50
# time in s to cache denial
denyTTL: 50
# time in ms to wait between retries
retryBackoff: 500
# determines behavior if the webhook backend fails
defaultAllow: true
Reference the ImagePolicyWebhook configuration file from the file provided to the API server's command line flag --admission-control-config-file:
apiVersion: apiserver.config.k8s.io/v1
kind: AdmissionConfiguration
plugins:
- name: ImagePolicyWebhook
path: imagepolicyconfig.yaml
...
Alternatively, you can embed the configuration directly in the file:
apiVersion: apiserver.config.k8s.io/v1
kind: AdmissionConfiguration
plugins:
- name: ImagePolicyWebhook
configuration:
imagePolicy:
kubeConfigFile: <path-to-kubeconfig-file>
allowTTL: 50
denyTTL: 50
retryBackoff: 500
defaultAllow: true
The ImagePolicyWebhook config file must reference a kubeconfig formatted file which sets up the connection to the backend. It is required that the backend communicate over TLS.
The kubeconfig file's cluster field must point to the remote service, and the user field
must contain the returned authorizer.
# clusters refers to the remote service.
clusters:
- name: name-of-remote-imagepolicy-service
cluster:
certificate-authority: /path/to/ca.pem # CA for verifying the remote service.
server: https://images.example.com/policy # URL of remote service to query. Must use 'https'.
# users refers to the API server's webhook configuration.
users:
- name: name-of-api-server
user:
client-certificate: /path/to/cert.pem # cert for the webhook admission controller to use
client-key: /path/to/key.pem # key matching the cert
For additional HTTP configuration, refer to the kubeconfig documentation.
When faced with an admission decision, the API Server POSTs a JSON serialized
imagepolicy.k8s.io/v1alpha1 ImageReview object describing the action.
This object contains fields describing the containers being admitted, as well as
any pod annotations that match *.image-policy.k8s.io/*.
{{ note }}
The webhook API objects are subject to the same versioning compatibility rules
as other Kubernetes API objects. Implementers should be aware of looser compatibility
promises for alpha objects and check the apiVersion field of the request to
ensure correct deserialization.
Additionally, the API Server must enable the imagepolicy.k8s.io/v1alpha1 API extensions
group (--runtime-config=imagepolicy.k8s.io/v1alpha1=true).
{{ /note }}
An example request body:
{
"apiVersion":"imagepolicy.k8s.io/v1alpha1",
"kind":"ImageReview",
"spec":{
"containers":[
{
"image":"myrepo/myimage:v1"
},
{
"image":"myrepo/myimage@sha256:beb6bd6a68f114c1dc2ea4b28db81bdf91de202a9014972bec5e4d9171d90ed"
}
],
"annotations":{
"mycluster.image-policy.k8s.io/ticket-1234": "break-glass"
},
"namespace":"mynamespace"
}
}
The remote service is expected to fill the ImageReviewStatus field of the request and
respond to either allow or disallow access. The response body's spec field is ignored and
may be omitted. A permissive response would return:
{
"apiVersion": "imagepolicy.k8s.io/v1alpha1",
"kind": "ImageReview",
"status": {
"allowed": true
}
}
To disallow access, the service would return:
{
"apiVersion": "imagepolicy.k8s.io/v1alpha1",
"kind": "ImageReview",
"status": {
"allowed": false,
"reason": "image currently blacklisted"
}
}
For further documentation refer to the
imagepolicy.v1alpha1 API.
All annotations on a Pod that match *.image-policy.k8s.io/* are sent to the webhook.
Sending annotations allows users who are aware of the image policy backend to
send extra information to it, and for different backends implementations to
accept different information.
Examples of information you might put here are:
In any case, the annotations are provided by the user and are not validated by Kubernetes in any way.
This admission controller denies any pod that defines AntiAffinity topology key other than
kubernetes.io/hostname in requiredDuringSchedulingRequiredDuringExecution.
This admission controller will observe the incoming request and ensure that it does not violate
any of the constraints enumerated in the LimitRange object in a Namespace. If you are using
LimitRange objects in your Kubernetes deployment, you MUST use this admission controller to
enforce those constraints. LimitRanger can also be used to apply default resource requests to Pods
that don't specify any; currently, the default LimitRanger applies a 0.1 CPU requirement to all
Pods in the default namespace.
See the LimitRange API reference and the example of LimitRange for more details.
This admission controller calls any mutating webhooks which match the request. Matching webhooks are called in serial; each one may modify the object if it desires.
This admission controller (as implied by the name) only runs in the mutating phase.
If a webhook called by this has side effects (for example, decrementing quota) it must have a reconciliation system, as it is not guaranteed that subsequent webhooks or validating admission controllers will permit the request to finish.
If you disable the MutatingAdmissionWebhook, you must also disable the
MutatingWebhookConfiguration object in the admissionregistration.k8s.io/v1
group/version via the --runtime-config flag, both are on by default.
This admission controller examines all incoming requests on namespaced resources and checks if the referenced namespace does exist. It creates a namespace if it cannot be found. This admission controller is useful in deployments that do not want to restrict creation of a namespace prior to its usage.
This admission controller checks all requests on namespaced resources other than Namespace itself.
If the namespace referenced from a request doesn't exist, the request is rejected.
This admission controller enforces that a Namespace that is undergoing termination cannot have
new objects created in it, and ensures that requests in a non-existent Namespace are rejected.
This admission controller also prevents deletion of three system reserved namespaces default,
kube-system, kube-public.
A Namespace deletion kicks off a sequence of operations that remove all objects (pods, services,
etc.) in that namespace. In order to enforce integrity of that process, we strongly recommend
running this admission controller.
This admission controller limits the Node and Pod objects a kubelet can modify. In order to be limited by this admission controller,
kubelets must use credentials in the system:nodes group, with a username in the form system:node:<nodeName>.
Such kubelets will only be allowed to modify their own Node API object, and only modify Pod API objects that are bound to their node.
kubelets are not allowed to update or remove taints from their Node API object.
The NodeRestriction admission plugin prevents kubelets from deleting their Node API object,
and enforces kubelet modification of labels under the kubernetes.io/ or k8s.io/ prefixes as follows:
node-restriction.kubernetes.io/ prefix.
This label prefix is reserved for administrators to label their Node objects for workload isolation purposes,
and kubelets will not be allowed to modify labels with that prefix.kubernetes.io/hostnamekubernetes.io/archkubernetes.io/osbeta.kubernetes.io/instance-typenode.kubernetes.io/instance-typefailure-domain.beta.kubernetes.io/region (deprecated)failure-domain.beta.kubernetes.io/zone (deprecated)topology.kubernetes.io/regiontopology.kubernetes.io/zonekubelet.kubernetes.io/-prefixed labelsnode.kubernetes.io/-prefixed labelsUse of any other labels under the kubernetes.io or k8s.io prefixes by kubelets is reserved,
and may be disallowed or allowed by the NodeRestriction admission plugin in the future.
Future versions may add additional restrictions to ensure kubelets have the minimal set of permissions required to operate correctly.
This admission controller protects the access to the metadata.ownerReferences of an object
so that only users with "delete" permission to the object can change it.
This admission controller also protects the access to metadata.ownerReferences[x].blockOwnerDeletion
of an object, so that only users with "update" permission to the finalizers
subresource of the referenced owner can change it.
Kubernetes v1.24 [stable]
This admission controller implements additional validations for checking incoming
PersistentVolumeClaim resize requests.
Enabling the PersistentVolumeClaimResize admission controller is recommended.
This admission controller prevents resizing of all claims by default unless a claim's StorageClass
explicitly enables resizing by setting allowVolumeExpansion to true.
For example: all PersistentVolumeClaims created from the following StorageClass support volume expansion:
apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
name: gluster-vol-default
provisioner: kubernetes.io/glusterfs
parameters:
resturl: "http://192.168.10.100:8080"
restuser: ""
secretNamespace: ""
secretName: ""
allowVolumeExpansion: true
For more information about persistent volume claims, see PersistentVolumeClaims.
Kubernetes v1.13 [deprecated]
This admission controller automatically attaches region or zone labels to PersistentVolumes as defined by the cloud provider (for example, GCE or AWS). It helps ensure the Pods and the PersistentVolumes mounted are in the same region and/or zone. If the admission controller doesn't support automatic labelling your PersistentVolumes, you may need to add the labels manually to prevent pods from mounting volumes from a different zone. PersistentVolumeLabel is DEPRECATED and labeling persistent volumes has been taken over by the cloud-controller-manager. Starting from 1.11, this admission controller is disabled by default.
Kubernetes v1.5 [alpha]
This admission controller defaults and limits what node selectors may be used within a namespace by reading a namespace annotation and a global configuration.
PodNodeSelector uses a configuration file to set options for the behavior of the backend.
Note that the configuration file format will move to a versioned file in a future release.
This file may be json or yaml and has the following format:
podNodeSelectorPluginConfig:
clusterDefaultNodeSelector: name-of-node-selector
namespace1: name-of-node-selector
namespace2: name-of-node-selector
Reference the PodNodeSelector configuration file from the file provided to the API server's
command line flag --admission-control-config-file:
apiVersion: apiserver.config.k8s.io/v1
kind: AdmissionConfiguration
plugins:
- name: PodNodeSelector
path: podnodeselector.yaml
...
PodNodeSelector uses the annotation key scheduler.alpha.kubernetes.io/node-selector to assign
node selectors to namespaces.
apiVersion: v1
kind: Namespace
metadata:
annotations:
scheduler.alpha.kubernetes.io/node-selector: name-of-node-selector
name: namespace3
This admission controller has the following behavior:
Namespace has an annotation with a key scheduler.alpha.kubernetes.io/node-selector,
use its value as the node selector.clusterDefaultNodeSelector defined in the
PodNodeSelector plugin configuration file as the node selector.Kubernetes v1.23 [beta]
This is the replacement for the deprecated PodSecurityPolicy admission controller defined in the next section. This admission controller acts on creation and modification of the pod and determines if it should be admitted based on the requested security context and the Pod Security Standards.
See the Pod Security Admission documentation for more information.
Kubernetes v1.21 [deprecated]
This admission controller acts on creation and modification of the pod and determines if it should be admitted based on the requested security context and the available Pod Security Policies.
See also the PodSecurityPolicy documentation for more information.
Kubernetes v1.7 [alpha]
The PodTolerationRestriction admission controller verifies any conflict between tolerations of a pod and the tolerations of its namespace. It rejects the pod request if there is a conflict. It then merges the tolerations annotated on the namespace into the tolerations of the pod. The resulting tolerations are checked against a list of allowed tolerations annotated to the namespace. If the check succeeds, the pod request is admitted otherwise it is rejected.
If the namespace of the pod does not have any associated default tolerations or allowed tolerations annotated, the cluster-level default tolerations or cluster-level list of allowed tolerations are used instead if they are specified.
Tolerations to a namespace are assigned via the scheduler.alpha.kubernetes.io/defaultTolerations annotation key.
The list of allowed tolerations can be added via the scheduler.alpha.kubernetes.io/tolerationsWhitelist annotation key.
Example for namespace annotations:
apiVersion: v1
kind: Namespace
metadata:
name: apps-that-need-nodes-exclusively
annotations:
scheduler.alpha.kubernetes.io/defaultTolerations: '[{"operator": "Exists", "effect": "NoSchedule", "key": "dedicated-node"}]'
scheduler.alpha.kubernetes.io/tolerationsWhitelist: '[{"operator": "Exists", "effect": "NoSchedule", "key": "dedicated-node"}]'
The priority admission controller uses the priorityClassName field and populates the integer
value of the priority.
If the priority class is not found, the Pod is rejected.
This admission controller will observe the incoming request and ensure that it does not violate
any of the constraints enumerated in the ResourceQuota object in a Namespace. If you are
using ResourceQuota objects in your Kubernetes deployment, you MUST use this admission
controller to enforce quota constraints.
See the ResourceQuota API reference and the example of Resource Quota for more details.
Kubernetes v1.20 [stable]
If you define a RuntimeClass with Pod overhead
configured, this admission controller checks incoming Pods.
When enabled, this admission controller rejects any Pod create requests
that have the overhead already set.
For Pods that have a RuntimeClass configured and selected in their .spec,
this admission controller sets .spec.overhead in the Pod based on the value
defined in the corresponding RuntimeClass.
See also Pod Overhead for more information.
This admission controller will deny any Pod that attempts to set certain escalating SecurityContext fields, as shown in the Configure a Security Context for a Pod or Container task. If you don't use Pod Security admission, PodSecurityPolicies, nor any external enforcement mechanism, then you could use this admission controller to restrict the set of values a security context can take.
See Pod Security Standards for more context on restricting pod privileges.
This admission controller implements automation for
serviceAccounts.
We strongly recommend using this admission controller if you intend to make use of Kubernetes
ServiceAccount objects.
The StorageObjectInUseProtection plugin adds the kubernetes.io/pvc-protection or kubernetes.io/pv-protection
finalizers to newly created Persistent Volume Claims (PVCs) or Persistent Volumes (PV).
In case a user deletes a PVC or PV the PVC or PV is not removed until the finalizer is removed
from the PVC or PV by PVC or PV Protection Controller.
Refer to the
Storage Object in Use Protection
for more detailed information.
Kubernetes v1.17 [stable]
This admission controller taints newly created
Nodes as NotReady and NoSchedule. That tainting avoids a race condition that could cause Pods
to be scheduled on new Nodes before their taints were updated to accurately reflect their reported
conditions.
This admission controller calls any validating webhooks which match the request. Matching
webhooks are called in parallel; if any of them rejects the request, the request
fails. This admission controller only runs in the validation phase; the webhooks it calls may not
mutate the object, as opposed to the webhooks called by the MutatingAdmissionWebhook admission controller.
If a webhook called by this has side effects (for example, decrementing quota) it must have a reconciliation system, as it is not guaranteed that subsequent webhooks or other validating admission controllers will permit the request to finish.
If you disable the ValidatingAdmissionWebhook, you must also disable the
ValidatingWebhookConfiguration object in the admissionregistration.k8s.io/v1
group/version via the --runtime-config flag (both are on by default in
versions 1.9 and later).
Yes. The recommended admission controllers are enabled by default
(shown here),
so you do not need to explicitly specify them.
You can enable additional admission controllers beyond the default set using the
--enable-admission-plugins flag (order doesn't matter).
In addition to compiled-in admission plugins, admission plugins can be developed as extensions and run as webhooks configured at runtime. This page describes how to build, configure, use, and monitor admission webhooks.
Admission webhooks are HTTP callbacks that receive admission requests and do something with them. You can define two types of admission webhooks, validating admission webhook and mutating admission webhook. Mutating admission webhooks are invoked first, and can modify objects sent to the API server to enforce custom defaults. After all object modifications are complete, and after the incoming object is validated by the API server, validating admission webhooks are invoked and can reject requests to enforce custom policies.
Admission webhooks are essentially part of the cluster control-plane. You should write and deploy them with great caution. Please read the user guides for instructions if you intend to write/deploy production-grade admission webhooks. In the following, we describe how to quickly experiment with admission webhooks.
Ensure that MutatingAdmissionWebhook and ValidatingAdmissionWebhook admission controllers are enabled. Here is a recommended set of admission controllers to enable in general.
Ensure that the admissionregistration.k8s.io/v1 API is enabled.
Please refer to the implementation of the admission webhook server
that is validated in a Kubernetes e2e test. The webhook handles the
AdmissionReview request sent by the API servers, and sends back its decision
as an AdmissionReview object in the same version it received.
See the webhook request section for details on the data sent to webhooks.
