Configure Authentication for Confluent Components using Confluent for Kubernetes

Confluent Platform components are configured without authentication by default. This document presents the supported authentication concepts and describes how to configure authentication for Confluent Platform using Confluent for Kubernetes (CFK).

For more details on security concepts in Confluent Platform, see Security in Confluent Platform.

For a comprehensive tutorial scenario for configuring authentication, see Deploy Secure Confluent Platform.

Configure authentication to access ZooKeeper

This section describes the following authentication mechanisms for Kafka to access ZooKeeper:

• SASL/DIGEST authentication
• mTLS authentication

SASL/DIGEST authentication

Server-side SASL/DIGEST authentication for ZooKeeper

ZooKeeper supports authentication using the SASL DIGEST-MD5 mechanism.

You can use the JAAS and JAAS pass-through mechanisms to set up the credentials.

Create server-side SASL/DIGEST credentials using JAAS config

When you use jaasConfig, you provide the user names and passwords, and CFK automates configuration. For example, when you add, remove, or update users, CFK automatically updates JAAS config. This is the recommended way to configure SASL/DIGEST for ZooKeeper.

The expected key for the server-side SASL/DIGEST credential is digest-users.json.

1. Create a .json file and add the expected value, in the following format:

   {
   "username1": "password1",
   "username2": "password2"
   }
   

2. Create a Kubernetes secret using the expected key (digest-users.json) and the value file you created in the previous step.

   The following example command creates a Kubernetes secret, using the ./digest-users.json file that contains the credentials:

   kubectl create secret generic credential \
     --from-file=digest-users.json=./digest-users.json \
     --namespace confluent
   

Create server-side SASL/DIGEST credentials using JAAS config pass-through

An alternate way to configure JAAS is to use jaasConfigPassThrough. If you have customizations, such as using custom login handlers, you can bypass the CFK automation and provide the configuration directly using jaasConfigPassThrough.

The expected server-side key for jaasConfigPassThrough is digest-jaas.conf.

1. Create a .conf file and add the expected value. For example:

   Server {
     org.apache.zookeeper.server.auth.DigestLoginModule required
       user_super="adminsecret"
       user_user1="user1-secret";
    };
   

2. You can use a Kubernetes secret or a directory path in the container to store the credentials.

   • Create a Kubernetes Secret with the expected key (digest-jaas.conf) and the value file you created in the previous step.

     The following example command creates a Kubernetes secret, using the ./digest-users.conf file that contains the credentials:

     kubectl create secret generic credential \
       --from-file=digest-jaas.conf=./digest-users.conf \
       --namespace confluent
     

   • Use a directory path in the container to provide the required credentials.

     If jaasConfigPassThrough.directoryPathInContainer is configured as /vaults/secrets in the ZooKeeper CR, the expected file, digest-jaas.conf, must exist in that directory path.

     See Provide secrets for Confluent Platform component CR for providing the credential and required annotations when using Vault.

     See CFK GitHub examples for more information on using the directoryPathInContainer property with Vault.

Configure ZooKeeper for server-side SASL/DIGEST authentication

Enable the server-side SASL/DIGEST authentication in the ZooKeeper CR as below:

kind: Zookeeper
spec:
  authentication:
    type: digest                 --- [1]
    jaasConfig:                  --- [2]
      secretRef:                 --- [3]
    jaasConfigPassThrough:       --- [4]
      secretRef:                 --- [5]
      directoryPathInContainer:  --- [6]

• [1] Required. Set to digest.
• [2] When you use jaasConfig, you provide the user names and passwords, and CFK automates configuration. For example, when you add, remove, or update users, CFK automatically updates JAAS config. This is the recommended way to configure SASL/DIGEST for ZooKeeper.
• One of [3], [5], or [6] is required. Only specify one.
• [3] Provide the name of the Kubernetes secret that you created in the previous section.
• [4] An alternate way to configure JAAS is to use jaasConfigPassThrough. If you have customizations, such as using custom login handlers, you can bypass the CFK automation and provide the configuration directly.
• [5] Provide a Kubernetes secret that you created in the previous section.
• [6] Provide the directory path in the container you set up in the previous section. The expected file, digest-jaas.conf, must exist in the directory path.

Client-side SASL/DIGEST authentication for ZooKeeper

Configure client-side Kafka to authenticate to ZooKeeper using SASL/DIGEST.

You can use the JAAS and JAAS pass-through mechanisms to set up the credentials.

Create client-side SASL/DIGEST credentials using JAAS config

When you use jaasConfig, you provide the user names and passwords, and CFK automates configuration. For example, when you add, remove, or update users, CFK automatically updates the JAAS config. This is the recommended way to configure SASL/DIGEST for ZooKeeper.

