Red Hat Operators

The Cloud Credential Operator (CCO) manages cloud provider credentials as Kubernetes custom resource definitions (CRDs). The CCO syncs on credentialsRequest custom resources (CRs) to allow OKD components to request cloud provider credentials with the specific permissions that are required for the cluster to run.

By setting different values for the credentialsMode parameter in the install-config.yaml file, the CCO can be configured to operate in several different modes. If no mode is specified, or the credentialsMode parameter is set to an empty string (""), the CCO operates in its default mode.

For platforms where multiple modes are supported (AWS, Azure, and GCP), when the CCO operates in its default mode, it checks the provided credentials dynamically to determine for which mode they are sufficient to process credentialsRequest CRs.

By default, the CCO determines whether the credentials are sufficient for mint mode, which is the preferred mode of operation, and uses those credentials to create appropriate credentials for components in the cluster. If the credentials are not sufficient for mint mode, it determines whether they are sufficient for passthrough mode. If the credentials are not sufficient for passthrough mode, the CCO cannot adequately process credentialsRequest CRs.

If the provided credentials are determined to be insufficient during installation, the installation fails. For AWS, the installer fails early in the process and indicates which required permissions are missing. Other providers might not provide specific information about the cause of the error until errors are encountered.

If the credentials are changed after a successful installation and the CCO determines that the new credentials are insufficient, the CCO puts conditions on any new credentialsRequest CRs to indicate that it cannot process them because of the insufficient credentials.

To resolve insufficient credentials issues, provide a credential with sufficient permissions. If an error occurred during installation, try installing again. For issues with new credentialsRequest CRs, wait for the CCO to try to process the CR again. As an alternative, you can manually create IAM for AWS, Azure, or GCP. For details, see the Manually creating IAM section of the installation content for AWS, Azure, or GCP.

By setting different values for the credentialsMode parameter in the install-config.yaml file, the CCO can be configured to operate in mint, passthrough, or manual mode. These options provide transparency and flexibility in how the CCO uses cloud credentials to process credentialsRequest CRs in the cluster, and allow the CCO to be configured to suit the security requirements of your organization. Not all CCO modes are supported for all cloud providers.

Mint mode is supported for AWS, Azure, and GCP.

Mint mode is the default and recommended best practice setting for the CCO to use. In this mode, the CCO uses the provided admin-level cloud credential to run the cluster.

If the credential is not removed after installation, it is stored and used by the CCO to process credentialsRequest CRs for components in the cluster and create new credentials for each with only the specific permissions that are required. The continuous reconciliation of cloud credentials in mint mode allows actions that require additional credentials or permissions, such as upgrading, to proceed.

The requirement that mint mode stores the admin-level credential in the cluster kube-system namespace might not suit the security requirements of every organization.

When using the CCO in mint mode, ensure that the credential you provide meets the requirements of the cloud on which you are running or installing OKD. If the provided credentials are not sufficient for mint mode, the CCO cannot create an IAM user.

Mint mode with removal or rotation of the admin-level credential is supported for AWS in OKD version 4.4 and later.

This option requires the presence of the admin-level credential during installation, but the credential is not stored in the cluster permanently and does not need to be long-lived.

After installing OKD in mint mode, you can remove the admin-level credential Secret from the cluster. If you remove the Secret, the CCO uses a previously minted read-only credential that allows it to verify whether all credentialsRequest CRs have their required permissions. Once removed, the associated credential can be destroyed on the underlying cloud if desired.

The admin-level credential is not required unless something that requires an admin-level credential needs to be changed, for instance during an upgrade. Prior to each upgrade, you must reinstate the credential Secret with the admin-level credential. If the credential is not present, the upgrade might be blocked.

Passthrough mode is supported for AWS, Azure, GCP, Red Hat OpenStack Platform (RHOSP), oVirt, and VMware vSphere.

In passthrough mode, the CCO passes the provided cloud credential to the components that request cloud credentials. The credential must have permissions to perform the installation and complete the operations that are required by components in the cluster, but does not need to be able to create new credentials. The CCO does not attempt to create additional limited-scoped credentials in passthrough mode.

When using the CCO in passthrough mode, ensure that the credential you provide meets the requirements of the cloud on which you are running or installing OKD. If the provided credentials the CCO passes to a component that creates a credentialsRequest CR are not sufficient, that component will report an error when it tries to call an API that it does not have permissions for.

The credential you provide for passthrough mode in AWS, Azure, or GCP must have all the requested permissions for all credentialsRequest CRs that are required by the version of OKD you are running or installing. To locate the credentialsRequest CRs that are required for your cloud provider, see the Manually creating IAM section of the installation content for AWS, Azure, or GCP.