See the webhook response section for the data expected from webhooks.
The example admission webhook server leaves the ClientAuth field
empty,
which defaults to NoClientCert. This means that the webhook server does not
authenticate the identity of the clients, supposedly API servers. If you need
mutual TLS or other ways to authenticate the clients, see
how to authenticate API servers.
The webhook server in the e2e test is deployed in the Kubernetes cluster, via the deployment API. The test also creates a service as the front-end of the webhook server. See code.
You may also deploy your webhooks outside of the cluster. You will need to update your webhook configurations accordingly.
You can dynamically configure what resources are subject to what admission webhooks via ValidatingWebhookConfiguration or MutatingWebhookConfiguration.
The following is an example ValidatingWebhookConfiguration, a mutating webhook configuration is similar.
See the webhook configuration section for details about each config field.
apiVersion: admissionregistration.k8s.io/v1
kind: ValidatingWebhookConfiguration
metadata:
name: "pod-policy.example.com"
webhooks:
- name: "pod-policy.example.com"
rules:
- apiGroups: [""]
apiVersions: ["v1"]
operations: ["CREATE"]
resources: ["pods"]
scope: "Namespaced"
clientConfig:
service:
namespace: "example-namespace"
name: "example-service"
caBundle: <CA_BUNDLE>
admissionReviewVersions: ["v1"]
sideEffects: None
timeoutSeconds: 5
<CA_BUNDLE> in the above example by a valid CA bundle
which is a PEM-encoded CA bundle for validating the webhook's server certificate.
The scope field specifies if only cluster-scoped resources ("Cluster") or namespace-scoped
resources ("Namespaced") will match this rule. "∗" means that there are no scope restrictions.
clientConfig.service, the server cert must be valid for
<svc_name>.<svc_namespace>.svc.
timeout and it is encouraged to use a short timeout for webhooks.
If the webhook call times out, the request is handled according to the webhook's
failure policy.
When an API server receives a request that matches one of the rules, the
API server sends an admissionReview request to webhook as specified in the
clientConfig.
After you create the webhook configuration, the system will take a few seconds to honor the new configuration.
If your admission webhooks require authentication, you can configure the API servers to use basic auth, bearer token, or a cert to authenticate itself to the webhooks. There are three steps to complete the configuration.
When starting the API server, specify the location of the admission control
configuration file via the --admission-control-config-file flag.
In the admission control configuration file, specify where the
MutatingAdmissionWebhook controller and ValidatingAdmissionWebhook controller
should read the credentials. The credentials are stored in kubeConfig files
(yes, the same schema that's used by kubectl), so the field name is
kubeConfigFile. Here is an example admission control configuration file:
apiVersion: apiserver.config.k8s.io/v1
kind: AdmissionConfiguration
plugins:
- name: ValidatingAdmissionWebhook
configuration:
apiVersion: apiserver.config.k8s.io/v1
kind: WebhookAdmissionConfiguration
kubeConfigFile: "<path-to-kubeconfig-file>"
- name: MutatingAdmissionWebhook
configuration:
apiVersion: apiserver.config.k8s.io/v1
kind: WebhookAdmissionConfiguration
kubeConfigFile: "<path-to-kubeconfig-file>"
# Deprecated in v1.17 in favor of apiserver.config.k8s.io/v1
apiVersion: apiserver.k8s.io/v1alpha1
kind: AdmissionConfiguration
plugins:
- name: ValidatingAdmissionWebhook
configuration:
# Deprecated in v1.17 in favor of apiserver.config.k8s.io/v1, kind=WebhookAdmissionConfiguration
apiVersion: apiserver.config.k8s.io/v1alpha1
kind: WebhookAdmission
kubeConfigFile: "<path-to-kubeconfig-file>"
- name: MutatingAdmissionWebhook
configuration:
# Deprecated in v1.17 in favor of apiserver.config.k8s.io/v1, kind=WebhookAdmissionConfiguration
apiVersion: apiserver.config.k8s.io/v1alpha1
kind: WebhookAdmission
kubeConfigFile: "<path-to-kubeconfig-file>"
For more information about AdmissionConfiguration, see the
AdmissionConfiguration (v1) reference.
See the webhook configuration section for details about each config field.
In the kubeConfig file, provide the credentials:
apiVersion: v1
kind: Config
users:
# name should be set to the DNS name of the service or the host (including port) of the URL the webhook is configured to speak to.
# If a non-443 port is used for services, it must be included in the name when configuring 1.16+ API servers.
#
# For a webhook configured to speak to a service on the default port (443), specify the DNS name of the service:
# - name: webhook1.ns1.svc
# user: ...
#
# For a webhook configured to speak to a service on non-default port (e.g. 8443), specify the DNS name and port of the service in 1.16+:
# - name: webhook1.ns1.svc:8443
# user: ...
# and optionally create a second stanza using only the DNS name of the service for compatibility with 1.15 API servers:
# - name: webhook1.ns1.svc
# user: ...
#
# For webhooks configured to speak to a URL, match the host (and port) specified in the webhook's URL. Examples:
# A webhook with `url: https://www.example.com`:
# - name: www.example.com
# user: ...
#
# A webhook with `url: https://www.example.com:443`:
# - name: www.example.com:443
# user: ...
#
# A webhook with `url: https://www.example.com:8443`:
# - name: www.example.com:8443
# user: ...
#
- name: 'webhook1.ns1.svc'
user:
client-certificate-data: "<pem encoded certificate>"
client-key-data: "<pem encoded key>"
# The `name` supports using * to wildcard-match prefixing segments.
- name: '*.webhook-company.org'
user:
password: "<password>"
username: "<name>"
# '*' is the default match.
- name: '*'
user:
token: "<token>"
Of course you need to set up the webhook server to handle these authentication requests.
Webhooks are sent as POST requests, with Content-Type: application/json,
with an AdmissionReview API object in the admission.k8s.io API group
serialized to JSON as the body.
Webhooks can specify what versions of AdmissionReview objects they accept
with the admissionReviewVersions field in their configuration:
apiVersion: admissionregistration.k8s.io/v1
kind: ValidatingWebhookConfiguration
webhooks:
- name: my-webhook.example.com
admissionReviewVersions: ["v1", "v1beta1"]
admissionReviewVersions is a required field when creating webhook configurations.
Webhooks are required to support at least one AdmissionReview
version understood by the current and previous API server.
API servers send the first AdmissionReview version in the admissionReviewVersions list they support.
If none of the versions in the list are supported by the API server, the configuration will not be allowed to be created.
If an API server encounters a webhook configuration that was previously created and does not support any of the AdmissionReview
versions the API server knows how to send, attempts to call to the webhook will fail and be subject to the failure policy.
This example shows the data contained in an AdmissionReview object
for a request to update the scale subresource of an apps/v1 Deployment:
apiVersion: admission.k8s.io/v1
kind: AdmissionReview
request:
# Random uid uniquely identifying this admission call
uid: 705ab4f5-6393-11e8-b7cc-42010a800002
# Fully-qualified group/version/kind of the incoming object
kind:
group: autoscaling
version: v1
kind: Scale
# Fully-qualified group/version/kind of the resource being modified
resource:
group: apps
version: v1
resource: deployments
# subresource, if the request is to a subresource
subResource: scale
# Fully-qualified group/version/kind of the incoming object in the original request to the API server.
# This only differs from `kind` if the webhook specified `matchPolicy: Equivalent` and the
# original request to the API server was converted to a version the webhook registered for.
requestKind:
group: autoscaling
version: v1
kind: Scale
# Fully-qualified group/version/kind of the resource being modified in the original request to the API server.
# This only differs from `resource` if the webhook specified `matchPolicy: Equivalent` and the
# original request to the API server was converted to a version the webhook registered for.
requestResource:
group: apps
version: v1
resource: deployments
# subresource, if the request is to a subresource
# This only differs from `subResource` if the webhook specified `matchPolicy: Equivalent` and the
# original request to the API server was converted to a version the webhook registered for.
requestSubResource: scale
# Name of the resource being modified
name: my-deployment
# Namespace of the resource being modified, if the resource is namespaced (or is a Namespace object)
namespace: my-namespace
# operation can be CREATE, UPDATE, DELETE, or CONNECT
operation: UPDATE
userInfo:
# Username of the authenticated user making the request to the API server
username: admin
# UID of the authenticated user making the request to the API server
uid: 014fbff9a07c
# Group memberships of the authenticated user making the request to the API server
groups:
- system:authenticated
- my-admin-group
# Arbitrary extra info associated with the user making the request to the API server.
# This is populated by the API server authentication layer and should be included
# if any SubjectAccessReview checks are performed by the webhook.
extra:
some-key:
- some-value1
- some-value2
# object is the new object being admitted.
# It is null for DELETE operations.
object:
apiVersion: autoscaling/v1
kind: Scale
# oldObject is the existing object.
# It is null for CREATE and CONNECT operations.
oldObject:
apiVersion: autoscaling/v1
kind: Scale
# options contains the options for the operation being admitted, like meta.k8s.io/v1 CreateOptions, UpdateOptions, or DeleteOptions.
# It is null for CONNECT operations.
options:
apiVersion: meta.k8s.io/v1
kind: UpdateOptions
# dryRun indicates the API request is running in dry run mode and will not be persisted.
# Webhooks with side effects should avoid actuating those side effects when dryRun is true.
# See http://k8s.io/docs/reference/using-api/api-concepts/#make-a-dry-run-request for more details.
dryRun: False
Webhooks respond with a 200 HTTP status code, Content-Type: application/json,
and a body containing an AdmissionReview object (in the same version they were sent),
with the response stanza populated, serialized to JSON.
At a minimum, the response stanza must contain the following fields:
uid, copied from the request.uid sent to the webhookallowed, either set to true or falseExample of a minimal response from a webhook to allow a request:
{
"apiVersion": "admission.k8s.io/v1",
"kind": "AdmissionReview",
"response": {
"uid": "<value from request.uid>",
"allowed": true
}
}
Example of a minimal response from a webhook to forbid a request:
{
"apiVersion": "admission.k8s.io/v1",
"kind": "AdmissionReview",
"response": {
"uid": "<value from request.uid>",
"allowed": false
}
}
When rejecting a request, the webhook can customize the http code and message returned to the user
using the status field. The specified status object is returned to the user.
See the API documentation
for details about the status type.
Example of a response to forbid a request, customizing the HTTP status code and message presented to the user:
{
"apiVersion": "admission.k8s.io/v1",
"kind": "AdmissionReview",
"response": {
"uid": "<value from request.uid>",
"allowed": false,
"status": {
"code": 403,
"message": "You cannot do this because it is Tuesday and your name starts with A"
}
}
}
When allowing a request, a mutating admission webhook may optionally modify the incoming object as well.
This is done using the patch and patchType fields in the response.
The only currently supported patchType is JSONPatch.
See JSON patch documentation for more details.
For patchType: JSONPatch, the patch field contains a base64-encoded array of JSON patch operations.
As an example, a single patch operation that would set spec.replicas would be
[{"op": "add", "path": "/spec/replicas", "value": 3}]
Base64-encoded, this would be W3sib3AiOiAiYWRkIiwgInBhdGgiOiAiL3NwZWMvcmVwbGljYXMiLCAidmFsdWUiOiAzfV0=
So a webhook response to add that label would be:
{
"apiVersion": "admission.k8s.io/v1",
"kind": "AdmissionReview",
"response": {
"uid": "<value from request.uid>",
"allowed": true,
"patchType": "JSONPatch",
"patch": "W3sib3AiOiAiYWRkIiwgInBhdGgiOiAiL3NwZWMvcmVwbGljYXMiLCAidmFsdWUiOiAzfV0="
}
}
Admission webhooks can optionally return warning messages that are returned to the requesting client
in HTTP Warning headers with a warning code of 299. Warnings can be sent with allowed or rejected admission responses.
If you're implementing a webhook that returns a warning:
{
"apiVersion": "admission.k8s.io/v1",
"kind": "AdmissionReview",
"response": {
"uid": "<value from request.uid>",
"allowed": true,
"warnings": [
"duplicate envvar entries specified with name MY_ENV",
"memory request less than 4MB specified for container mycontainer, which will not start successfully"
]
}
}
To register admission webhooks, create MutatingWebhookConfiguration or ValidatingWebhookConfiguration API objects.
The name of a MutatingWebhookConfiguration or a ValidatingWebhookConfiguration object must be a valid
DNS subdomain name.
Each configuration can contain one or more webhooks. If multiple webhooks are specified in a single configuration, each must be given a unique name. This is required in order to make resulting audit logs and metrics easier to match up to active configurations.
Each webhook defines the following things.
Each webhook must specify a list of rules used to determine if a request to the API server should be sent to the webhook. Each rule specifies one or more operations, apiGroups, apiVersions, and resources, and a resource scope:
operations lists one or more operations to match. Can be "CREATE", "UPDATE", "DELETE", "CONNECT",
or "*" to match all.
apiGroups lists one or more API groups to match. "" is the core API group. "*" matches all API groups.
apiVersions lists one or more API versions to match. "*" matches all API versions.
resources lists one or more resources to match.
"*" matches all resources, but not subresources."*/*" matches all resources and subresources."pods/*" matches all subresources of pods."*/status" matches all status subresources.scope specifies a scope to match. Valid values are "Cluster", "Namespaced", and "*".
Subresources match the scope of their parent resource. Default is "*".
"Cluster" means that only cluster-scoped resources will match this rule (Namespace API objects are cluster-scoped)."Namespaced" means that only namespaced resources will match this rule."*" means that there are no scope restrictions.If an incoming request matches one of the specified operations, groups, versions,
resources, and scope for any of a webhook's rules, the request is sent to the webhook.
Here are other examples of rules that could be used to specify which resources should be intercepted.
Match CREATE or UPDATE requests to apps/v1 and apps/v1beta1 deployments and replicasets:
apiVersion: admissionregistration.k8s.io/v1
kind: ValidatingWebhookConfiguration
...
webhooks:
- name: my-webhook.example.com
rules:
- operations: ["CREATE", "UPDATE"]
apiGroups: ["apps"]
apiVersions: ["v1", "v1beta1"]
resources: ["deployments", "replicasets"]
scope: "Namespaced"
...
Match create requests for all resources (but not subresources) in all API groups and versions:
apiVersion: admissionregistration.k8s.io/v1
kind: ValidatingWebhookConfiguration
webhooks:
- name: my-webhook.example.com
rules:
- operations: ["CREATE"]
apiGroups: ["*"]
apiVersions: ["*"]
resources: ["*"]
scope: "*"
Match update requests for all status subresources in all API groups and versions:
apiVersion: admissionregistration.k8s.io/v1
kind: ValidatingWebhookConfiguration
webhooks:
- name: my-webhook.example.com
rules:
- operations: ["UPDATE"]
apiGroups: ["*"]
apiVersions: ["*"]
resources: ["*/status"]
scope: "*"
Webhooks may optionally limit which requests are intercepted based on the labels of the
objects they would be sent, by specifying an objectSelector. If specified, the objectSelector
is evaluated against both the object and oldObject that would be sent to the webhook,
and is considered to match if either object matches the selector.
A null object (oldObject in the case of create, or newObject in the case of delete),
or an object that cannot have labels (like a DeploymentRollback or a PodProxyOptions object)
is not considered to match.
Use the object selector only if the webhook is opt-in, because end users may skip the admission webhook by setting the labels.
