Authorization

This section pertains to Redpanda Console in a self-hosted deployment, and this feature requires an Enterprise license. To upgrade, contact Redpanda sales.

Redpanda Console uses role-based access control (RBAC) to restrict system access to authorized users. All configurations are applied through YAML files which Redpanda Console loads at startup. This page describes the concepts, the available roles, and how you can bind roles to your users.

This page describes the RBAC implementation for Redpanda Console which only manages access to the Console and its functions. This does not apply to clients that interact with a system through other methods such as the Kafka API. To manage client access, you must use either access control lists or Redpanda role-based access control.

RBAC concepts

You can control access to different resources in Console by binding one of the existing roles (admin, editor, viewer) to one or more users. Besides binding roles to individual users, you can also use user groups (for example, Google groups) from your configured identity providers. Users who have multiple roles assigned will receive the union of all permissions that have been defined in the bound roles.

Roles

Each role has a name, as well as a set of permissions, which define the user permissions when the role is bound to a user. Console comes with three primitive roles:

Role viewer

The viewer role grants you the permissions to view all resources within Console. This includes:

  • Viewing all topic aspects (messages, configs, partition, using search filters)

  • Viewing all cluster aspects (node configs, acls, service accounts, quotas)

  • Viewing all consumer group aspects (consumer groups, group offsets and lags)

  • Viewing all schema registry aspects (registered schemas with their contents)

  • Viewing all Kafka connect aspects (list configured clusters and their connectors including the status and connector configs)

It does not include permissions to view the list of Console users (admin tab) that are allowed to use Console.

Role editor

The editor role grants all permissions that come with the viewer role and additionally includes:

  • Managing all topic aspects, such as creating topics, editing topic configuration, deleting topics, and publishing and deleting topic records.

  • Managing all cluster configuration aspects, such as editing node or cluster configs.

  • Managing all consumer group aspects, such as editing group offsets or deleting group offsets.

  • Managing all Kafka connect aspects, such as creating/updating/deleting or starting/pausing/stopping connectors.

It does not include the permission to create/remove ACLs or to create or remove a service account.

Role admin

The editor role grants all permissions that come with the editor role and additionally includes:

  • Managing all service account aspects (create/remove service account)

  • Manage all ACL aspects (create/remove ACLs)

You cannot create your own roles with a custom set of permissions.

Role bindings

To grant users the permissions defined in a role, you must bind the role to a user or a group. Each role binding holds a list of subjects (users or groups) and a reference to the role being granted. Optionally you can add metadata (key-value pairs) which may help you to manage your role bindings. Console shows the metadata in the UI, so that it helps you to understand how a specific user got its permissions.

Example role-bindings.yaml configuration file:

roleBindings:
  - roleName: admin
    metadata:
      # Metadata properties will be shown in the UI. You can omit it if you want to
      name: Developers
      creator: John Doe
    subjects:
      - kind: user
        provider: Google
        name: john.doe@redpanda.com
      - kind: group
        provider: Google
        name: dev-team-console@redpanda.com

This role binding binds Google account john.doe@redpanda.com as well as all Google accounts which are a member of Google group dev-team-console@redpanda.com to the role with name admin. Group memberships are resolved recursively, so that members from nested groups are also considered. Users who have multiple roles assigned through role bindings will inherit the union of these roles' permissions.

To use groups in role bindings you need to configure the Groups sync for your respective provider.

Subjects

Each subject has three properties that are configurable to bind a role to one or more users.

kind: Supported kinds are: group and user.

providers: One of Google, GitHub, Okta, OIDC and Plain.

name: Depending on your kind and provider the name property may refer to different things. This is an overview to what it refers for each provider:

Kind Provider Name Reference

user

Google

Google E-Mail address

group

Google

Google Group Name (which is an E-Mail address)

user

GitHub

Login handle / GitHub username

group

GitHub

GitHub team name within your GitHub organization

user

Okta

Login handle / email

group

Okta

Okta Group ID (not name), for example "00gra1ajmZa1G1ks04x9"

user

OIDC

Configurable via login.oidc.userIdentifyingClaimKey (by default it uses the sub claim from the issued access token)

user

Plain

Username as defined in the login.plain.credentials configuration

Groups sync

If you want to bind Roles to a set of users (for example GitHub teams or Google Groups) you need to grant Console additional permissions, so that it can resolve the memberships of these user sets. You will find more information about the setup in the respective identity provider documentation page.

All group memberships that are used in rolebindings, are resolved at startup. Group memberships will be cached for 15 minutes. Resolving group memberships proactively is required in order to support nested groups.