The expected client-side key for jaasConfig is digest.txt.

1. Create a .txt file and add the expected value, in the following format:

   username=<user>
   password=<password>
   

2. Create a Kubernetes Secret with the expected key (digest.txt) and the value file you created in the previous step.

   The following example command creates a Kubernetes secret, using the ./digest-users.txt file that contains the credentials:

   kubectl create secret generic credential \
     --from-file=digest.txt=./digest-users.txt \
     --namespace confluent
   

Create client-side SASL/DIGEST credentials using JAAS config pass-through

If you have customizations, such as using custom login handlers, you can bypass the CFK automation and provide the configuration directly using jaasConfigPassThrough.

The expected server-side key for jaasConfigPassThrough is digest-jaas.conf.

1. Create a .conf file and add the expected value to the file.

   The following is an example value (the data in the file) for digest-jaas.conf with a standard login module and a user, bob.

   Client { //zookeeper dependencies
       org.apache.zookeeper.server.auth.DigestLoginModule required
       username="bob"
       password="password";
   };
   

2. You can use a Kubernetes secret or a directory path in the container to store the credentials.

   • Create a Kubernetes secret using the expected key (digest-jaas.conf) and the value file you created in the previous step.

     The following example command creates a Kubernetes secret, using the ./digest-jaas-users.conf file that contains the credentials:

     kubectl create secret generic credential \
       --from-file=digest-jaas.conf=./digest-jaas-users.conf \
       --namespace confluent
     

   • Use a directory path in the container to provide the required credentials.

     If jaasConfigPassThrough.directoryPathInContainer is configured as /vaults/secrets in the Kafka CR, the expected file, digest-jaas.conf, must exist in that directory path.

     See Provide secrets for Confluent Platform component CR for providing the credential and required annotations when using Vault.

     See CFK GitHub examples for more information on using the directoryPathInContainer property with Vault.

Configure Kafka for client-side SASL/DIGEST authentication

For Kafka to authenticate to ZooKeeper using SASL/DIGEST authentication, configure the Kafka CR as below:

kind: Kafka
spec:
  dependencies:
    zookeeper:
      authentication:
        type: digest                --- [1]
        jaasConfig:                 --- [2]
          secretRef:                --- [3]
        jaasConfigPassThrough:      --- [4]
          secretRef:                --- [5]
          directoryPathInContainer: --- [6]

• [1] Required. Set to digest.

• [2] When you use jaasConfig, you provide the user names and passwords, and CFK automates configuration. For example, when you add, remove, or update users, CFK automatically updates the JAAS config. This is the recommended way to configure SASL/DIGEST for ZooKeeper.

• One of [3], [5], or [6] is required. Only specify one.

• [3] Provide the name of the Kubernetes secret you set up in the previous section for Kafka to authenticate to ZooKeeper.

• [4] An alternate way to configure JAAS is to use jaasConfigPassThrough. If you have customizations, such as using custom login handlers, you can bypass the CFK automation and provide the configuration directly.

• [5] Provide the name of the Kubernetes secret you set up in the previous section for Kafka to authenticate to ZooKeeper.

• [6] Provide the container directory path you set up in the previous section.

  See Provide secrets for Confluent Platform component CR for providing the credential and required annotations when using Vault.

  See CFK GitHub examples for more information on using the directoryPathInContainer property with Vault.

mTLS authentication

Server-side mTLS authentication for ZooKeeper

Enable mTLS authentication in the ZooKeeper CR as below:

kind: Zookeeper
spec:
  authentication:
    type: mtls

Client-side mTLS authentication ZooKeeper

For Kafka to authenticate to ZooKeeper using mTLS authentication, configure the Kafka CR as below:

kind: Kafka
spec:
  dependencies:
    zookeeper:
      authentication:
        type: mtls               --- [1]
      tls:
        enabled: true            --- [2]

• [1] Required. Set to mtls.
• [2] Required for mTLS authentication. Set to true.

Configure authentication to access other Confluent Platform components

This section describes the following authentication methods for the Confluent Platform components (other than Kafka and ZooKeeper):

• Basic authentication
• mTLS authentication
• LDAP authentication (for Control Center only)
• Single sign-on (for Control Center only)

Basic authentication

Server-side basic authentication for Confluent components

Create server-side basic credentials

The expected server-side key for basic authentication is basic.txt.