This example shows a mutating webhook that would match a CREATE of any resource with the label foo: bar:
apiVersion: admissionregistration.k8s.io/v1
kind: MutatingWebhookConfiguration
webhooks:
- name: my-webhook.example.com
objectSelector:
matchLabels:
foo: bar
rules:
- operations: ["CREATE"]
apiGroups: ["*"]
apiVersions: ["*"]
resources: ["*"]
scope: "*"
See labels concept for more examples of label selectors.
Webhooks may optionally limit which requests for namespaced resources are intercepted,
based on the labels of the containing namespace, by specifying a namespaceSelector.
The namespaceSelector decides whether to run the webhook on a request for a namespaced resource
(or a Namespace object), based on whether the namespace's labels match the selector.
If the object itself is a namespace, the matching is performed on object.metadata.labels.
If the object is a cluster scoped resource other than a Namespace, namespaceSelector has no effect.
This example shows a mutating webhook that matches a CREATE of any namespaced resource inside a namespace
that does not have a "runlevel" label of "0" or "1":
apiVersion: admissionregistration.k8s.io/v1
kind: MutatingWebhookConfiguration
webhooks:
- name: my-webhook.example.com
namespaceSelector:
matchExpressions:
- key: runlevel
operator: NotIn
values: ["0","1"]
rules:
- operations: ["CREATE"]
apiGroups: ["*"]
apiVersions: ["*"]
resources: ["*"]
scope: "Namespaced"
This example shows a validating webhook that matches a CREATE of any namespaced resource inside
a namespace that is associated with the "environment" of "prod" or "staging":
apiVersion: admissionregistration.k8s.io/v1
kind: ValidatingWebhookConfiguration
webhooks:
- name: my-webhook.example.com
namespaceSelector:
matchExpressions:
- key: environment
operator: In
values: ["prod","staging"]
rules:
- operations: ["CREATE"]
apiGroups: ["*"]
apiVersions: ["*"]
resources: ["*"]
scope: "Namespaced"
See labels concept for more examples of label selectors.
API servers can make objects available via multiple API groups or versions.
For example, if a webhook only specified a rule for some API groups/versions
(like apiGroups:["apps"], apiVersions:["v1","v1beta1"]),
and a request was made to modify the resource via another API group/version (like extensions/v1beta1),
the request would not be sent to the webhook.
The matchPolicy lets a webhook define how its rules are used to match incoming requests.
Allowed values are Exact or Equivalent.
Exact means a request should be intercepted only if it exactly matches a specified rule.Equivalent means a request should be intercepted if modifies a resource listed in rules,
even via another API group or version.In the example given above, the webhook that only registered for apps/v1 could use matchPolicy:
matchPolicy: Exact would mean the extensions/v1beta1 request would not be sent to the webhookmatchPolicy: Equivalent means the extensions/v1beta1 request would be sent to the webhook
(with the objects converted to a version the webhook had specified: apps/v1)Specifying Equivalent is recommended, and ensures that webhooks continue to intercept the
resources they expect when upgrades enable new versions of the resource in the API server.
When a resource stops being served by the API server, it is no longer considered equivalent to
other versions of that resource that are still served.
For example, extensions/v1beta1 deployments were first deprecated and then removed (in Kubernetes v1.16).
Since that removal, a webhook with a apiGroups:["extensions"], apiVersions:["v1beta1"], resources:["deployments"] rule
does not intercept deployments created via apps/v1 APIs. For that reason, webhooks should prefer registering
for stable versions of resources.
This example shows a validating webhook that intercepts modifications to deployments (no matter the API group or version),
and is always sent an apps/v1 Deployment object:
apiVersion: admissionregistration.k8s.io/v1
kind: ValidatingWebhookConfiguration
webhooks:
- name: my-webhook.example.com
matchPolicy: Equivalent
rules:
- operations: ["CREATE","UPDATE","DELETE"]
apiGroups: ["apps"]
apiVersions: ["v1"]
resources: ["deployments"]
scope: "Namespaced"
The matchPolicy for an admission webhooks defaults to Equivalent.
Once the API server has determined a request should be sent to a webhook,
it needs to know how to contact the webhook. This is specified in the clientConfig
stanza of the webhook configuration.
Webhooks can either be called via a URL or a service reference, and can optionally include a custom CA bundle to use to verify the TLS connection.
url gives the location of the webhook, in standard URL form
(scheme://host:port/path).
The host should not refer to a service running in the cluster; use
a service reference by specifying the service field instead.
The host might be resolved via external DNS in some API servers
(e.g., kube-apiserver cannot resolve in-cluster DNS as that would
be a layering violation). host may also be an IP address.
Please note that using localhost or 127.0.0.1 as a host is
risky unless you take great care to run this webhook on all hosts
which run an API server which might need to make calls to this
webhook. Such installations are likely to be non-portable or not readily
run in a new cluster.
The scheme must be "https"; the URL must begin with "https://".
Attempting to use a user or basic auth (for example user:password@) is not allowed.
Fragments (#...) and query parameters (?...) are also not allowed.
Here is an example of a mutating webhook configured to call a URL (and expects the TLS certificate to be verified using system trust roots, so does not specify a caBundle):
apiVersion: admissionregistration.k8s.io/v1
kind: MutatingWebhookConfiguration
webhooks:
- name: my-webhook.example.com
clientConfig:
url: "https://my-webhook.example.com:9443/my-webhook-path"
The service stanza inside clientConfig is a reference to the service for this webhook.
If the webhook is running within the cluster, then you should use service instead of url.
The service namespace and name are required. The port is optional and defaults to 443.
The path is optional and defaults to "/".
Here is an example of a mutating webhook configured to call a service on port "1234"
at the subpath "/my-path", and to verify the TLS connection against the ServerName
my-service-name.my-service-namespace.svc using a custom CA bundle:
apiVersion: admissionregistration.k8s.io/v1
kind: MutatingWebhookConfiguration
webhooks:
- name: my-webhook.example.com
clientConfig:
caBundle: <CA_BUNDLE>
service:
namespace: my-service-namespace
name: my-service-name
path: /my-path
port: 1234
<CA_BUNDLE> in the above example by a valid CA bundle
which is a PEM-encoded CA bundle for validating the webhook's server certificate.
Webhooks typically operate only on the content of the AdmissionReview sent to them.
Some webhooks, however, make out-of-band changes as part of processing admission requests.
Webhooks that make out-of-band changes ("side effects") must also have a reconciliation mechanism (like a controller) that periodically determines the actual state of the world, and adjusts the out-of-band data modified by the admission webhook to reflect reality. This is because a call to an admission webhook does not guarantee the admitted object will be persisted as is, or at all. Later webhooks can modify the content of the object, a conflict could be encountered while writing to storage, or the server could power off before persisting the object.
Additionally, webhooks with side effects must skip those side-effects when dryRun: true admission requests are handled.
A webhook must explicitly indicate that it will not have side-effects when run with dryRun,
or the dry-run request will not be sent to the webhook and the API request will fail instead.
Webhooks indicate whether they have side effects using the sideEffects field in the webhook configuration:
None: calling the webhook will have no side effects.NoneOnDryRun: calling the webhook will possibly have side effects, but if a request with
dryRun: true is sent to the webhook, the webhook will suppress the side effects (the webhook
is dryRun-aware).Here is an example of a validating webhook indicating it has no side effects on dryRun: true requests:
apiVersion: admissionregistration.k8s.io/v1
kind: ValidatingWebhookConfiguration
webhooks:
- name: my-webhook.example.com
sideEffects: NoneOnDryRun
Because webhooks add to API request latency, they should evaluate as quickly as possible.
timeoutSeconds allows configuring how long the API server should wait for a webhook to respond
before treating the call as a failure.
If the timeout expires before the webhook responds, the webhook call will be ignored or the API call will be rejected based on the failure policy.
The timeout value must be between 1 and 30 seconds.
Here is an example of a validating webhook with a custom timeout of 2 seconds:
apiVersion: admissionregistration.k8s.io/v1
kind: ValidatingWebhookConfiguration
webhooks:
- name: my-webhook.example.com
timeoutSeconds: 2
The timeout for an admission webhook defaults to 10 seconds.
A single ordering of mutating admissions plugins (including webhooks) does not work for all cases
(see https://issue.k8s.io/64333 as an example). A mutating webhook can add a new sub-structure
to the object (like adding a container to a pod), and other mutating plugins which have already
run may have opinions on those new structures (like setting an imagePullPolicy on all containers).
To allow mutating admission plugins to observe changes made by other plugins,
built-in mutating admission plugins are re-run if a mutating webhook modifies an object,
and mutating webhooks can specify a reinvocationPolicy to control whether they are reinvoked as well.
reinvocationPolicy may be set to Never or IfNeeded. It defaults to Never.
Never: the webhook must not be called more than once in a single admission evaluation.IfNeeded: the webhook may be called again as part of the admission evaluation if the object
being admitted is modified by other admission plugins after the initial webhook call.The important elements to note are:
Here is an example of a mutating webhook opting into being re-invoked if later admission plugins modify the object:
apiVersion: admissionregistration.k8s.io/v1
kind: MutatingWebhookConfiguration
webhooks:
- name: my-webhook.example.com
reinvocationPolicy: IfNeeded
Mutating webhooks must be idempotent, able to successfully process an object they have already admitted and potentially modified. This is true for all mutating admission webhooks, since any change they can make in an object could already exist in the user-provided object, but it is essential for webhooks that opt into reinvocation.
failurePolicy defines how unrecognized errors and timeout errors from the admission webhook
are handled. Allowed values are Ignore or Fail.
Ignore means that an error calling the webhook is ignored and the API request is allowed to continue.Fail means that an error calling the webhook causes the admission to fail and the API request to be rejected.Here is a mutating webhook configured to reject an API request if errors are encountered calling the admission webhook:
apiVersion: admissionregistration.k8s.io/v1
kind: MutatingWebhookConfiguration
webhooks:
- name: my-webhook.example.com
failurePolicy: Fail
The default failurePolicy for an admission webhooks is Fail.
The API server provides ways to monitor admission webhook behaviors. These monitoring mechanisms help cluster admins to answer questions like:
Which mutating webhook mutated the object in a API request?
What change did the mutating webhook applied to the object?
Which webhooks are frequently rejecting API requests? What's the reason for a rejection?
Sometimes it's useful to know which mutating webhook mutated the object in a API request, and what change did the webhook apply.
The Kubernetes API server performs auditing on each mutating webhook invocation. Each invocation generates an auditing annotation capturing if a request object is mutated by the invocation, and optionally generates an annotation capturing the applied patch from the webhook admission response. The annotations are set in the audit event for given request on given stage of its execution, which is then pre-processed according to a certain policy and written to a backend.
The audit level of a event determines which annotations get recorded:
At Metadata audit level or higher, an annotation with key
mutation.webhook.admission.k8s.io/round_{round idx}_index_{order idx} gets logged with JSON
payload indicating a webhook gets invoked for given request and whether it mutated the object or not.
For example, the following annotation gets recorded for a webhook being reinvoked. The webhook is ordered the third in the mutating webhook chain, and didn't mutated the request object during the invocation.
# the audit event recorded
{
"kind": "Event",
"apiVersion": "audit.k8s.io/v1",
"annotations": {
"mutation.webhook.admission.k8s.io/round_1_index_2": "{\"configuration\":\"my-mutating-webhook-configuration.example.com\",\"webhook\":\"my-webhook.example.com\",\"mutated\": false}"
# other annotations
...
}
# other fields
...
}
# the annotation value deserialized
{
"configuration": "my-mutating-webhook-configuration.example.com",
"webhook": "my-webhook.example.com",
"mutated": false
}
The following annotation gets recorded for a webhook being invoked in the first round. The webhook is ordered the first in the mutating webhook chain, and mutated the request object during the invocation.
# the audit event recorded
{
"kind": "Event",
"apiVersion": "audit.k8s.io/v1",
"annotations": {
"mutation.webhook.admission.k8s.io/round_0_index_0": "{\"configuration\":\"my-mutating-webhook-configuration.example.com\",\"webhook\":\"my-webhook-always-mutate.example.com\",\"mutated\": true}"
# other annotations
...
}
# other fields
...
}
# the annotation value deserialized
{
"configuration": "my-mutating-webhook-configuration.example.com",
"webhook": "my-webhook-always-mutate.example.com",
"mutated": true
}
At Request audit level or higher, an annotation with key
patch.webhook.admission.k8s.io/round_{round idx}_index_{order idx} gets logged with JSON payload indicating
a webhook gets invoked for given request and what patch gets applied to the request object.
For example, the following annotation gets recorded for a webhook being reinvoked. The webhook is ordered the fourth in the mutating webhook chain, and responded with a JSON patch which got applied to the request object.
# the audit event recorded
{
"kind": "Event",
"apiVersion": "audit.k8s.io/v1",
"annotations": {
"patch.webhook.admission.k8s.io/round_1_index_3": "{\"configuration\":\"my-other-mutating-webhook-configuration.example.com\",\"webhook\":\"my-webhook-always-mutate.example.com\",\"patch\":[{\"op\":\"add\",\"path\":\"/data/mutation-stage\",\"value\":\"yes\"}],\"patchType\":\"JSONPatch\"}"
# other annotations
...
}
# other fields
...
}
# the annotation value deserialized
{
"configuration": "my-other-mutating-webhook-configuration.example.com",
"webhook": "my-webhook-always-mutate.example.com",
"patchType": "JSONPatch",
"patch": [
{
"op": "add",
"path": "/data/mutation-stage",
"value": "yes"
}
]
}
The API server exposes Prometheus metrics from the /metrics endpoint, which can be used for monitoring and
diagnosing API server status. The following metrics record status related to admission webhooks.
Sometimes it's useful to know which admission webhooks are frequently rejecting API requests, and the reason for a rejection.
The API server exposes a Prometheus counter metric recording admission webhook rejections. The metrics are labelled to identify the causes of webhook rejection(s):
name: the name of the webhook that rejected a request.
operation: the operation type of the request, can be one of CREATE,
UPDATE, DELETE and CONNECT.
type: the admission webhook type, can be one of admit and validating.
error_type: identifies if an error occurred during the webhook invocation
that caused the rejection. Its value can be one of:
calling_webhook_error: unrecognized errors or timeout errors from the admission webhook happened and the
webhook's Failure policy is set to Fail.no_error: no error occurred. The webhook rejected the request with allowed: false in the admission
response. The metrics label rejection_code records the .status.code set in the admission response.apiserver_internal_error: an API server internal error happened.rejection_code: the HTTP status code set in the admission response when a
webhook rejected a request.
Example of the rejection count metrics:
# HELP apiserver_admission_webhook_rejection_count [ALPHA] Admission webhook rejection count, identified by name and broken out for each admission type (validating or admit) and operation. Additional labels specify an error type (calling_webhook_error or apiserver_internal_error if an error occurred; no_error otherwise) and optionally a non-zero rejection code if the webhook rejects the request with an HTTP status code (honored by the apiserver when the code is greater or equal to 400). Codes greater than 600 are truncated to 600, to keep the metrics cardinality bounded.
# TYPE apiserver_admission_webhook_rejection_count counter
apiserver_admission_webhook_rejection_count{error_type="calling_webhook_error",name="always-timeout-webhook.example.com",operation="CREATE",rejection_code="0",type="validating"} 1
apiserver_admission_webhook_rejection_count{error_type="calling_webhook_error",name="invalid-admission-response-webhook.example.com",operation="CREATE",rejection_code="0",type="validating"} 1
apiserver_admission_webhook_rejection_count{error_type="no_error",name="deny-unwanted-configmap-data.example.com",operation="CREATE",rejection_code="400",type="validating"} 13
An idempotent mutating admission webhook is able to successfully process an object it has already admitted and potentially modified. The admission can be applied multiple times without changing the result beyond the initial application.
