Ns4Kafka

Ns4Kafka introduces namespace functionality to Apache Kafka, as well as a new deployment model for Kafka resources using Kafkactl, which follows best practices from Kubernetes.

Table of Contents

• Principles
  • Namespace Isolation
  • Desired State
  • Server Side Validation
  • CLI
• Download
• Install
• Demo Environment
• Configuration
  • Authentication
    • Local
    • GitLab
  • Kafka Broker
  • Managed clusters
  • AKHQ
• Administration
• Contribution

Principles

Ns4Kafka is an API that provides controllers for listing, creating, and deleting various Kafka resources, including topics, connectors, schemas, and Kafka Connect clusters. The solution is built on several principles.

Namespace Isolation

Ns4Kafka implements the concept of namespaces, which enable encapsulation of Kafka resources within specific namespaces. Each namespace can only view and manage the resources that belong to it, with other namespaces being isolated from each other. This isolation is achieved by assigning ownership of names and prefixes to specific namespaces.

Desired State

Whenever you deploy a Kafka resource using Ns4Kafka, the solution saves it to a dedicated topic and synchronizes the Kafka cluster to ensure that the resource's desired state is achieved.

Server Side Validation

Ns4Kafka allows you to apply customizable validation rules to ensure that your resources are configured with the appropriate values.

CLI

Ns4Kafka includes Kafkactl, a command-line interface (CLI) that enables you to deploy your Kafka resources 'as code' within your namespace using YAML descriptors. This tool can also be used in continuous integration/continuous delivery (CI/CD) pipelines.

Download

You can download Ns4Kafka as a fat jar from the project's releases page on GitHub at https://github.com/michelin/ns4kafka/releases. Please note that Java 21 is required to run the application.

Additionally, a Docker image of the solution is available at https://hub.docker.com/repository/docker/michelin/ns4kafka.

Install

To operate, Ns4Kafka requires a Kafka broker for data storage and GitLab for user authentication.

The solution is built on the Micronaut framework and can be configured with any Micronaut property source loader.

To override the default properties from the application.yml file, you can set the micronaut.config.file system property when running the fat jar file, like so:

java -Dmicronaut.config.file=application.yml -jar ns4kafka.jar

Alternatively, you can set the MICRONAUT_CONFIG_FILE environment variable and then run the jar file without additional parameters, as shown below:

MICRONAUT_CONFIG_FILE=application.yml 
java -jar ns4kafka.jar

Demo Environment

To run and try out the application, you can use the provided docker-compose files located in the .docker directory.

docker-compose up -d

This command will start multiple containers, including:

• 1 Kafka broker (KRaft mode)
• 1 Schema registry
• 1 Kafka Connect
• 1 Control Center
• Ns4Kafka, with customizable config.yml and logback.xml files
• Kafkactl, with multiple deployable resources in /resources

Please note that SASL/SCRAM authentication and authorization using ACLs are enabled on the broker.

To get started, you'll need to perform the following steps:

1. Define a GitLab admin group for Ns4Kafka in the application.yml file. You can find an example here. It is recommended to choose a GitLab group you belong to in order to have admin rights.
2. Define a GitLab token for Kafkactl in the config.yml file. You can refer to the installation instructions here.
3. Define a GitLab group you belong to in the role bindings of the resources/admin/namespace.yml file. This is demonstrated in the example here.