1. Create a .txt file and add the expected value, in the following format:

   <username1>: <password1>, <role that username1 is assigned to>
   ...
   <usernameN>: <passwordN>, <role that usernameN is assigned to>
   

   To be authorized to access a Confluent Platform component, an authenticated user must belong to one of the basic authentication roles. The Roles associated with a user does not differentiate permissions, except Confluent Control Center where the users belonging to restricted roles have limited permissions as stated in Control Center Security.

   The following default basic authentication roles are supported:

   • For REST Proxy: The admin, developer, user, and krp-user roles are available.
   • For ksqlDB: The admin, developer, user, and ksql-user roles are available.
   • For Schema Registry: The admin, developer, user, and sr-user roles are available.
   • For Confluent Control Center: The Administrators and Restricted roles are available.

   Warning

   For Connect, there is no support for roles.

2. You can use a Kubernetes secret or a directory path in the container to store the basic credentials.

   • Create a Kubernetes secret using the expected key (basic.txt) and the value file you created in the previous step.

     The following example command creates a Kubernetes secret, using the ./creds-basic.txt file that contains the credentials:

     kubectl create secret generic credential \
       --from-file=basic.txt=./creds-basic.txt \
       --namespace confluent
     

   • Use a directory path in the container to provide the required credentials.

     See Provide secrets for Confluent Platform component CR for providing the credential and required annotations when using Vault.

     See CFK GitHub examples for more information on using the directoryPathInContainer property with Vault.

Configure Confluent component for server-side basic authentication

Configure the server-side basic authentication in the component CR as below:

kind: <Confluent component>
spec:
  authentication:
    type: basic                  --- [1]
    basic:
      secretRef:                 --- [2]
      directoryPathInContainer:  --- [3]
      restrictedRoles:           --- [4]
      roles:                     --- [5]

• [1] Required. Set to basic.

• [2] or [3] Required. Do not specify both.

• [2] Provide the name of the Kubernetes secret you created in the previous section.

• [3] Provide the path in the container in the previous section. The expected file, basic.txt, must exist in the specified directory path.

• [4] Optional. A list of restricted roles on the server in Confluent Control Center. The setting only applies to Control Center and is ignored for other Confluent Platform components.

  For Control Center, the users who belong to restricted roles have limited permissions as stated in Control Center Security.

• [5] Optional. A list of roles on the server side. To be authorized to access this Confluent Platform component, users must belong to one of these roles. See Create server-side basic credentials for the component default roles.

Client-side basic authentication for Confluent components

Configure client-side Confluent components to authenticate to other Confluent Platform component using basic authentication.

Create client-side basic credentials

The expected client-side key for basic authentication is basic.txt.

1. Create a .txt file and add the expected value, in the following format:

   username=<username>
   password=<password>
   

2. The username/password for basic authentication are either loaded through the secretRef or through directoryPathInContainer.

Configure Confluent components for client-side basic authentication

Enable the client-side basic authentication in the component CR as below. <component> is the Confluent Platform component that this component needs to authenticate to:

kind: <this Confluent component>
spec:
  dependencies:
    <component>:
      url:
      authentication:
        type: basic                --- [1]
        basic:
          secretRef:               --- [2]
          directoryPathInContainer --- [3]

• [1] Required. Set to basic.

• [2] or [3] is required. Do not specify both.

• [2] Provide the name of the Kubernetes secret that you created in the previous section.

• [3] Provide the directory in the container in the previous section.

  See Provide secrets for Confluent Platform component CR for providing the credential and required annotations when using Vault.

mTLS authentication

Server-side mTLS authentication for Confluent components

Configure mTLS authentication in the Confluent component CR as below:

kind: <Confluent component>
spec:
  authentication:
    type: mtls                --- [1]

• [1] Required. Set to mtls.

Client-side mTLS authentication for Confluent components

To configure Confluent components to authenticate to other Confluent Platform components using mTLS authentication, enable mTLS authentication in the component CR as below. <component> is the Confluent Platform component that this component needs to authenticate to:

kind: <this client Confluent component>
spec:
  dependencies:
    <component>:
      url:
      authentication:
        type: mtls               --- [1]
      tls:
        enabled: true            --- [2]

• [1] Required. Set to mtls.
• [2] Required for mTLS authentication. Set to true.

LDAP authentication for Confluent Control Center

Confluent Control Center supports LDAP as an authentication method.

Create LDAP credentials

The expected server-side key for LDAP authentication is ldap.txt.

1. Create a .txt file and add the expected value, in the following format:

   username=<bindDn_value>
   password=<bindPassword_value>
   

   For the password for bindDn, escape any restricted LDAP characters. For best results, avoid characters that require escaping.