For a CREATE pod request, set the field .spec.securityContext.runAsNonRoot of the
pod to true, to enforce security best practices.
For a CREATE pod request, if the field .spec.containers[].resources.limits
of a container is not set, set default resource limits.
For a CREATE pod request, inject a sidecar container with name foo-sidecar if no container
with the name foo-sidecar already exists.
In the cases above, the webhook can be safely reinvoked, or admit an object that already has the fields set.
For a CREATE pod request, inject a sidecar container with name foo-sidecar
suffixed with the current timestamp (e.g. foo-sidecar-19700101-000000).
For a CREATE/UPDATE pod request, reject if the pod has label "env" set,
otherwise add an "env": "prod" label to the pod.
For a CREATE pod request, blindly append a sidecar container named
foo-sidecar without looking to see if there is already a foo-sidecar
container in the pod.
In the first case above, reinvoking the webhook can result in the same sidecar being injected multiple times to a pod, each time with a different container name. Similarly the webhook can inject duplicated containers if the sidecar already exists in a user-provided pod.
In the second case above, reinvoking the webhook will result in the webhook failing on its own output.
In the third case above, reinvoking the webhook will result in duplicated containers in the pod spec, which makes the request invalid and rejected by the API server.
It is recommended that admission webhooks should always intercept all versions of an object by setting .webhooks[].matchPolicy
to Equivalent. It is also recommended that admission webhooks should prefer registering for stable versions of resources.
Failure to intercept all versions of an object can result in admission policies not being enforced for requests in certain
versions. See Matching requests: matchPolicy for examples.
It is recommended that admission webhooks should evaluate as quickly as possible (typically in milliseconds), since they add to API request latency. It is encouraged to use a small timeout for webhooks. See Timeouts for more detail.
It is recommended that admission webhooks should leverage some format of load-balancing, to provide high availability and performance benefits. If a webhook is running within the cluster, you can run multiple webhook backends behind a service to leverage the load-balancing that service supports.
Admission webhooks that need to guarantee they see the final state of the object in order to enforce policy should use a validating admission webhook, since objects can be modified after being seen by mutating webhooks.
For example, a mutating admission webhook is configured to inject a sidecar container with name
"foo-sidecar" on every CREATE pod request. If the sidecar must be present, a validating
admisson webhook should also be configured to intercept CREATE pod requests, and validate that a
container with name "foo-sidecar" with the expected configuration exists in the to-be-created
object.
A webhook running inside the cluster might cause deadlocks for its own deployment if it is configured to intercept resources required to start its own pods.
For example, a mutating admission webhook is configured to admit CREATE pod requests only if a certain label is set in the
pod (e.g. "env": "prod"). The webhook server runs in a deployment which doesn't set the "env" label.
When a node that runs the webhook server pods
becomes unhealthy, the webhook deployment will try to reschedule the pods to another node. However the requests will
get rejected by the existing webhook server since the "env" label is unset, and the migration cannot happen.
It is recommended to exclude the namespace where your webhook is running with a namespaceSelector.
It is recommended that admission webhooks should avoid side effects if possible, which means the webhooks operate only on the
content of the AdmissionReview sent to them, and do not make out-of-band changes. The .webhooks[].sideEffects field should
be set to None if a webhook doesn't have any side effect.
If side effects are required during the admission evaluation, they must be suppressed when processing an
AdmissionReview object with dryRun set to true, and the .webhooks[].sideEffects field should be
set to NoneOnDryRun. See Side effects for more detail.
The kube-system namespace contains objects created by the Kubernetes system,
e.g. service accounts for the control plane components, pods like kube-dns.
Accidentally mutating or rejecting requests in the kube-system namespace may
cause the control plane components to stop functioning or introduce unknown behavior.
If your admission webhooks don't intend to modify the behavior of the Kubernetes control
plane, exclude the kube-system namespace from being intercepted using a
namespaceSelector.
This is a Cluster Administrator guide to service accounts. You should be familiar with configuring Kubernetes service accounts.
Support for authorization and user accounts is planned but incomplete. Sometimes incomplete features are referred to in order to better describe service accounts.
Kubernetes distinguishes between the concept of a user account and a service account for a number of reasons:
Three separate components cooperate to implement the automation around service accounts:
ServiceAccount admission controllerServiceAccount controllerThe modification of pods is implemented via a plugin called an Admission Controller. It is part of the API server. It acts synchronously to modify pods as they are created or updated. When this plugin is active (and it is by default on most distributions), then it does the following when a pod is created or modified:
ServiceAccount set, it sets the ServiceAccount to default.ServiceAccount referenced by the pod exists, and otherwise rejects it.volume to the pod which contains a token for API access if neither the ServiceAccount automountServiceAccountToken nor the Pod's automountServiceAccountToken is set to false.volumeSource to each container of the pod mounted at /var/run/secrets/kubernetes.io/serviceaccount, if the previous step has created a volume for ServiceAccount token.imagePullSecrets, then imagePullSecrets of the ServiceAccount are added to the pod.Kubernetes v1.22 [stable]
The ServiceAccount admission controller will add the following projected volume instead of a Secret-based volume for the non-expiring service account token created by Token Controller.
- name: kube-api-access-<random-suffix>
projected:
defaultMode: 420 # 0644
sources:
- serviceAccountToken:
expirationSeconds: 3607
path: token
- configMap:
items:
- key: ca.crt
path: ca.crt
name: kube-root-ca.crt
- downwardAPI:
items:
- fieldRef:
apiVersion: v1
fieldPath: metadata.namespace
path: namespace
This projected volume consists of three sources:
RootCAConfigMap feature gate, which publishes a "kube-root-ca.crt" ConfigMap to every namespace. RootCAConfigMap feature gate is graduated to GA in 1.21 and default to true. (This flag will be removed from --feature-gate arg in 1.22)See more details about projected volumes.
TokenController runs as part of kube-controller-manager. It acts asynchronously. It:
You must pass a service account private key file to the token controller in
the kube-controller-manager using the --service-account-private-key-file
flag. The private key is used to sign generated service account tokens.
Similarly, you must pass the corresponding public key to the kube-apiserver
using the --service-account-key-file flag. The public key will be used to
verify the tokens during authentication.
A controller loop ensures a Secret with an API token exists for each
ServiceAccount. To create additional API tokens for a ServiceAccount, create a
Secret of type kubernetes.io/service-account-token with an annotation
referencing the ServiceAccount, and the controller will update it with a
generated token:
Below is a sample configuration for such a Secret:
apiVersion: v1
kind: Secret
metadata:
name: mysecretname
annotations:
kubernetes.io/service-account.name: myserviceaccount
type: kubernetes.io/service-account-token
kubectl create -f ./secret.yaml
kubectl describe secret mysecretname
kubectl delete secret mysecretname
A ServiceAccount controller manages the ServiceAccounts inside namespaces, and ensures a ServiceAccount named "default" exists in every active namespace.
Learn more about Kubernetes authorization, including details about creating policies using the supported authorization modules.
In Kubernetes, you must be authenticated (logged in) before your request can be authorized (granted permission to access). For information about authentication, see Controlling Access to the Kubernetes API.
Kubernetes expects attributes that are common to REST API requests. This means that Kubernetes authorization works with existing organization-wide or cloud-provider-wide access control systems which may handle other APIs besides the Kubernetes API.
Kubernetes authorizes API requests using the API server. It evaluates all of the request attributes against all policies and allows or denies the request. All parts of an API request must be allowed by some policy in order to proceed. This means that permissions are denied by default.
(Although Kubernetes uses the API server, access controls and policies that depend on specific fields of specific kinds of objects are handled by Admission Controllers.)
When multiple authorization modules are configured, each is checked in sequence. If any authorizer approves or denies a request, that decision is immediately returned and no other authorizer is consulted. If all modules have no opinion on the request, then the request is denied. A deny returns an HTTP status code 403.
Kubernetes reviews only the following API request attributes:
user string provided during authentication./api or /healthz.get, list, create, update, patch, watch, delete, and deletecollection are used for resource requests. To determine the request verb for a resource API endpoint, see Determine the request verb.get, post, put, and delete are used for non-resource requests.get, update, patch, and delete verbs, you must provide the resource name.Non-resource requests
Requests to endpoints other than /api/v1/... or /apis/<group>/<version>/...
are considered "non-resource requests", and use the lower-cased HTTP method of the request as the verb.
For example, a GET request to endpoints like /api or /healthz would use get as the verb.
Resource requests To determine the request verb for a resource API endpoint, review the HTTP verb used and whether or not the request acts on an individual resource or a collection of resources:
| HTTP verb | request verb |
|---|---|
| POST | create |
| GET, HEAD | get (for individual resources), list (for collections, including full object content), watch (for watching an individual resource or collection of resources) |
| PUT | update |
| PATCH | patch |
| DELETE | delete (for individual resources), deletecollection (for collections) |
get, list and watch verbs can all return the full details of a resource. In terms of the returned data they are equivalent. For example, list on secrets will still reveal the data attributes of any returned resources.
Kubernetes sometimes checks authorization for additional permissions using specialized verbs. For example:
use verb on podsecuritypolicies resources in the policy API group.bind and escalate verbs on roles and clusterroles resources in the rbac.authorization.k8s.io API group.impersonate verb on users, groups, and serviceaccounts in the core API group, and the userextras in the authentication.k8s.io API group.The Kubernetes API server may authorize a request using one of several authorization modes:
rbac.authorization.k8s.io API group to drive authorization decisions, allowing admins to dynamically configure permission policies through the Kubernetes API.--authorization-mode=RBAC.kubectl provides the auth can-i subcommand for quickly querying the API authorization layer.
The command uses the SelfSubjectAccessReview API to determine if the current user can perform
a given action, and works regardless of the authorization mode used.
kubectl auth can-i create deployments --namespace dev
The output is similar to this:
yes
kubectl auth can-i create deployments --namespace prod
The output is similar to this:
no
Administrators can combine this with user impersonation to determine what action other users can perform.
kubectl auth can-i list secrets --namespace dev --as dave
The output is similar to this:
no
Similarly, to check whether a ServiceAccount named dev-sa in Namespace dev
can list Pods in the Namespace target:
kubectl auth can-i list pods \
--namespace target \
--as system:serviceaccount:dev:dev-sa
The output is similar to this:
yes
SelfSubjectAccessReview is part of the authorization.k8s.io API group, which
exposes the API server authorization to external services. Other resources in
this group include:
SubjectAccessReview - Access review for any user, not only the current one. Useful for delegating authorization decisions to the API server. For example, the kubelet and extension API servers use this to determine user access to their own APIs.LocalSubjectAccessReview - Like SubjectAccessReview but restricted to a specific namespace.SelfSubjectRulesReview - A review which returns the set of actions a user can perform within a namespace. Useful for users to quickly summarize their own access, or for UIs to hide/show actions.These APIs can be queried by creating normal Kubernetes resources, where the response "status" field of the returned object is the result of the query.
kubectl create -f - -o yaml << EOF
apiVersion: authorization.k8s.io/v1
kind: SelfSubjectAccessReview
spec:
resourceAttributes:
group: apps
resource: deployments
verb: create
namespace: dev
EOF
The generated SelfSubjectAccessReview is:
apiVersion: authorization.k8s.io/v1
kind: SelfSubjectAccessReview
metadata:
creationTimestamp: null
spec:
resourceAttributes:
group: apps
resource: deployments
namespace: dev
verb: create
status:
allowed: true
denied: false
You must include a flag in your policy to indicate which authorization module your policies include:
The following flags can be used:
--authorization-mode=ABAC Attribute-Based Access Control (ABAC) mode allows you to configure policies using local files.--authorization-mode=RBAC Role-based access control (RBAC) mode allows you to create and store policies using the Kubernetes API.--authorization-mode=Webhook WebHook is an HTTP callback mode that allows you to manage authorization using a remote REST endpoint.--authorization-mode=Node Node authorization is a special-purpose authorization mode that specifically authorizes API requests made by kubelets.--authorization-mode=AlwaysDeny This flag blocks all requests. Use this flag only for testing.--authorization-mode=AlwaysAllow This flag allows all requests. Use this flag only if you do not require authorization for your API requests.You can choose more than one authorization module. Modules are checked in order so an earlier module has higher priority to allow or deny a request.
Users who can create/edit pods in a namespace, either directly or through a controller such as an operator, could escalate their privileges in that namespace.
Role-based access control (RBAC) is a method of regulating access to computer or network resources based on the roles of individual users within your organization.
RBAC authorization uses the rbac.authorization.k8s.io
API group to drive authorization
decisions, allowing you to dynamically configure policies through the Kubernetes API.
To enable RBAC, start the API server
with the --authorization-mode flag set to a comma-separated list that includes RBAC;
for example:
kube-apiserver --authorization-mode=Example,RBAC --other-options --more-options
The RBAC API declares four kinds of Kubernetes object: Role, ClusterRole,
RoleBinding and ClusterRoleBinding. You can
describe objects,
or amend them, using tools such as kubectl, just like any other Kubernetes object.
An RBAC Role or ClusterRole contains rules that represent a set of permissions. Permissions are purely additive (there are no "deny" rules).
A Role always sets permissions within a particular namespace; when you create a Role, you have to specify the namespace it belongs in.
ClusterRole, by contrast, is a non-namespaced resource. The resources have different names (Role and ClusterRole) because a Kubernetes object always has to be either namespaced or not namespaced; it can't be both.
ClusterRoles have several uses. You can use a ClusterRole to:
If you want to define a role within a namespace, use a Role; if you want to define a role cluster-wide, use a ClusterRole.
Here's an example Role in the "default" namespace that can be used to grant read access to pods:
apiVersion: rbac.authorization.k8s.io/v1
kind: Role
metadata:
namespace: default
name: pod-reader
rules:
- apiGroups: [""] # "" indicates the core API group
resources: ["pods"]
verbs: ["get", "watch", "list"]
A ClusterRole can be used to grant the same permissions as a Role. Because ClusterRoles are cluster-scoped, you can also use them to grant access to:
cluster-scoped resources (like nodes)
non-resource endpoints (like /healthz)
namespaced resources (like Pods), across all namespaces
For example: you can use a ClusterRole to allow a particular user to run
kubectl get pods --all-namespaces
Here is an example of a ClusterRole that can be used to grant read access to secrets in any particular namespace, or across all namespaces (depending on how it is bound):
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
# "namespace" omitted since ClusterRoles are not namespaced
name: secret-reader
rules:
- apiGroups: [""]
#
# at the HTTP level, the name of the resource for accessing Secret
# objects is "secrets"
resources: ["secrets"]
verbs: ["get", "watch", "list"]
The name of a Role or a ClusterRole object must be a valid path segment name.
A role binding grants the permissions defined in a role to a user or set of users. It holds a list of subjects (users, groups, or service accounts), and a reference to the role being granted. A RoleBinding grants permissions within a specific namespace whereas a ClusterRoleBinding grants that access cluster-wide.
A RoleBinding may reference any Role in the same namespace. Alternatively, a RoleBinding can reference a ClusterRole and bind that ClusterRole to the namespace of the RoleBinding. If you want to bind a ClusterRole to all the namespaces in your cluster, you use a ClusterRoleBinding.
The name of a RoleBinding or ClusterRoleBinding object must be a valid path segment name.