Alternatively, a docker-compose file running AKHQ instead of Control Center is available in the .docker directory.

docker-compose -f docker-compose-akhq.yml up -d

Configuration

Authentication

Ns4Kafka supports two types of authentication:

• Basic authentication

curl -u username:password http://localhost:8080/api/namespaces/myNamespace/topics

• Bearer token authentication

By generating a JWT access token from the built-in Micronaut LoginController and using it in the Authorization header.

curl -X POST -d '{"username":"username","password":"password"}' -H "Content-Type: application/json" http://localhost:8080/login

The delivered JWT token will have the following format:

{
  "roleBindings": [
    {
      "namespace": "myNamespace",
      "verbs": [
        "GET",
        "POST",
        "PUT",
        "DELETE"
      ],
      "resourceTypes": [
        "schemas",
        "schemas/config",
        "topics",
        "topics/import",
        "topics/delete-records",
        "connectors",
        "connectors/import",
        "connectors/change-state",
        "connect-clusters",
        "connect-clusters/vaults",
        "acls",
        "consumer-groups/reset",
        "streams"
      ]
    }
  ],
  "sub": "user.name@mail.com",
  "nbf": 1711905057,
  "roles": [
    "isAdmin()"
  ],
  "iss": "ns4kafka",
  "exp": 1711908657,
  "iat": 1711905057
}

The token will be valid for 1 hour by default.

The roleBindings field contains the permissions granted to the user.

An ID provider is required to authenticate users. The following ID providers are supported.

Local

The local ID provider is used for testing purposes. It is not recommended for production environments.

To set up authentication with the local ID provider, you can use the following configuration:

ns4kafka:
  security:
    admin-group: adminGroup
    local-users:
      - username: admin
        password: 8c6976e5b5410415bde908bd4dee15dfb167a9c873fc4bb8a81f6f2ab448a918
        groups:
          - "adminGroup"
      - username: user
        password: 8c6976e5b5410415bde908bd4dee15dfb167a9c873fc4bb8a81f6f2ab448a918
        groups:
          - "userGroup"

Identities are stored in the local-users configuration. The password is hashed using the SHA-256 algorithm. The groups used to grant access to namespaces are defined in the groups field.

The admin group is set to "adminGroup" in the example above. Users will be granted admin privileges if they belong to the local group "adminGroup".

GitLab

GitLab is recommended for production environments. It uses GitLab groups to grant access to namespaces. From a given GitLab token, it retrieves the user's GitLab groups and checks if any of them match any of the role bindings.

To set up authentication with GitLab, you can use the following configuration:

micronaut:
  gitlab:
    enabled: true
    url: https://gitlab.com
  token:
    jwt:
      signatures:
        secret:
          generator:
            secret: "changeit"

ns4kafka:
  security:
    admin-group: ADMIN_GROUP

The micronaut.gitlab.url property is set to the GitLab instance URL. The micronaut.token.jwt.signatures.secret.generator.secret property is used to sign the JWT token and should be changed update to a secure value.

The admin group is set to "ADMIN_GROUP" in the example above. Users will be granted admin privileges if they belong to the GitLab group "ADMIN_GROUP".

Kafka Broker

You can configure authentication to the Kafka brokers using the following:

kafka:
  bootstrap.servers: "localhost:9092"
  sasl.mechanism: "PLAIN"
  security.protocol: "SASL_PLAINTEXT"
  sasl.jaas.config: "org.apache.kafka.common.security.scram.ScramLoginModule required username=\"admin\" password=\"admin\";"

The configuration will depend on the authentication method selected for your broker.

Managed clusters

Managed clusters are the clusters where Ns4Kafka namespaces are deployed, and Kafka resources are managed.

You can configure your managed clusters with the following properties:

ns4kafka:
  managed-clusters:
    clusterNameOne:
      manage-users: true
      manage-acls: true
      manage-topics: true
      manage-connectors: true
      drop-unsync-acls: true
      provider: "SELF_MANAGED"
      config:
        bootstrap.servers: "localhost:9092"
        sasl.mechanism: "PLAIN"
        security.protocol: "SASL_PLAINTEXT"
        sasl.jaas.config: "org.apache.kafka.common.security.scram.ScramLoginModule required username=\"admin\" password=\"admin\";"
        cluster.id: "lkc-abcde"
      schema-registry:
        url: "http://localhost:8081"
        basicAuthUsername: "user"
        basicAuthPassword: "password"
      connects:
        connectOne:
          url: "http://localhost:8083"
          basicAuthUsername: "user"
          basicAuthPassword: "password"

The name for each managed cluster has to be unique. This is this name you have to set in the field metadata.cluster of your namespace descriptors.