2. Create a Kubernetes secret using the expected key (ldap.txt) and the value file you created in the previous step.

   The following example command creates a Kubernetes secret, using the ./creds-ldap.txt file that contains the credentials:

   kubectl create secret generic credential \
     --from-file=ldap.txt=./creds-ldap.txt \
     --namespace confluent
   

Configure Confluent Control Center for server-side LDAP authentication

Configure the Control Center CR to pull users and groups from an LDAP server:

kind: ControlCenter
spec:
  authentication:
    ldap:
      secretRef:                                    --- [1]
      roles:                                        --- [2]
      restrictedRoles:                              --- [3]
      property:                                     --- [4]
        #  useLdaps: "true"
        #  contextFactory: "com.sun.jndi.ldap.LdapCtxFactory"
        #  hostname: ""
        #  port: "389"
        #  bindDn: ""                               --- [5]
        #  bindPassword: ""                         --- [6]
        #  authenticationMethod: ""
        #  forceBindingLogin: "true"
        #  userBaseDn: ""
        #  userRdnAttribute: "sAMAccountName"
        #  userIdAttribute: "sAMAccountName"
        #  userPasswordAttribute: "userPassword"
        #  userObjectClass: "user"
        #  roleBaseDn: ""
        #  roleNameAttribute: "cn"
        #  roleMemberAttribute: "member"
        #  roleObjectClass: "group"

• [1] Provide the Kubernetes secret you created in the previous section for LDAP credentials.
• [2] Optional. A list of roles on the server side. To be authorized to access Confluent Control Center, users must belong to one of these roles. By default, it’s set to ["Administrators", "Restricted"].
• [3] Optional. List of roles with limited read-only access. No editing or creating allowed in Control Center.
• [4] Required. See Configure LdapLoginModule for details.
• [5] [6] If you have security considerations, pass empty values for bindDn and bindPassword in the CR. The CFK will replace them and add them as appropriately.

LDAP over SSL authentication for Confluent Control Center

When configuring Control Center for encryption and authentication using LDAP over SSL (LDAPS), the LDAPS trust store must be exported to the JVM using the configOverrides feature.

This requirement does not apply to RBAC-enabled environments where Control Center relies on the external MDS for authentication.

Add the LDAP SSL certificate to the JVM trust store in the Confluent Control Center CR as shown below:

kind:ControlCenter
spec:
  configOverrides:
    jvm:
      -Djavax.net.ssl.trustStore=<path to truststore.jks>
      -Djavax.net.ssl.trustStorePassword=<password for the truststore>

Single sign-on authentication for Confluent Control Center

In CFK, you can configure single sign-on (SSO) authentication for Control Center using OpenID Connect (OIDC). OIDC is an identity layer that allows third-party applications to verify the identity of the end user.

To enable SSO to Control Center:

1. In the Kafka CR, configure MDS, LDAP and OIDC as below.

   To enable SSO, currently MDS, LDAP, and OIDC need to be configured.

   kind: Kafka
   spec:
     services:
       mds:                                --- [1]
         tokenKeyPair:                     --- [2]
           secretRef:
           directoryPathInContainer:
         provider:
           ldap:                           --- [3]
             address:
             authentication:
             configurations:
           oidc:
             clientCredentials:
               secretRef:                  --- [4]
             configurations:
               groupsClaimName:            --- [5]
               subClaimName: sub           --- [6]
               groupsClaimScope:           --- [7]
               issuer:                     --- [8]
               jwksEndpointUri:            --- [9]
               authorizeBaseEndpointUri:   --- [10]
               tokenBaseEndpointUri:       --- [11]
               refreshToken:               --- [12]
               sessionTokenExpiry:         --- [13]
               sessionMaxTimeout:          --- [14]
   

   • [1] Required.

   • [2] Required. The token key pair to authenticate to the MDS. Use secretRef or directoryPathInContainer to specify.

     You need to add the public key and the token key pair to the secret or directoryPathInContainer. For details, see Create a PEM key pair for MDS.

     An example command to create a secret is:

     kubectl create secret generic mds-token \
       --from-file=mdsPublicKey.pem=mds-publickey.txt \
       --from-file=mdsTokenKeyPair.pem=mds-tokenkeypair.txt \
       --namespace confluent
     

   • [3] Required. See the snippet below for examples configuration properties.

   • [4] The secret that contains a OIDC client ID and the client secret for authorization and token request to IDP.

     Create the secret with the file name oidcClientSecret.txt that contains two keys with their respective values, clientId and clientSecret.

   • [5] Groups in JSON Web Tokens (JWT). The default value is groups.

     JWT (Session Token) is issued by Confluent (MDS) to maintain the session. This contains information passed by IDP in id token along with custom claims (if any) added by the MDS.