Here is an example of a RoleBinding that grants the "pod-reader" Role to the user "jane" within the "default" namespace. This allows "jane" to read pods in the "default" namespace.
apiVersion: rbac.authorization.k8s.io/v1
# This role binding allows "jane" to read pods in the "default" namespace.
# You need to already have a Role named "pod-reader" in that namespace.
kind: RoleBinding
metadata:
name: read-pods
namespace: default
subjects:
# You can specify more than one "subject"
- kind: User
name: jane # "name" is case sensitive
apiGroup: rbac.authorization.k8s.io
roleRef:
# "roleRef" specifies the binding to a Role / ClusterRole
kind: Role #this must be Role or ClusterRole
name: pod-reader # this must match the name of the Role or ClusterRole you wish to bind to
apiGroup: rbac.authorization.k8s.io
A RoleBinding can also reference a ClusterRole to grant the permissions defined in that ClusterRole to resources inside the RoleBinding's namespace. This kind of reference lets you define a set of common roles across your cluster, then reuse them within multiple namespaces.
For instance, even though the following RoleBinding refers to a ClusterRole, "dave" (the subject, case sensitive) will only be able to read Secrets in the "development" namespace, because the RoleBinding's namespace (in its metadata) is "development".
apiVersion: rbac.authorization.k8s.io/v1
# This role binding allows "dave" to read secrets in the "development" namespace.
# You need to already have a ClusterRole named "secret-reader".
kind: RoleBinding
metadata:
name: read-secrets
#
# The namespace of the RoleBinding determines where the permissions are granted.
# This only grants permissions within the "development" namespace.
namespace: development
subjects:
- kind: User
name: dave # Name is case sensitive
apiGroup: rbac.authorization.k8s.io
roleRef:
kind: ClusterRole
name: secret-reader
apiGroup: rbac.authorization.k8s.io
To grant permissions across a whole cluster, you can use a ClusterRoleBinding. The following ClusterRoleBinding allows any user in the group "manager" to read secrets in any namespace.
apiVersion: rbac.authorization.k8s.io/v1
# This cluster role binding allows anyone in the "manager" group to read secrets in any namespace.
kind: ClusterRoleBinding
metadata:
name: read-secrets-global
subjects:
- kind: Group
name: manager # Name is case sensitive
apiGroup: rbac.authorization.k8s.io
roleRef:
kind: ClusterRole
name: secret-reader
apiGroup: rbac.authorization.k8s.io
After you create a binding, you cannot change the Role or ClusterRole that it refers to.
If you try to change a binding's roleRef, you get a validation error. If you do want
to change the roleRef for a binding, you need to remove the binding object and create
a replacement.
There are two reasons for this restriction:
roleRef immutable allows granting someone update permission on an existing binding
object, so that they can manage the list of subjects, without being able to change
the role that is granted to those subjects.roleRef
ensures the full list of subjects in the binding is intended to be granted
the new role (as opposed to enabling or accidentally modifying only the roleRef
without verifying all of the existing subjects should be given the new role's
permissions).The kubectl auth reconcile command-line utility creates or updates a manifest file containing RBAC objects,
and handles deleting and recreating binding objects if required to change the role they refer to.
See command usage and examples for more information.
In the Kubernetes API, most resources are represented and accessed using a string representation of
their object name, such as pods for a Pod. RBAC refers to resources using exactly the same
name that appears in the URL for the relevant API endpoint.
Some Kubernetes APIs involve a
subresource, such as the logs for a Pod. A request for a Pod's logs looks like:
GET /api/v1/namespaces/{namespace}/pods/{name}/log
In this case, pods is the namespaced resource for Pod resources, and log is a
subresource of pods. To represent this in an RBAC role, use a slash (/) to
delimit the resource and subresource. To allow a subject to read pods and
also access the log subresource for each of those Pods, you write:
apiVersion: rbac.authorization.k8s.io/v1
kind: Role
metadata:
namespace: default
name: pod-and-pod-logs-reader
rules:
- apiGroups: [""]
resources: ["pods", "pods/log"]
verbs: ["get", "list"]
You can also refer to resources by name for certain requests through the resourceNames list.
When specified, requests can be restricted to individual instances of a resource.
Here is an example that restricts its subject to only get or update a
ConfigMap named my-configmap:
apiVersion: rbac.authorization.k8s.io/v1
kind: Role
metadata:
namespace: default
name: configmap-updater
rules:
- apiGroups: [""]
#
# at the HTTP level, the name of the resource for accessing ConfigMap
# objects is "configmaps"
resources: ["configmaps"]
resourceNames: ["my-configmap"]
verbs: ["update", "get"]
create or deletecollection requests by their resource name.
For create, this limitation is because the name of the new object may not be known at authorization time.
If you restrict list or watch by resourceName, clients must include a metadata.name field selector in their list or watch request that matches the specified resourceName in order to be authorized.
For example, kubectl get configmaps --field-selector=metadata.name=my-configmap
You can aggregate several ClusterRoles into one combined ClusterRole.
A controller, running as part of the cluster control plane, watches for ClusterRole
objects with an aggregationRule set. The aggregationRule defines a label
selector that the controller
uses to match other ClusterRole objects that should be combined into the rules
field of this one.
Here is an example aggregated ClusterRole:
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
name: monitoring
aggregationRule:
clusterRoleSelectors:
- matchLabels:
rbac.example.com/aggregate-to-monitoring: "true"
rules: [] # The control plane automatically fills in the rules
If you create a new ClusterRole that matches the label selector of an existing aggregated ClusterRole,
that change triggers adding the new rules into the aggregated ClusterRole.
Here is an example that adds rules to the "monitoring" ClusterRole, by creating another
ClusterRole labeled rbac.example.com/aggregate-to-monitoring: true.
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
name: monitoring-endpoints
labels:
rbac.example.com/aggregate-to-monitoring: "true"
# When you create the "monitoring-endpoints" ClusterRole,
# the rules below will be added to the "monitoring" ClusterRole.
rules:
- apiGroups: [""]
resources: ["services", "endpoints", "pods"]
verbs: ["get", "list", "watch"]
The default user-facing roles use ClusterRole aggregation. This lets you, as a cluster administrator, include rules for custom resources, such as those served by CustomResourceDefinitions or aggregated API servers, to extend the default roles.
For example: the following ClusterRoles let the "admin" and "edit" default roles manage the custom resource
named CronTab, whereas the "view" role can perform only read actions on CronTab resources.
You can assume that CronTab objects are named "crontabs" in URLs as seen by the API server.
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
name: aggregate-cron-tabs-edit
labels:
# Add these permissions to the "admin" and "edit" default roles.
rbac.authorization.k8s.io/aggregate-to-admin: "true"
rbac.authorization.k8s.io/aggregate-to-edit: "true"
rules:
- apiGroups: ["stable.example.com"]
resources: ["crontabs"]
verbs: ["get", "list", "watch", "create", "update", "patch", "delete"]
---
kind: ClusterRole
apiVersion: rbac.authorization.k8s.io/v1
metadata:
name: aggregate-cron-tabs-view
labels:
# Add these permissions to the "view" default role.
rbac.authorization.k8s.io/aggregate-to-view: "true"
rules:
- apiGroups: ["stable.example.com"]
resources: ["crontabs"]
verbs: ["get", "list", "watch"]
The following examples are excerpts from Role or ClusterRole objects, showing only
the rules section.
Allow reading "pods" resources in the core
API Group:
rules:
- apiGroups: [""]
#
# at the HTTP level, the name of the resource for accessing Pod
# objects is "pods"
resources: ["pods"]
verbs: ["get", "list", "watch"]
Allow reading/writing Deployments (at the HTTP level: objects with "deployments"
in the resource part of their URL) in the "apps" API groups:
rules:
- apiGroups: ["apps"]
#
# at the HTTP level, the name of the resource for accessing Deployment
# objects is "deployments"
resources: ["deployments"]
verbs: ["get", "list", "watch", "create", "update", "patch", "delete"]
Allow reading Pods in the core API group, as well as reading or writing Job
resources in the "batch" API group:
rules:
- apiGroups: [""]
#
# at the HTTP level, the name of the resource for accessing Pod
# objects is "pods"
resources: ["pods"]
verbs: ["get", "list", "watch"]
- apiGroups: ["batch"]
#
# at the HTTP level, the name of the resource for accessing Job
# objects is "jobs"
resources: ["jobs"]
verbs: ["get", "list", "watch", "create", "update", "patch", "delete"]
Allow reading a ConfigMap named "my-config" (must be bound with a RoleBinding to limit to a single ConfigMap in a single namespace):
rules:
- apiGroups: [""]
#
# at the HTTP level, the name of the resource for accessing ConfigMap
# objects is "configmaps"
resources: ["configmaps"]
resourceNames: ["my-config"]
verbs: ["get"]
Allow reading the resource "nodes" in the core group (because a
Node is cluster-scoped, this must be in a ClusterRole bound with a
ClusterRoleBinding to be effective):
rules:
- apiGroups: [""]
#
# at the HTTP level, the name of the resource for accessing Node
# objects is "nodes"
resources: ["nodes"]
verbs: ["get", "list", "watch"]
Allow GET and POST requests to the non-resource endpoint /healthz and
all subpaths (must be in a ClusterRole bound with a ClusterRoleBinding
to be effective):
rules:
- nonResourceURLs: ["/healthz", "/healthz/*"] # '*' in a nonResourceURL is a suffix glob match
verbs: ["get", "post"]
A RoleBinding or ClusterRoleBinding binds a role to subjects. Subjects can be groups, users or ServiceAccounts.
Kubernetes represents usernames as strings. These can be: plain names, such as "alice"; email-style names, like "bob@example.com"; or numeric user IDs represented as a string. It is up to you as a cluster administrator to configure the authentication modules so that authentication produces usernames in the format you want.
system: is reserved for Kubernetes system use, so you should ensure
that you don't have users or groups with names that start with system: by
accident.
Other than this special prefix, the RBAC authorization system does not require any format
for usernames.
In Kubernetes, Authenticator modules provide group information.
Groups, like users, are represented as strings, and that string has no format requirements,
other than that the prefix system: is reserved.
ServiceAccounts have names prefixed
with system:serviceaccount:, and belong to groups that have names prefixed with system:serviceaccounts:.
system:serviceaccount: (singular) is the prefix for service account usernames.system:serviceaccounts: (plural) is the prefix for service account groups.The following examples are RoleBinding excerpts that only
show the subjects section.
For a user named alice@example.com:
subjects:
- kind: User
name: "alice@example.com"
apiGroup: rbac.authorization.k8s.io
For a group named frontend-admins:
subjects:
- kind: Group
name: "frontend-admins"
apiGroup: rbac.authorization.k8s.io
For the default service account in the "kube-system" namespace:
subjects:
- kind: ServiceAccount
name: default
namespace: kube-system
For all service accounts in the "qa" namespace:
subjects:
- kind: Group
name: system:serviceaccounts:qa
apiGroup: rbac.authorization.k8s.io
For all service accounts in any namespace:
subjects:
- kind: Group
name: system:serviceaccounts
apiGroup: rbac.authorization.k8s.io
For all authenticated users:
subjects:
- kind: Group
name: system:authenticated
apiGroup: rbac.authorization.k8s.io
For all unauthenticated users:
subjects:
- kind: Group
name: system:unauthenticated
apiGroup: rbac.authorization.k8s.io
For all users:
subjects:
- kind: Group
name: system:authenticated
apiGroup: rbac.authorization.k8s.io
- kind: Group
name: system:unauthenticated
apiGroup: rbac.authorization.k8s.io
API servers create a set of default ClusterRole and ClusterRoleBinding objects.
Many of these are system: prefixed, which indicates that the resource is directly
managed by the cluster control plane.
All of the default ClusterRoles and ClusterRoleBindings are labeled with kubernetes.io/bootstrapping=rbac-defaults.
system: prefix.
Modifications to these resources can result in non-functional clusters.
At each start-up, the API server updates default cluster roles with any missing permissions, and updates default cluster role bindings with any missing subjects. This allows the cluster to repair accidental modifications, and helps to keep roles and role bindings up-to-date as permissions and subjects change in new Kubernetes releases.
To opt out of this reconciliation, set the rbac.authorization.kubernetes.io/autoupdate
annotation on a default cluster role or rolebinding to false.
Be aware that missing default permissions and subjects can result in non-functional clusters.
Auto-reconciliation is enabled by default if the RBAC authorizer is active.
Default role bindings authorize unauthenticated and authenticated users to read API information that is deemed safe to be publicly accessible (including CustomResourceDefinitions). To disable anonymous unauthenticated access, add --anonymous-auth=false to the API server configuration.
To view the configuration of these roles via kubectl run:
kubectl get clusterroles system:discovery -o yaml
| Default ClusterRole | Default ClusterRoleBinding | Description |
|---|---|---|
| system:basic-user | system:authenticated group | Allows a user read-only access to basic information about themselves. Prior to v1.14, this role was also bound to system:unauthenticated by default. |
| system:discovery | system:authenticated group | Allows read-only access to API discovery endpoints needed to discover and negotiate an API level. Prior to v1.14, this role was also bound to system:unauthenticated by default. |
| system:public-info-viewer | system:authenticated and system:unauthenticated groups | Allows read-only access to non-sensitive information about the cluster. Introduced in Kubernetes v1.14. |
Some of the default ClusterRoles are not system: prefixed. These are intended to be user-facing roles.
They include super-user roles (cluster-admin), roles intended to be granted cluster-wide
using ClusterRoleBindings, and roles intended to be granted within particular
namespaces using RoleBindings (admin, edit, view).
User-facing ClusterRoles use ClusterRole aggregation to allow admins to include
rules for custom resources on these ClusterRoles. To add rules to the admin, edit, or view roles, create
a ClusterRole with one or more of the following labels:
metadata:
labels:
rbac.authorization.k8s.io/aggregate-to-admin: "true"
rbac.authorization.k8s.io/aggregate-to-edit: "true"
rbac.authorization.k8s.io/aggregate-to-view: "true"
| Default ClusterRole | Default ClusterRoleBinding | Description |
|---|---|---|
| cluster-admin | system:masters group | Allows super-user access to perform any action on any resource. When used in a ClusterRoleBinding, it gives full control over every resource in the cluster and in all namespaces. When used in a RoleBinding, it gives full control over every resource in the role binding's namespace, including the namespace itself. |
| admin | None | Allows admin access, intended to be granted within a namespace using a RoleBinding.
If used in a RoleBinding, allows read/write access to most resources in a namespace, including the ability to create roles and role bindings within the namespace. This role does not allow write access to resource quota or to the namespace itself. This role also does not allow write access to Endpoints in clusters created using Kubernetes v1.22+. More information is available in the "Write Access for Endpoints" section. |
| edit | None | Allows read/write access to most objects in a namespace.
This role does not allow viewing or modifying roles or role bindings. However, this role allows accessing Secrets and running Pods as any ServiceAccount in the namespace, so it can be used to gain the API access levels of any ServiceAccount in the namespace. This role also does not allow write access to Endpoints in clusters created using Kubernetes v1.22+. More information is available in the "Write Access for Endpoints" section. |
| view | None | Allows read-only access to see most objects in a namespace.
It does not allow viewing roles or role bindings.
This role does not allow viewing Secrets, since reading the contents of Secrets enables access to ServiceAccount credentials in the namespace, which would allow API access as any ServiceAccount in the namespace (a form of privilege escalation). |
| Default ClusterRole | Default ClusterRoleBinding | Description |
|---|---|---|
| system:kube-scheduler | system:kube-scheduler user | Allows access to the resources required by the scheduler component. |
| system:volume-scheduler | system:kube-scheduler user | Allows access to the volume resources required by the kube-scheduler component. |
| system:kube-controller-manager | system:kube-controller-manager user | Allows access to the resources required by the controller manager component. The permissions required by individual controllers are detailed in the controller roles. |
| system:node | None | Allows access to resources required by the kubelet, including read access to all secrets, and write access to all pod status objects.