Property	type	description
manage-users	boolean	Does the cluster manages users ?
manage-acls	boolean	Does the cluster manages access control entries ?
manage-topics	boolean	Does the cluster manages topics ?
manage-connectors	boolean	Does the cluster manages connects ?
drop-unsync-acls	boolean	Should Ns4Kafka drop unsynchronized ACLs
provider	boolean	The kind of cluster. Either SELF_MANAGED or CONFLUENT_CLOUD
config.bootstrap.servers	string	The location of the clusters servers
config.cluster.id	string	The cluster id. Required to use Confluent Cloud tags.
schema-registry.url	string	The location of the Schema Registry
schema-registry.basicAuthUsername	string	Basic authentication username to the Schema Registry
schema-registry.basicAuthPassword	string	Basic authentication password to the Schema Registry
connects.connect-name.url	string	The location of the kafka connect
connects.connect-name.basicAuthUsername	string	Basic authentication username to the Kafka Connect
connects.connect-name.basicAuthPassword	string	Basic authentication password to the Kafka Connect

The configuration will depend on the authentication method selected for your broker, schema registry and Kafka Connect.

AKHQ

AKHQ can be integrated with Ns4Kafka to provide access to resources within your namespace during the authentication process.

To enable this integration, follow these steps:

1. Configure LDAP authentication in AKHQ.
2. Add the Ns4Kafka claim endpoint to AKHQ's configuration:

akhq:
  security:
    rest:
      enabled: true
      url: https://ns4kafka/akhq-claim/v3

For AKHQ versions from v0.20 to v0.24, use the /akhq-claim/v2 endpoint. For AKHQ versions prior to v0.20, use the /akhq-claim/v1 endpoint.

3. In your Ns4Kafka configuration, specify the following settings for AKHQ:

• For AKHQ versions v0.25 and later

ns4kafka:
  akhq:
    admin-group: LDAP-ADMIN-GROUP
    roles:
      TOPIC: topic-read
      CONNECT: connect-rw
      SCHEMA: registry-read
      GROUP: group-read
      CONNECT_CLUSTER: connect-cluster-read
    admin-roles:
      TOPIC: topic-admin
      CONNECT: connect-admin
      SCHEMA: registry-admin
      GROUP: group-read
      CONNECT_CLUSTER: connect-cluster-read 

• For AKHQ versions prior to v0.25

ns4kafka:
  akhq:
    admin-group: LDAP-ADMIN-GROUP
    former-admin-roles:
      - topic/read
      - topic/data/read
      - group/read
      - registry/read
      - connect/read
      - connect/state/update
      - users/reset-password
    group-label: support-group
    former-roles:
      - topic/read
      - topic/data/read
      - group/read
      - registry/read
      - connect/read
      - connect/state/update

If the admin group is set to "LDAP-ADMIN-GROUP", users belonging to this LDAP group will be granted admin privileges.

4. In your namespace configuration, define an LDAP group:

apiVersion: v1
kind: Namespace
metadata:
  name: myNamespace
  cluster: local
  labels:
    contacts: namespace.owner@example.com
    support-group: NAMESPACE-LDAP-GROUP

Once the configuration is in place, after successful authentication in AKHQ, users belonging to the NAMESPACE-LDAP-GROUP will be able to access the resources within the myNamespace namespace.

Administration

The setup of namespaces, owner ACLs, role bindings, and quotas is the responsibility of Ns4Kafka administrators, as these resources define the context in which project teams will work. To create your first namespace, please refer to the Kafkactl documentation.

Contribution

We welcome contributions from the community! Before you get started, please take a look at our contribution guide to learn about our guidelines and best practices. We appreciate your help in making Ns4Kafka a better tool for everyone.