The more prominent and complex Kubernetes deployments become, the more important it is to define strict access controls and tighter security. In this blog, Kasun has explained how RBAC can be implemented in Kubernetes clusters to restrict user permissions to relevant resources only.
With the inception of Kubernetes as an open-source project in 2015, it became the new standard of managing software in the cloud. Modern applications are moving from monolithic architecture to microservice architecture and containerization. When managing multiple microservice applications, managing the number of containers will be a challenging task. To overcome this, we can use Kubernetes for container orchestration. Kubernetes will manage the entire lifecycle of individual containers, spinning up and shutting down resources as needed. If a container shuts down unexpectedly, the orchestration platform will react by launching another container in its place.
To give you an idea, these are some of well known features of Kubernetes:
- Automated Scheduling
- Self-Healing Capabilities
- Automated rollouts & rollback
- Horizontal Scaling & Load Balancing
- Environment consistency for development, testing, and production
- Infrastructure loosely coupled to all component can act as a separate unit
- Provides a higher density of resource utilization
- Offers enterprise-ready features
- Application-centric management
- Auto-scalable infrastructure
- Predictable infrastructure
RBAC is a security measurement, designed to restrict access to valuable resources based on the roles assigned to the entity(user, group, or service account). We can see RBAC in modern applications and cloud providers. Let's identify why RBAC is important for us.
As an example, let's assume your application doesn't use any RBAC system. An administrator uses his username and password to log in to the system and has full access to the system. In this case, if we add regular users, they will also have full permission in the system.
With the expansion of the Kubernetes environment, complexity grows, and role-based security becomes essential. Therefore, we need to divide the cluster into multiple namespaces and limit access from service accounts and users for each resource.
Not every user needs unrestricted access to the cluster to create, modify or delete resources. As the users and cluster resources increase, you will need to limit access to resources and allow relevant permissions required.
Kubernetes have two kinds of roles for defining the action users can perform within the cluster. ClusterRole and Roles operate in Kubernetes within a cluster or namespace, respectively. In Kubernetes, you can use RBAC to assign roles to Kubernetes subjects(users, groups and service accounts) with RoleBindings and ClusterRoleBindings. Kubernetes allowed administrators to create custom roles including but not limited to built-in roles as follows:
Cluster-Admin: The "superuser" role of Kubernetes, this user can perform any action on any resource. We can use ClusterRoleBinding to grant cluster-wide full access or RoleBinding for namespace-wide full access.
Admin: Has unlimited read/write access to resources within a namespace. This role can be created with Role and RoleBindings with a specific namespace.
Edit: Provide full read/write access to the namespace. It can't view or modify Role or RoleBindings.
View: This role allows read-only access within the given namespace.
Kubernetes Service Accounts are used for creating managed Kubernetes resources via Kubernetes API, meant to be used by Kubernetes entities such as Pods for authenticating Kubernetes API server or external services.
A Role always sets permissions within a particular namespace; when you create a Role, you have to specify the namespace it belongs to. Following is an example of a YAML manifest for Kubernetes Role.
Note: The below commands were tested in Kubernetes v1.17
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 RoleBinding 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 references the role granted. A RoleBinding grants permissions within a specific namespace. Following is an example of a YAML manifest for Kubernetes RoleBinding.
apiVersion: rbac.authorization.k8s.io/v1 namespace. kind: RoleBinding metadata: name: read-pods namespace: default subjects: - kind: User name: jane apiGroup: rbac.authorization.k8s.io roleRef: kind: Role name: pod-reader you wish to bind to apiGroup: rbac.authorization.k8s.io
ClusterRole is used to provide permissions to a non-namespaced resource. ClusterRoles have several uses. You can use a ClusterRole to:
- define permissions on namespaced resources and be granted within an individual namespace(s)
- define permissions on namespaced resources and be granted across all namespaces
- define permissions on cluster-scoped resources
To define a role within a namespace, use a Role; to define a cluster-wide role we use ClusterRole.
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"]
ClusterRoleBinding is used to grant cluster-wide access to a user or set of users, which we call as subjects (users, groups or service accounts). Following is an example for ClusterRoleBinding:
# This cluster role binding allows anyone in the "manager" group to read secrets in any namespace. apiVersion: rbac.authorization.k8s.io/v1 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
We can get started in simplest way by applying RBAC into Kubernetes using Role and RoleBinding.
Scenario - We have a service account name
serviceAccount1 and need access to all the resources in the namespace
apiVersion: v1 kind: ServiceAccount metadata: namespace: default name: serviceAccount1
Note: We assume the serviceAccount1 has already been created.
kind: Role apiVersion: rbac.authorization.k8s.io/v1 metadata: namespace: app1 name: app1admin rules: - apiGroups: ["*"] resources: ["*"] verbs: ["*"]
In the above YAML role, we grant full access to all the resources within the namespace app1. As I mentioned previously, roles are namespaced, so we need to specify which namespace role is applied when creating a Role. Next, we look into RoleBinding.
kind: RoleBinding apiVersion: rbac.authorization.k8s.io/v1 metadata: name: app1adminrolebinding namespace: app1 subjects: - kind: ServiceAccount name: serviceAccount1 apiGroup: "" roleRef: kind: Role name: app1admin apiGroup: ""
From RoleBinding we bind the service account with the role we created previously. So now, the service account has only access to the resources inside the app1 namespace.
It's possible to bind a service account from a different namespace, so that the service account can access the resources in app1 namespace with the role attached to the RoleBinding.
ClusterRoles do not belong to any namespace. Therefore ClusterRoles does not scope to any single namespace. But if ClusterRole is linked with a service account via a RoleBinding, ClusterRole permissions are only applied to the namespace that RoleBinding has created.
kind: RoleBinding apiVersion: rbac.authorization.k8s.io/v1 metadata: name: adminbinding namespace: app1 subjects: - kind: ServiceAccount name: serviceAccount1 namespace: app1 apiGroup: "" roleRef: kind: ClusterRole name: cluster-admin apiGroup: ""
ClusterRole is used to give permissions to cluster-wide resources. ClusterRole or ClusterRoleBinding does not belong to any namespace. Therefore the permissions assigned are cluster-wide. For example, a service account needs to access all the resources in the cluster. So we can use ClusterRoles and ClusterRoleBindings to grant him access to all persistent volumes in the cluster. Example of ClusterRole and ClusterRoleBinding are shown below:
apiVersion: rbac.authorization.k8s.io/v1 kind: ClusterRoleBinding metadata: name: storageadminclusterrolebinding subjects: - kind: ServiceAccount name: storageaccess apiGroup: "" namespace: app1 roleRef: kind: ClusterRole name: cluster-admin apiGroup: ""
With the above examples, we can understand the benefits and use cases of Kubernetes RBAC. Below is the summary of what we discussed:
- Roles and RoleBindings must exist in the same namespace.
- RoleBindings can exist in separate namespaces to service accounts.
- RoleBindings can link cluster roles, but they only grant access to the namespace of the role binding.
- ClusterRoleBindings link accounts to cluster roles and grant access across all resources.
- ClusterRoleBindings can not reference roles. Squadcast is an incident management tool that’s purpose-built for SRE. Your team can get rid of unwanted alerts, receive relevant notifications, work in collaboration using the virtual incident war rooms, and use automated tools like runbooks to eliminate toil.