You should use the Node authorizer and NodeRestriction admission plugin instead of the system:node role, and allow granting API access to kubelets based on the Pods scheduled to run on them. The system:node role only exists for compatibility with Kubernetes clusters upgraded from versions prior to v1.8. |
| system:node-proxier | system:kube-proxy user | Allows access to the resources required by the kube-proxy component. |
| Default ClusterRole | Default ClusterRoleBinding | Description |
|---|---|---|
| system:auth-delegator | None | Allows delegated authentication and authorization checks. This is commonly used by add-on API servers for unified authentication and authorization. |
| system:heapster | None | Role for the Heapster component (deprecated). |
| system:kube-aggregator | None | Role for the kube-aggregator component. |
| system:kube-dns | kube-dns service account in the kube-system namespace | Role for the kube-dns component. |
| system:kubelet-api-admin | None | Allows full access to the kubelet API. |
| system:node-bootstrapper | None | Allows access to the resources required to perform kubelet TLS bootstrapping. |
| system:node-problem-detector | None | Role for the node-problem-detector component. |
| system:persistent-volume-provisioner | None | Allows access to the resources required by most dynamic volume provisioners. |
| system:monitoring | system:monitoring group | Allows read access to control-plane monitoring endpoints (i.e. kube-apiserver liveness and readiness endpoints (/healthz, /livez, /readyz), the individual health-check endpoints (/healthz/*, /livez/*, /readyz/*), and /metrics). Note that individual health check endpoints and the metric endpoint may expose sensitive information. |
The Kubernetes controller manager runs
controllers that are built in to the Kubernetes
control plane.
When invoked with --use-service-account-credentials, kube-controller-manager starts each controller
using a separate service account.
Corresponding roles exist for each built-in controller, prefixed with system:controller:.
If the controller manager is not started with --use-service-account-credentials, it runs all control loops
using its own credential, which must be granted all the relevant roles.
These roles include:
system:controller:attachdetach-controllersystem:controller:certificate-controllersystem:controller:clusterrole-aggregation-controllersystem:controller:cronjob-controllersystem:controller:daemon-set-controllersystem:controller:deployment-controllersystem:controller:disruption-controllersystem:controller:endpoint-controllersystem:controller:expand-controllersystem:controller:generic-garbage-collectorsystem:controller:horizontal-pod-autoscalersystem:controller:job-controllersystem:controller:namespace-controllersystem:controller:node-controllersystem:controller:persistent-volume-bindersystem:controller:pod-garbage-collectorsystem:controller:pv-protection-controllersystem:controller:pvc-protection-controllersystem:controller:replicaset-controllersystem:controller:replication-controllersystem:controller:resourcequota-controllersystem:controller:root-ca-cert-publishersystem:controller:route-controllersystem:controller:service-account-controllersystem:controller:service-controllersystem:controller:statefulset-controllersystem:controller:ttl-controllerThe RBAC API prevents users from escalating privileges by editing roles or role bindings. Because this is enforced at the API level, it applies even when the RBAC authorizer is not in use.
You can only create/update a role if at least one of the following things is true:
escalate verb on the roles or clusterroles resource in the rbac.authorization.k8s.io API group.For example, if user-1 does not have the ability to list Secrets cluster-wide, they cannot create a ClusterRole
containing that permission. To allow a user to create/update roles:
Role or ClusterRole by giving them permission to perform the escalate verb on roles or clusterroles resources in the rbac.authorization.k8s.io API groupYou can only create/update a role binding if you already have all the permissions contained in the referenced role
(at the same scope as the role binding) or if you have been authorized to perform the bind verb on the referenced role.
For example, if user-1 does not have the ability to list Secrets cluster-wide, they cannot create a ClusterRoleBinding
to a role that grants that permission. To allow a user to create/update role bindings:
bind verb on the particular Role (or ClusterRole).For example, this ClusterRole and RoleBinding would allow user-1 to grant other users the admin, edit, and view roles in the namespace user-1-namespace:
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
name: role-grantor
rules:
- apiGroups: ["rbac.authorization.k8s.io"]
resources: ["rolebindings"]
verbs: ["create"]
- apiGroups: ["rbac.authorization.k8s.io"]
resources: ["clusterroles"]
verbs: ["bind"]
# omit resourceNames to allow binding any ClusterRole
resourceNames: ["admin","edit","view"]
---
apiVersion: rbac.authorization.k8s.io/v1
kind: RoleBinding
metadata:
name: role-grantor-binding
namespace: user-1-namespace
roleRef:
apiGroup: rbac.authorization.k8s.io
kind: ClusterRole
name: role-grantor
subjects:
- apiGroup: rbac.authorization.k8s.io
kind: User
name: user-1
When bootstrapping the first roles and role bindings, it is necessary for the initial user to grant permissions they do not yet have. To bootstrap initial roles and role bindings:
--insecure-port), you can also make API calls via that port, which does not enforce authentication or authorization.kubectl create roleCreates a Role object defining permissions within a single namespace. Examples:
Create a Role named "pod-reader" that allows users to perform get, watch and list on pods:
kubectl create role pod-reader --verb=get --verb=list --verb=watch --resource=pods
Create a Role named "pod-reader" with resourceNames specified:
kubectl create role pod-reader --verb=get --resource=pods --resource-name=readablepod --resource-name=anotherpod
Create a Role named "foo" with apiGroups specified:
kubectl create role foo --verb=get,list,watch --resource=replicasets.apps
Create a Role named "foo" with subresource permissions:
kubectl create role foo --verb=get,list,watch --resource=pods,pods/status
Create a Role named "my-component-lease-holder" with permissions to get/update a resource with a specific name:
kubectl create role my-component-lease-holder --verb=get,list,watch,update --resource=lease --resource-name=my-component
kubectl create clusterroleCreates a ClusterRole. Examples:
Create a ClusterRole named "pod-reader" that allows user to perform get, watch and list on pods:
kubectl create clusterrole pod-reader --verb=get,list,watch --resource=pods
Create a ClusterRole named "pod-reader" with resourceNames specified:
kubectl create clusterrole pod-reader --verb=get --resource=pods --resource-name=readablepod --resource-name=anotherpod
Create a ClusterRole named "foo" with apiGroups specified:
kubectl create clusterrole foo --verb=get,list,watch --resource=replicasets.apps
Create a ClusterRole named "foo" with subresource permissions:
kubectl create clusterrole foo --verb=get,list,watch --resource=pods,pods/status
Create a ClusterRole named "foo" with nonResourceURL specified:
kubectl create clusterrole "foo" --verb=get --non-resource-url=/logs/*
Create a ClusterRole named "monitoring" with an aggregationRule specified:
kubectl create clusterrole monitoring --aggregation-rule="rbac.example.com/aggregate-to-monitoring=true"
kubectl create rolebindingGrants a Role or ClusterRole within a specific namespace. Examples:
Within the namespace "acme", grant the permissions in the "admin" ClusterRole to a user named "bob":
kubectl create rolebinding bob-admin-binding --clusterrole=admin --user=bob --namespace=acme
Within the namespace "acme", grant the permissions in the "view" ClusterRole to the service account in the namespace "acme" named "myapp":
kubectl create rolebinding myapp-view-binding --clusterrole=view --serviceaccount=acme:myapp --namespace=acme
Within the namespace "acme", grant the permissions in the "view" ClusterRole to a service account in the namespace "myappnamespace" named "myapp":
kubectl create rolebinding myappnamespace-myapp-view-binding --clusterrole=view --serviceaccount=myappnamespace:myapp --namespace=acme
kubectl create clusterrolebindingGrants a ClusterRole across the entire cluster (all namespaces). Examples:
Across the entire cluster, grant the permissions in the "cluster-admin" ClusterRole to a user named "root":
kubectl create clusterrolebinding root-cluster-admin-binding --clusterrole=cluster-admin --user=root
Across the entire cluster, grant the permissions in the "system:node-proxier" ClusterRole to a user named "system:kube-proxy":
kubectl create clusterrolebinding kube-proxy-binding --clusterrole=system:node-proxier --user=system:kube-proxy
Across the entire cluster, grant the permissions in the "view" ClusterRole to a service account named "myapp" in the namespace "acme":
kubectl create clusterrolebinding myapp-view-binding --clusterrole=view --serviceaccount=acme:myapp
kubectl auth reconcileCreates or updates rbac.authorization.k8s.io/v1 API objects from a manifest file.
Missing objects are created, and the containing namespace is created for namespaced objects, if required.
Existing roles are updated to include the permissions in the input objects,
and remove extra permissions if --remove-extra-permissions is specified.
Existing bindings are updated to include the subjects in the input objects,
and remove extra subjects if --remove-extra-subjects is specified.
Examples:
Test applying a manifest file of RBAC objects, displaying changes that would be made:
kubectl auth reconcile -f my-rbac-rules.yaml --dry-run=client
Apply a manifest file of RBAC objects, preserving any extra permissions (in roles) and any extra subjects (in bindings):
kubectl auth reconcile -f my-rbac-rules.yaml
Apply a manifest file of RBAC objects, removing any extra permissions (in roles) and any extra subjects (in bindings):
kubectl auth reconcile -f my-rbac-rules.yaml --remove-extra-subjects --remove-extra-permissions
Default RBAC policies grant scoped permissions to control-plane components, nodes,
and controllers, but grant no permissions to service accounts outside the kube-system namespace
(beyond discovery permissions given to all authenticated users).
This allows you to grant particular roles to particular ServiceAccounts as needed. Fine-grained role bindings provide greater security, but require more effort to administrate. Broader grants can give unnecessary (and potentially escalating) API access to ServiceAccounts, but are easier to administrate.
In order from most secure to least secure, the approaches are:
Grant a role to an application-specific service account (best practice)
This requires the application to specify a serviceAccountName in its pod spec,
and for the service account to be created (via the API, application manifest, kubectl create serviceaccount, etc.).
For example, grant read-only permission within "my-namespace" to the "my-sa" service account:
kubectl create rolebinding my-sa-view \
--clusterrole=view \
--serviceaccount=my-namespace:my-sa \
--namespace=my-namespace
Grant a role to the "default" service account in a namespace
If an application does not specify a serviceAccountName, it uses the "default" service account.
serviceAccountName.
For example, grant read-only permission within "my-namespace" to the "default" service account:
kubectl create rolebinding default-view \
--clusterrole=view \
--serviceaccount=my-namespace:default \
--namespace=my-namespace
Many add-ons run as the
"default" service account in the kube-system namespace.
To allow those add-ons to run with super-user access, grant cluster-admin
permissions to the "default" service account in the kube-system namespace.
kube-system namespace contains Secrets
that grant super-user access to your cluster's API.
kubectl create clusterrolebinding add-on-cluster-admin \
--clusterrole=cluster-admin \
--serviceaccount=kube-system:default
Grant a role to all service accounts in a namespace
If you want all applications in a namespace to have a role, no matter what service account they use, you can grant a role to the service account group for that namespace.
For example, grant read-only permission within "my-namespace" to all service accounts in that namespace:
kubectl create rolebinding serviceaccounts-view \
--clusterrole=view \
--group=system:serviceaccounts:my-namespace \
--namespace=my-namespace
Grant a limited role to all service accounts cluster-wide (discouraged)
If you don't want to manage permissions per-namespace, you can grant a cluster-wide role to all service accounts.
For example, grant read-only permission across all namespaces to all service accounts in the cluster:
kubectl create clusterrolebinding serviceaccounts-view \
--clusterrole=view \
--group=system:serviceaccounts
Grant super-user access to all service accounts cluster-wide (strongly discouraged)
If you don't care about partitioning permissions at all, you can grant super-user access to all service accounts.
kubectl create clusterrolebinding serviceaccounts-cluster-admin \
--clusterrole=cluster-admin \
--group=system:serviceaccounts
Kubernetes clusters created before Kubernetes v1.22 include write access to Endpoints in the aggregated "edit" and "admin" roles. As a mitigation for CVE-2021-25740, this access is not part of the aggregated roles in clusters that you create using Kubernetes v1.22 or later.
Existing clusters that have been upgraded to Kubernetes v1.22 will not be subject to this change. The CVE announcement includes guidance for restricting this access in existing clusters.
If you want new clusters to retain this level of access in the aggregated roles, you can create the following ClusterRole:
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
annotations:
kubernetes.io/description: |-
Add endpoints write permissions to the edit and admin roles. This was
removed by default in 1.22 because of CVE-2021-25740. See
https://issue.k8s.io/103675. This can allow writers to direct LoadBalancer
or Ingress implementations to expose backend IPs that would not otherwise
be accessible, and can circumvent network policies or security controls
intended to prevent/isolate access to those backends.
labels:
rbac.authorization.k8s.io/aggregate-to-edit: "true"
name: custom:aggregate-to-edit:endpoints # you can change this if you wish
rules:
- apiGroups: [""]
resources: ["endpoints"]
verbs: ["create", "delete", "deletecollection", "patch", "update"]
Clusters that originally ran older Kubernetes versions often used permissive ABAC policies, including granting full API access to all service accounts.
Default RBAC policies grant scoped permissions to control-plane components, nodes,
and controllers, but grant no permissions to service accounts outside the kube-system namespace
(beyond discovery permissions given to all authenticated users).
While far more secure, this can be disruptive to existing workloads expecting to automatically receive API permissions. Here are two approaches for managing this transition:
Run both the RBAC and ABAC authorizers, and specify a policy file that contains the legacy ABAC policy:
--authorization-mode=...,RBAC,ABAC --authorization-policy-file=mypolicy.json
To explain that first command line option in detail: if earlier authorizers, such as Node, deny a request, then the RBAC authorizer attempts to authorize the API request. If RBAC also denies that API request, the ABAC authorizer is then run. This means that any request allowed by either the RBAC or ABAC policies is allowed.
When the kube-apiserver is run with a log level of 5 or higher for the RBAC component
(--vmodule=rbac*=5 or --v=5), you can see RBAC denials in the API server log
(prefixed with RBAC).
You can use that information to determine which roles need to be granted to which users, groups, or service accounts.
Once you have granted roles to service accounts and workloads are running with no RBAC denial messages in the server logs, you can remove the ABAC authorizer.
You can replicate a permissive ABAC policy using RBAC role bindings.
The following policy allows ALL service accounts to act as cluster administrators. Any application running in a container receives service account credentials automatically, and could perform any action against the API, including viewing secrets and modifying permissions. This is not a recommended policy.
kubectl create clusterrolebinding permissive-binding \
--clusterrole=cluster-admin \
--user=admin \
--user=kubelet \
--group=system:serviceaccounts
After you have transitioned to use RBAC, you should adjust the access controls for your cluster to ensure that these meet your information security needs.
Attribute-based access control (ABAC) defines an access control paradigm whereby access rights are granted to users through the use of policies which combine attributes together.
To enable ABAC mode, specify --authorization-policy-file=SOME_FILENAME and --authorization-mode=ABAC on startup.
The file format is one JSON object per line. There should be no enclosing list or map, only one map per line.
Each line is a "policy object", where each such object is a map with the following properties:
apiVersion, type string; valid values are "abac.authorization.kubernetes.io/v1beta1". Allows versioning and conversion of the policy format.kind, type string: valid values are "Policy". Allows versioning and conversion of the policy format.spec property set to a map with the following properties:
user, type string; the user-string from --token-auth-file. If you specify user, it must match the username of the authenticated user.group, type string; if you specify group, it must match one of the groups of the authenticated user. system:authenticated matches all authenticated requests. system:unauthenticated matches all unauthenticated requests.apiGroup, type string; an API group.
apps, networking.k8s.io* matches all API groups.namespace, type string; a namespace.
kube-system* matches all resource requests.resource, type string; a resource type
pods, deployments* matches all resource requests.nonResourcePath, type string; non-resource request paths.