   • [6] Sub in JWT (session token). The default value is sub.

   • [7] If any additional scope is needed to include groups in the JWT, set the optional variable based on the IDP. The setting lets the MDS pass these scope(s) during the authorization process.

     Possible values are groups, openid, offline_access, etc.

   • [8] The issuer URL, which is typically the authorization server’s URL. This value is used to compare to the issuer claim in the JWT for verification.

   • [9] The JSON Web Key Set (JWKS) URI. It is used to verify any JWT issued by the IDP.

   • [10] The base endpoint URI for the authorize that initiates an OAuth authorization request.

   • [11] The IDP token endpoint from where a token is requested by the MDS.

   • [12] Specify whether offline access scope would be requested in the authorization URI. Set this to false if offline tokens are not allowed for the user or client in the IDP.

   • [13] Time before the JWT (session token) needs to be renewed. Default value 15 minutes.

     The actual expiry would be the lesser value between the ID token lifetime ([14]) value and the value of this setting.

   • [14] Max time up to which JWT (session token) could be renewed.

   An example snippet of a Kafka CR:

   kind: Kafka
   spec:
     services:
       mds:
         tls:
           enabled: true
         tokenKeyPair:
           secretRef: mds-token
         provider:
           ldap:
             address: ldaps://ldap.operator.svc.cluster.local:636
             authentication:
               type: simple
               simple:
                 secretRef: credential
             tls:
               enabled: true
             configurations:
               groupNameAttribute: cn
               groupObjectClass: group
               groupMemberAttribute: member
               groupMemberAttributePattern: CN=(.*),DC=test,DC=com
               groupSearchBase: dc=test,dc=com
               userNameAttribute: cn
               userMemberOfAttributePattern: CN=(.*),DC=test,DC=com
               userObjectClass: organizationalRole
               userSearchBase: dc=test,dc=com
           oidc:
             clientCredentials:
               secretRef: oidc-client  // to pass oidc client id and client secret
             configurations:
               groupsClaimName: groups
               subClaimName: sub
               groupsClaimScope: groups
               issuer: https://dev-59009577.okta.com/oauth2/aus96p2og3u7Cpwu65d7
               jwksEndpointUri: https://dev-59009577.okta.com/oauth2/aus96p2og3u7Cpwu65d7/v1/keys
               authorizeBaseEndpointUri: https://dev-59009577.okta.com/oauth2/aus96p2og3u7Cpwu65d7/v1/authorize
               tokenBaseEndpointUri: https://dev-59009577.okta.com/oauth2/aus96p2og3u7Cpwu65d7/v1/token
               refreshToken: true
               sessionTokenExpiry: 900000
               sessionMaxTimeout: 21600000
   

2. In the Control Center CR, specify OIDC to enable SSO:

   kind: ControlCenter
     spec:
       dependencies:
         mds:
           ssoProtocol: oidc
   

Configure authentication to access MDS

Bearer authentication

When RBAC is enabled (spec.authorization.type: rbac), CFK always uses the Bearer authentication for Confluent components, ignoring the spec.authentication setting. It is not possible to set the component authentication type to mTLS when RBAC is enabled.

Create client-side Bearer credentials for MDS

Provide the required Bearer credentials.

The expected key is bearer.txt.

1. Create a .txt file with the expected value in the following format:

   username=<username>
   password=<password>
   

2. You can use a Kubernetes secret or a directory path in the container to store the bearer credentials.

   • Create a Kubernetes secret using the expected key (bearer.txt) and the value.

     For example, using the ./c3-mds-client.txt file that you create for Control Center credentials:

     kubectl create secret generic c3-mds-client \
       --from-file=bearer.txt=./c3-mds-client.txt \
       --namespace confluent
     

   • Use a directory path in the container to provide the required credentials.

     See Provide secrets for Confluent Platform component CR for providing the credential and required annotations when using Vault.

     See CFK GitHub examples for more information on using the directoryPathInContainer property with Vault.

Client-side Bearer authentication for MDS

For each of the Confluent components that communicate with MDS, configure bearer authentication mechanism in the component CR:

kind: <Confluent component>
spec:
  dependencies:
    mds:
      authentication:
        type: bearer               --- [1]
        bearer:
          secretRef:               --- [2]
          directoryPathInContainer --- [3]

• [1] Required. Set to bearer.
• [2] or [3] Required. Do not specify both.
• [2] To load username and secret, set to the Kubernetes secret you created in the previous section.
• [3] Provide the path in the container in the previous section. The expected file, bearer.txt, must exist in the specified directory path.