/version or /apis* matches all non-resource requests./foo/* matches all subpaths of /foo/.readonly, type boolean, when true, means that the Resource-matching policy only applies to get, list, and watch operations, Non-resource-matching policy only applies to get operation.An unset property is the same as a property set to the zero value for its type (e.g. empty string, 0, false). However, unset should be preferred for readability.
In the future, policies may be expressed in a JSON format, and managed via a REST interface.
A request has attributes which correspond to the properties of a policy object.
When a request is received, the attributes are determined. Unknown attributes are set to the zero value of its type (e.g. empty string, 0, false).
A property set to "*" will match any value of the corresponding attribute.
The tuple of attributes is checked for a match against every policy in the policy file. If at least one line matches the request attributes, then the request is authorized (but may fail later validation).
To permit any authenticated user to do something, write a policy with the
group property set to "system:authenticated".
To permit any unauthenticated user to do something, write a policy with the
group property set to "system:unauthenticated".
To permit a user to do anything, write a policy with the apiGroup, namespace,
resource, and nonResourcePath properties set to "*".
Kubectl uses the /api and /apis endpoints of api-server to discover
served resource types, and validates objects sent to the API by create/update
operations using schema information located at /openapi/v2.
When using ABAC authorization, those special resources have to be explicitly
exposed via the nonResourcePath property in a policy (see examples below):
/api, /api/*, /apis, and /apis/* for API version negotiation./version for retrieving the server version via kubectl version./swaggerapi/* for create/update operations.To inspect the HTTP calls involved in a specific kubectl operation you can turn up the verbosity:
kubectl --v=8 version
Alice can do anything to all resources:
{"apiVersion": "abac.authorization.kubernetes.io/v1beta1", "kind": "Policy", "spec": {"user": "alice", "namespace": "*", "resource": "*", "apiGroup": "*"}}
The Kubelet can read any pods:
{"apiVersion": "abac.authorization.kubernetes.io/v1beta1", "kind": "Policy", "spec": {"user": "kubelet", "namespace": "*", "resource": "pods", "readonly": true}}
The Kubelet can read and write events:
{"apiVersion": "abac.authorization.kubernetes.io/v1beta1", "kind": "Policy", "spec": {"user": "kubelet", "namespace": "*", "resource": "events"}}
Bob can just read pods in namespace "projectCaribou":
{"apiVersion": "abac.authorization.kubernetes.io/v1beta1", "kind": "Policy", "spec": {"user": "bob", "namespace": "projectCaribou", "resource": "pods", "readonly": true}}
Anyone can make read-only requests to all non-resource paths:
{"apiVersion": "abac.authorization.kubernetes.io/v1beta1", "kind": "Policy", "spec": {"group": "system:authenticated", "readonly": true, "nonResourcePath": "*"}}
{"apiVersion": "abac.authorization.kubernetes.io/v1beta1", "kind": "Policy", "spec": {"group": "system:unauthenticated", "readonly": true, "nonResourcePath": "*"}}
Every service account has a corresponding ABAC username, and that service account's user name is generated according to the naming convention:
system:serviceaccount:<namespace>:<serviceaccountname>
Creating a new namespace leads to the creation of a new service account in the following format:
system:serviceaccount:<namespace>:default
For example, if you wanted to grant the default service account (in the kube-system namespace) full
privilege to the API using ABAC, you would add this line to your policy file:
{"apiVersion":"abac.authorization.kubernetes.io/v1beta1","kind":"Policy","spec":{"user":"system:serviceaccount:kube-system:default","namespace":"*","resource":"*","apiGroup":"*"}}
The apiserver will need to be restarted to pickup the new policy lines.
Node authorization is a special-purpose authorization mode that specifically authorizes API requests made by kubelets.
The Node authorizer allows a kubelet to perform API operations. This includes:
Read operations:
Write operations:
NodeRestriction admission plugin to limit a kubelet to modify its own node)NodeRestriction admission plugin to limit a kubelet to modify pods bound to itself)Auth-related operations:
In future releases, the node authorizer may add or remove permissions to ensure kubelets have the minimal set of permissions required to operate correctly.
In order to be authorized by the Node authorizer, kubelets must use a credential that identifies them as
being in the system:nodes group, with a username of system:node:<nodeName>.
This group and user name format match the identity created for each kubelet as part of
kubelet TLS bootstrapping.
The value of <nodeName> must match precisely the name of the node as registered by the kubelet. By default, this is the host name as provided by hostname, or overridden via the kubelet option --hostname-override. However, when using the --cloud-provider kubelet option, the specific hostname may be determined by the cloud provider, ignoring the local hostname and the --hostname-override option.
For specifics about how the kubelet determines the hostname, see the kubelet options reference.
To enable the Node authorizer, start the apiserver with --authorization-mode=Node.
To limit the API objects kubelets are able to write, enable the NodeRestriction admission plugin by starting the apiserver with --enable-admission-plugins=...,NodeRestriction,...
system:nodes groupKubelets outside the system:nodes group would not be authorized by the Node authorization mode,
and would need to continue to be authorized via whatever mechanism currently authorizes them.
The node admission plugin would not restrict requests from these kubelets.
In some deployments, kubelets have credentials that place them in the system:nodes group,
but do not identify the particular node they are associated with,
because they do not have a username in the system:node:... format.
These kubelets would not be authorized by the Node authorization mode,
and would need to continue to be authorized via whatever mechanism currently authorizes them.
The NodeRestriction admission plugin would ignore requests from these kubelets,
since the default node identifier implementation would not consider that a node identity.
Upgraded pre-1.7 clusters using RBAC will continue functioning as-is because the system:nodes group binding will already exist.
If a cluster admin wishes to start using the Node authorizer and NodeRestriction admission plugin
to limit node access to the API, that can be done non-disruptively:
Node authorization mode (--authorization-mode=Node,RBAC) and the NodeRestriction admission pluginNode authorizer is not rejecting requests from kubelets (no persistent NODE DENY messages logged)system:node cluster role bindingIn 1.6, the system:node cluster role was automatically bound to the system:nodes group when using the RBAC Authorization mode.
In 1.7, the automatic binding of the system:nodes group to the system:node role is deprecated
because the node authorizer accomplishes the same purpose with the benefit of additional restrictions
on secret and configmap access. If the Node and RBAC authorization modes are both enabled,
the automatic binding of the system:nodes group to the system:node role is not created in 1.7.
In 1.8, the binding will not be created at all.
When using RBAC, the system:node cluster role will continue to be created,
for compatibility with deployment methods that bind other users or groups to that role.
The tables below enumerate the configuration parameters on PodSecurityPolicy objects, whether the field mutates and/or validates pods, and how the configuration values map to the Pod Security Standards.
For each applicable parameter, the allowed values for the Baseline and Restricted profiles are listed. Anything outside the allowed values for those profiles would fall under the Privileged profile. "No opinion" means all values are allowed under all Pod Security Standards.
For a step-by-step migration guide, see Migrate from PodSecurityPolicy to the Built-In PodSecurity Admission Controller.
The fields enumerated in this table are part of the PodSecurityPolicySpec, which is specified
under the .spec field path.
PodSecurityPolicySpec |
Type | Pod Security Standards Equivalent |
|---|---|---|
privileged |
Validating | Baseline & Restricted: false / undefined / nil |
defaultAddCapabilities |
Mutating & Validating | Requirements match allowedCapabilities below. |
allowedCapabilities |
Validating |
Baseline: subset of
Restricted: empty / undefined / nil OR a list containing only |
requiredDropCapabilities |
Mutating & Validating |
Baseline: no opinion Restricted: must include |
volumes |
Validating |
Baseline: anything except
Restricted: subset of
|
hostNetwork |
Validating | Baseline & Restricted: false / undefined / nil |
hostPorts |
Validating | Baseline & Restricted: undefined / nil / empty |
hostPID |
Validating | Baseline & Restricted: false / undefined / nil |
hostIPC |
Validating | Baseline & Restricted: false / undefined / nil |
seLinux |
Mutating & Validating |
Baseline & Restricted:
|
runAsUser |
Mutating & Validating |
Baseline: Anything Restricted: |
runAsGroup |
Mutating (MustRunAs) & Validating | No opinion |
supplementalGroups |
Mutating & Validating | No opinion |
fsGroup |
Mutating & Validating | No opinion |
readOnlyRootFilesystem |
Mutating & Validating | No opinion |
defaultAllowPrivilegeEscalation |
Mutating | No opinion (non-validating) |
allowPrivilegeEscalation |
Mutating & Validating |
Only mutating if set to Baseline: No opinion Restricted: |
allowedHostPaths |
Validating | No opinion (volumes takes precedence) |
allowedFlexVolumes |
Validating | No opinion (volumes takes precedence) |
allowedCSIDrivers |
Validating | No opinion (volumes takes precedence) |
allowedUnsafeSysctls |
Validating | Baseline & Restricted: undefined / nil / empty |
forbiddenSysctls |
Validating | No opinion |
allowedProcMountTypes(alpha feature) |
Validating | Baseline & Restricted: ["Default"] OR undefined / nil / empty |
runtimeClass .defaultRuntimeClassName |
Mutating | No opinion |
runtimeClass .allowedRuntimeClassNames |
Validating | No opinion |
The annotations enumerated in this
table can be specified under .metadata.annotations on the PodSecurityPolicy object.
PSP Annotation |
Type | Pod Security Standards Equivalent |
|---|---|---|
seccomp.security.alpha.kubernetes.io/defaultProfileName |
Mutating | No opinion |
seccomp.security.alpha.kubernetes.io/allowedProfileNames |
Validating |
Baseline: Restricted:
|
apparmor.security.beta.kubernetes.io/defaultProfileName |
Mutating | No opinion |
apparmor.security.beta.kubernetes.io/allowedProfileNames |
Validating |
Baseline: Restricted:
|
A WebHook is an HTTP callback: an HTTP POST that occurs when something happens; a simple event-notification via HTTP POST. A web application implementing WebHooks will POST a message to a URL when certain things happen.
When specified, mode Webhook causes Kubernetes to query an outside REST
service when determining user privileges.
Mode Webhook requires a file for HTTP configuration, specify by the
--authorization-webhook-config-file=SOME_FILENAME flag.
The configuration file uses the kubeconfig file format. Within the file "users" refers to the API Server webhook and "clusters" refers to the remote service.
A configuration example which uses HTTPS client auth:
# Kubernetes API version
apiVersion: v1
# kind of the API object
kind: Config
# clusters refers to the remote service.
clusters:
- name: name-of-remote-authz-service
cluster:
# CA for verifying the remote service.
certificate-authority: /path/to/ca.pem
# URL of remote service to query. Must use 'https'. May not include parameters.
server: https://authz.example.com/authorize
# users refers to the API Server's webhook configuration.
users:
- name: name-of-api-server
user:
client-certificate: /path/to/cert.pem # cert for the webhook plugin to use
client-key: /path/to/key.pem # key matching the cert
# kubeconfig files require a context. Provide one for the API Server.
current-context: webhook
contexts:
- context:
cluster: name-of-remote-authz-service
user: name-of-api-server
name: webhook
When faced with an authorization decision, the API Server POSTs a JSON-
serialized authorization.k8s.io/v1beta1 SubjectAccessReview object describing the
action. This object contains fields describing the user attempting to make the
request, and either details about the resource being accessed or requests
attributes.
Note that webhook API objects are subject to the same versioning compatibility rules
as other Kubernetes API objects. Implementers should be aware of looser
compatibility promises for beta objects and check the "apiVersion" field of the
request to ensure correct deserialization. Additionally, the API Server must
enable the authorization.k8s.io/v1beta1 API extensions group (--runtime-config=authorization.k8s.io/v1beta1=true).
An example request body:
{
"apiVersion": "authorization.k8s.io/v1beta1",
"kind": "SubjectAccessReview",
"spec": {
"resourceAttributes": {
"namespace": "kittensandponies",
"verb": "get",
"group": "unicorn.example.org",
"resource": "pods"
},
"user": "jane",
"group": [
"group1",
"group2"
]
}
}
The remote service is expected to fill the status field of
the request and respond to either allow or disallow access. The response body's
spec field is ignored and may be omitted. A permissive response would return:
{
"apiVersion": "authorization.k8s.io/v1beta1",
"kind": "SubjectAccessReview",
"status": {
"allowed": true
}
}
For disallowing access there are two methods.
The first method is preferred in most cases, and indicates the authorization webhook does not allow, or has "no opinion" about the request, but if other authorizers are configured, they are given a chance to allow the request. If there are no other authorizers, or none of them allow the request, the request is forbidden. The webhook would return:
{
"apiVersion": "authorization.k8s.io/v1beta1",
"kind": "SubjectAccessReview",
"status": {
"allowed": false,
"reason": "user does not have read access to the namespace"
}
}
The second method denies immediately, short-circuiting evaluation by other configured authorizers. This should only be used by webhooks that have detailed knowledge of the full authorizer configuration of the cluster. The webhook would return:
{
"apiVersion": "authorization.k8s.io/v1beta1",
"kind": "SubjectAccessReview",
"status": {
"allowed": false,
"denied": true,
"reason": "user does not have read access to the namespace"
}
}
Access to non-resource paths are sent as:
{
"apiVersion": "authorization.k8s.io/v1beta1",
"kind": "SubjectAccessReview",
"spec": {
"nonResourceAttributes": {
"path": "/debug",
"verb": "get"
},
"user": "jane",
"group": [
"group1",
"group2"
]
}
}
Non-resource paths include: /api, /apis, /metrics,
/logs, /debug, /healthz, /livez, /openapi/v2, /readyz, and
/version. Clients require access to /api, /api/*, /apis, /apis/*,
and /version to discover what resources and versions are present on the server.
Access to other non-resource paths can be disallowed without restricting access
to the REST api.
For further documentation refer to the authorization.v1beta1 API objects and webhook.go.
A kubelet's HTTPS endpoint exposes APIs which give access to data of varying sensitivity, and allow you to perform operations with varying levels of power on the node and within containers.
This document describes how to authenticate and authorize access to the kubelet's HTTPS endpoint.
By default, requests to the kubelet's HTTPS endpoint that are not rejected by other configured
authentication methods are treated as anonymous requests, and given a username of system:anonymous
and a group of system:unauthenticated.
To disable anonymous access and send 401 Unauthorized responses to unauthenticated requests:
--anonymous-auth=false flagTo enable X509 client certificate authentication to the kubelet's HTTPS endpoint:
--client-ca-file flag, providing a CA bundle to verify client certificates with--kubelet-client-certificate and --kubelet-client-key flagsTo enable API bearer tokens (including service account tokens) to be used to authenticate to the kubelet's HTTPS endpoint:
authentication.k8s.io/v1beta1 API group is enabled in the API server--authentication-token-webhook and --kubeconfig flagsTokenReview API on the configured API server to determine user information from bearer tokensAny request that is successfully authenticated (including an anonymous request) is then authorized. The default authorization mode is AlwaysAllow, which allows all requests.
There are many possible reasons to subdivide access to the kubelet API:
To subdivide access to the kubelet API, delegate authorization to the API server:
authorization.k8s.io/v1beta1 API group is enabled in the API server--authorization-mode=Webhook and the --kubeconfig flagsSubjectAccessReview API on the configured API server to determine whether each request is authorizedThe kubelet authorizes API requests using the same request attributes approach as the apiserver.
The verb is determined from the incoming request's HTTP verb:
| HTTP verb | request verb |
|---|---|
| POST | create |
| GET, HEAD | get |
| PUT | update |
| PATCH | patch |
| DELETE | delete |
The resource and subresource is determined from the incoming request's path:
| Kubelet API | resource | subresource |
|---|---|---|
| /stats/* | nodes | stats |
| /metrics/* | nodes | metrics |
| /logs/* | nodes | log |
| /spec/* | nodes | spec |
| all others | nodes | proxy |
The namespace and API group attributes are always an empty string, and
the resource name is always the name of the kubelet's Node API object.
When running in this mode, ensure the user identified by the --kubelet-client-certificate and --kubelet-client-key
flags passed to the apiserver is authorized for the following attributes:
In a Kubernetes cluster, the components on the worker nodes - kubelet and kube-proxy - need to communicate with Kubernetes control plane components, specifically kube-apiserver. In order to ensure that communication is kept private, not interfered with, and ensure that each component of the cluster is talking to another trusted component, we strongly recommend using client TLS certificates on nodes.
The normal process of bootstrapping these components, especially worker nodes that need certificates so they can communicate safely with kube-apiserver, can be a challenging process as it is often outside of the scope of Kubernetes and requires significant additional work. This in turn, can make it challenging to initialize or scale a cluster.
In order to simplify the process, beginning in version 1.4, Kubernetes introduced a certificate request and signing API. The proposal can be found here.
This document describes the process of node initialization, how to set up TLS client certificate bootstrapping for kubelets, and how it works.
When a worker node starts up, the kubelet does the following:
kubeconfig filekubeconfig fileAssuming that the kube-apiserver successfully validates the kubelet's credentials, it will treat the kubelet as a valid node, and begin to assign pods to it.
Note that the above process depends upon:
kubeconfigAll of the following are responsibilities of whoever sets up and manages the cluster:
The TLS Bootstrapping described in this document is intended to simplify, and partially or even completely automate, steps 3 onwards, as these are the most common when initializing or scaling a cluster.
In the bootstrap initialization process, the following occurs:
kubeconfig filebootstrap-kubeconfig filekubernetes.io/kube-apiserver-client-kubeletkubectlkubeconfig with the key and signed certificateThe rest of this document describes the necessary steps to configure TLS Bootstrapping, and its limitations.
To configure for TLS bootstrapping and optional automatic approval, you must configure options on the following components:
ClusterRoleBinding and potentially ClusterRoleIn addition, you need your Kubernetes Certificate Authority (CA).
As without bootstrapping, you will need a Certificate Authority (CA) key and certificate. As without bootstrapping, these will be used to sign the kubelet certificate. As before, it is your responsibility to distribute them to control plane nodes.
For the purposes of this document, we will assume these have been distributed to control plane nodes at /var/lib/kubernetes/ca.pem (certificate) and /var/lib/kubernetes/ca-key.pem (key).
We will refer to these as "Kubernetes CA certificate and key".
All Kubernetes components that use these certificates - kubelet, kube-apiserver, kube-controller-manager - assume the key and certificate to be PEM-encoded.
The kube-apiserver has several requirements to enable TLS bootstrapping:
system:bootstrappers groupThis is normal for all client certificate authentication.
If not already set, add the --client-ca-file=FILENAME flag to the kube-apiserver command to enable
client certificate authentication, referencing a certificate authority bundle
containing the signing certificate, for example
--client-ca-file=/var/lib/kubernetes/ca.pem.
In order for the bootstrapping kubelet to connect to kube-apiserver and request a certificate, it must first authenticate to the server. You can use any authenticator that can authenticate the kubelet.
While any authentication strategy can be used for the kubelet's initial bootstrap credentials, the following two authenticators are recommended for ease of provisioning.
Bootstrap tokens are a simpler and more easily managed method to authenticate kubelets, and do not require any additional flags when starting kube-apiserver.
Whichever method you choose, the requirement is that the kubelet be able to authenticate as a user with the rights to:
A kubelet authenticating using bootstrap tokens is authenticated as a user in the group system:bootstrappers, which is the standard method to use.
As this feature matures, you should ensure tokens are bound to a Role Based Access Control (RBAC) policy which limits requests (using the bootstrap token) strictly to client requests related to certificate provisioning. With RBAC in place, scoping the tokens to a group allows for great flexibility. For example, you could disable a particular bootstrap group's access when you are done provisioning the nodes.
Bootstrap tokens are described in detail here. These are tokens that are stored as secrets in the Kubernetes cluster, and then issued to the individual kubelet. You can use a single token for an entire cluster, or issue one per worker node.
The process is two-fold:
From the kubelet's perspective, one token is like another and has no special meaning.
From the kube-apiserver's perspective, however, the bootstrap token is special. Due to its type, namespace and name, kube-apiserver recognizes it as a special token,
and grants anyone authenticating with that token special bootstrap rights, notably treating them as a member of the system:bootstrappers group. This fulfills a basic requirement
for TLS bootstrapping.
The details for creating the secret are available here.
If you want to use bootstrap tokens, you must enable it on kube-apiserver with the flag:
--enable-bootstrap-token-auth=true
kube-apiserver has the ability to accept tokens as authentication.
These tokens are arbitrary but should represent at least 128 bits of entropy derived
from a secure random number generator (such as /dev/urandom on most modern Linux
systems). There are multiple ways you can generate a token. For example:
head -c 16 /dev/urandom | od -An -t x | tr -d ' '
will generate tokens that look like 02b50b05283e98dd0fd71db496ef01e8.
The token file should look like the following example, where the first three values can be anything and the quoted group name should be as depicted:
02b50b05283e98dd0fd71db496ef01e8,kubelet-bootstrap,10001,"system:bootstrappers"
Add the --token-auth-file=FILENAME flag to the kube-apiserver command (in your
systemd unit file perhaps) to enable the token file. See docs
here for
further details.
Now that the bootstrapping node is authenticated as part of the
system:bootstrappers group, it needs to be authorized to create a
certificate signing request (CSR) as well as retrieve it when done.
Fortunately, Kubernetes ships with a ClusterRole with precisely these (and
only these) permissions, system:node-bootstrapper.
To do this, you only need to create a ClusterRoleBinding that binds the system:bootstrappers group to the cluster role system:node-bootstrapper.
# enable bootstrapping nodes to create CSR
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRoleBinding
metadata:
name: create-csrs-for-bootstrapping
subjects:
- kind: Group
name: system:bootstrappers
apiGroup: rbac.authorization.k8s.io
roleRef:
kind: ClusterRole
name: system:node-bootstrapper
apiGroup: rbac.authorization.k8s.io
While the apiserver receives the requests for certificates from the kubelet and authenticates those requests, the controller-manager is responsible for issuing actual signed certificates.
The controller-manager performs this function via a certificate-issuing control loop. This takes the form of a cfssl local signer using assets on disk. Currently, all certificates issued have one year validity and a default set of key usages.
In order for the controller-manager to sign certificates, it needs the following:
As described earlier, you need to create a Kubernetes CA key and certificate, and distribute it to the control plane nodes. These will be used by the controller-manager to sign the kubelet certificates.
Since these signed certificates will, in turn, be used by the kubelet to authenticate as a regular kubelet to kube-apiserver, it is important that the CA
provided to the controller-manager at this stage also be trusted by kube-apiserver for authentication. This is provided to kube-apiserver
with the flag --client-ca-file=FILENAME (for example, --client-ca-file=/var/lib/kubernetes/ca.pem), as described in the kube-apiserver configuration section.
To provide the Kubernetes CA key and certificate to kube-controller-manager, use the following flags:
--cluster-signing-cert-file="/etc/path/to/kubernetes/ca/ca.crt" --cluster-signing-key-file="/etc/path/to/kubernetes/ca/ca.key"
for example:
--cluster-signing-cert-file="/var/lib/kubernetes/ca.pem" --cluster-signing-key-file="/var/lib/kubernetes/ca-key.pem"
The validity duration of signed certificates can be configured with flag:
--cluster-signing-duration
In order to approve CSRs, you need to tell the controller-manager that it is acceptable to approve them. This is done by granting RBAC permissions to the correct group.
There are two distinct sets of permissions:
nodeclient: If a node is creating a new certificate for a node, then it does not have a certificate yet. It is authenticating using one of the tokens listed above, and thus is part of the group system:bootstrappers.selfnodeclient: If a node is renewing its certificate, then it already has a certificate (by definition), which it uses continuously to authenticate as part of the group system:nodes.To enable the kubelet to request and receive a new certificate, create a ClusterRoleBinding that binds the group in which the bootstrapping node is a member system:bootstrappers to the ClusterRole that grants it permission, system:certificates.k8s.io:certificatesigningrequests:nodeclient:
# Approve all CSRs for the group "system:bootstrappers"
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRoleBinding
metadata:
name: auto-approve-csrs-for-group
subjects:
- kind: Group
name: system:bootstrappers
apiGroup: rbac.authorization.k8s.io
roleRef:
kind: ClusterRole
name: system:certificates.k8s.io:certificatesigningrequests:nodeclient
apiGroup: rbac.authorization.k8s.io
To enable the kubelet to renew its own client certificate, create a ClusterRoleBinding that binds the group in which the fully functioning node is a member system:nodes to the ClusterRole that
grants it permission, system:certificates.k8s.io:certificatesigningrequests:selfnodeclient:
# Approve renewal CSRs for the group "system:nodes"
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRoleBinding
metadata:
name: auto-approve-renewals-for-nodes
subjects:
- kind: Group
name: system:nodes
apiGroup: rbac.authorization.k8s.io
roleRef:
kind: ClusterRole
name: system:certificates.k8s.io:certificatesigningrequests:selfnodeclient
apiGroup: rbac.authorization.k8s.io
The csrapproving controller that ships as part of
kube-controller-manager and is enabled
by default. The controller uses the
SubjectAccessReview API to
determine if a given user is authorized to request a CSR, then approves based on
the authorization outcome. To prevent conflicts with other approvers, the
builtin approver doesn't explicitly deny CSRs. It only ignores unauthorized
requests. The controller also prunes expired certificates as part of garbage
collection.
Finally, with the control plane nodes properly set up and all of the necessary authentication and authorization in place, we can configure the kubelet.
The kubelet requires the following configuration to bootstrap:
kubeconfig file that does not yet exist; it will place the bootstrapped config file herekubeconfig file to provide the URL for the server and bootstrap credentials, e.g. a bootstrap tokenThe bootstrap kubeconfig should be in a path available to the kubelet, for example /var/lib/kubelet/bootstrap-kubeconfig.
Its format is identical to a normal kubeconfig file. A sample file might look as follows:
apiVersion: v1
kind: Config
clusters:
- cluster:
certificate-authority: /var/lib/kubernetes/ca.pem
server: https://my.server.example.com:6443
name: bootstrap
contexts:
- context:
cluster: bootstrap
user: kubelet-bootstrap
name: bootstrap
current-context: bootstrap
preferences: {}
users:
- name: kubelet-bootstrap
user:
token: 07401b.f395accd246ae52d
The important elements to note are:
certificate-authority: path to a CA file, used to validate the server certificate presented by kube-apiserverserver: URL to kube-apiservertoken: the token to useThe format of the token does not matter, as long as it matches what kube-apiserver expects. In the above example, we used a bootstrap token. As stated earlier, any valid authentication method can be used, not only tokens.
Because the bootstrap kubeconfig is a standard kubeconfig, you can use kubectl to generate it. To create the above example file:
kubectl config --kubeconfig=/var/lib/kubelet/bootstrap-kubeconfig set-cluster bootstrap --server='https://my.server.example.com:6443' --certificate-authority=/var/lib/kubernetes/ca.pem
kubectl config --kubeconfig=/var/lib/kubelet/bootstrap-kubeconfig set-credentials kubelet-bootstrap --token=07401b.f395accd246ae52d
kubectl config --kubeconfig=/var/lib/kubelet/bootstrap-kubeconfig set-context bootstrap --user=kubelet-bootstrap --cluster=bootstrap
kubectl config --kubeconfig=/var/lib/kubelet/bootstrap-kubeconfig use-context bootstrap
To indicate to the kubelet to use the bootstrap kubeconfig, use the following kubelet flag:
--bootstrap-kubeconfig="/var/lib/kubelet/bootstrap-kubeconfig" --kubeconfig="/var/lib/kubelet/kubeconfig"
When starting the kubelet, if the file specified via --kubeconfig does not
exist, the bootstrap kubeconfig specified via --bootstrap-kubeconfig is used
to request a client certificate from the API server. On approval of the
certificate request and receipt back by the kubelet, a kubeconfig file
referencing the generated key and obtained certificate is written to the path
specified by --kubeconfig. The certificate and key file will be placed in the
directory specified by --cert-dir.
All of the above relate to kubelet client certificates, specifically, the certificates a kubelet uses to authenticate to kube-apiserver.
A kubelet also can use serving certificates. The kubelet itself exposes an https endpoint for certain features. To secure these, the kubelet can do one of:
--tls-private-key-file and --tls-cert-file flagsThe client certificate provided by TLS bootstrapping is signed, by default, for client auth only, and thus cannot
be used as serving certificates, or server auth.
However, you can enable its server certificate, at least partially, via certificate rotation.
Kubernetes v1.8 and higher kubelet implements beta features for enabling
rotation of its client and/or serving certificates. These can be enabled through
the respective RotateKubeletClientCertificate and
RotateKubeletServerCertificate feature flags on the kubelet and are enabled by
default.
RotateKubeletClientCertificate causes the kubelet to rotate its client
certificates by creating new CSRs as its existing credentials expire. To enable
this feature pass the following flag to the kubelet:
--rotate-certificates
RotateKubeletServerCertificate causes the kubelet both to request a serving
certificate after bootstrapping its client credentials and to rotate that
certificate. To enable this feature pass the following flag to the kubelet:
--rotate-server-certificates
The CSR approving controllers implemented in core Kubernetes do not
approve node serving certificates for security
reasons. To use
RotateKubeletServerCertificate operators need to run a custom approving
controller, or manually approve the serving certificate requests.
A deployment-specific approval process for kubelet serving certificates should typically only approve CSRs which:
spec.username field is of the form
system:node:<nodeName> and spec.groups contains system:nodes)spec.usages contains server auth,
optionally contains digital signature and key encipherment, and contains no other usages)spec.request to verify subjectAltNames)All of TLS bootstrapping described in this document relates to the kubelet. However, other components may need to communicate directly with kube-apiserver. Notable is kube-proxy, which is part of the Kubernetes control plane and runs on every node, but may also include other components such as monitoring or networking.
Like the kubelet, these other components also require a method of authenticating to kube-apiserver. You have several options for generating these credentials:
kube-system namespace. Since it will be in-cluster, you can give it a proper service account with appropriate permissions to perform its activities. This may be the simplest way to configure such services.CSRs can be approved outside of the approval flows builtin to the controller manager.
The signing controller does not immediately sign all certificate requests.
Instead, it waits until they have been flagged with an "Approved" status by an
appropriately-privileged user. This flow is intended to allow for automated
approval handled by an external approval controller or the approval controller
implemented in the core controller-manager. However cluster administrators can
also manually approve certificate requests using kubectl. An administrator can
list CSRs with kubectl get csr and describe one in detail with kubectl describe csr <name>. An administrator can approve or deny a CSR with kubectl certificate approve <name> and kubectl certificate deny <name>.