To install an OKD cluster on Red Hat OpenStack Platform (RHOSP), the CCO requires a credential with the permissions of a member user role.

To install an OKD cluster on oVirt, the CCO requires a credential with the following privileges:

To install an OKD cluster on VMware vSphere, the CCO requires a credential with the following vSphere privileges:

If credentialsRequest CRs change over time as the cluster is upgraded, you must manually update the passthrough mode credential to meet the requirements. To avoid credentials issues during an upgrade, check the credentialsRequest CRs in the release image for the new version of OKD before upgrading. To locate the credentialsRequest CRs that are required for your cloud provider, see the Manually creating IAM section of the installation content for AWS, Azure, or GCP.

When using passthrough mode, each component has the same permissions used by all other components. If you do not reduce the permissions after installing, all components have the broad permissions that are required to run the installer.

After installation, you can reduce the permissions on your credential to only those that are required to run the cluster, as defined by the credentialsRequest CRs in the release image for the version of OKD that you are using.

To locate the credentialsRequest CRs that are required for AWS, Azure, or GCP and learn how to change the permissions the CCO uses, see the Manually creating IAM section of the installation content for AWS, Azure, or GCP.

Manual mode is supported for AWS.

In manual mode, a user manages cloud credentials instead of the CCO. To use this mode, you must examine the credentialsRequest CRs in the release image for the version of OKD that you are running or installing, create corresponding credentials in the underlying cloud provider, and create Kubernetes Secrets in the correct namespaces to satisfy all credentialsRequest CRs for the cluster’s cloud provider.

Using manual mode allows each cluster component to have only the permissions it requires, without storing an admin-level credential in the cluster. This mode also does not require connectivity to the AWS public IAM endpoint. However, you must manually reconcile permissions with new release images for every upgrade.

For information about configuring AWS to use manual mode, see Manually creating IAM for AWS.

Disabled CCO is supported for Azure and GCP.

To manually manage credentials for Azure or GCP, you must disable the CCO. Disabling the CCO has many of the same configuration and maintenance requirements as running the CCO in manual mode, but is accomplished by a different process. For more information, see the Manually creating IAM section of the installation content for Azure or GCP.

openshift-cloud-credential-operator

No configuration required.

The Cluster Authentication Operator installs and maintains the Authentication custom resource in a cluster and can be viewed with:

cluster-authentication-operator

The Cluster Autoscaler Operator manages deployments of the OpenShift Cluster Autoscaler using the cluster-api provider.

cluster-autoscaler-operator

The Cluster Image Registry Operator manages a singleton instance of the OKD registry. It manages all configuration of the registry, including creating storage.

On initial start up, the Operator creates a default image-registry resource instance based on the configuration detected in the cluster. This indicates what cloud storage type to use based on the cloud provider.

If insufficient information is available to define a complete image-registry resource, then an incomplete resource is defined and the Operator updates the resource status with information about what is missing.

The Cluster Image Registry Operator runs in the openshift-image-registry namespace and it also manages the registry instance in that location. All configuration and workload resources for the registry reside in that namespace.

cluster-image-registry-operator

The Cluster Monitoring Operator manages and updates the Prometheus-based cluster monitoring stack deployed on top of OKD.

openshift-monitoring

The Cluster Network Operator installs and upgrades the networking components on an OKD cluster.

The OpenShift Controller Manager Operator installs and maintains the OpenShiftControllerManager custom resource in a cluster and can be viewed with:

The custom resource definitino (CRD) openshiftcontrollermanagers.operator.openshift.io can be viewed in a cluster with:

cluster-openshift-controller-manager-operator

The Cluster Samples Operator manages the sample image streams and templates stored in the openshift namespace.

On initial start up, the Operator creates the default samples configuration resource to initiate the creation of the image streams and templates. The configuration object is a cluster scoped object with the key cluster and type configs.samples.

The image streams are the Fedora CoreOS (FCOS)-based OKD image streams pointing to images on registry.redhat.io. Similarly, the templates are those categorized as OKD templates.

The Cluster Samples Operator deployment is contained within the openshift-cluster-samples-operator namespace. On start up, the install pull secret is used by the image stream import logic in the internal registry and API server to authenticate with registry.redhat.io. An administrator can create any additional secrets in the openshift namespace if they change the registry used for the sample image streams. If created, those secrets contain the content of a config.json for docker needed to facilitate image import.

The image for the Cluster Samples Operator contains image stream and template definitions for the associated OKD release. After the Cluster Samples Operator creates a sample, it adds an annotation that denotes the OKD version that it is compatible with. The Operator uses this annotation to ensure that each sample matches the compatible release version. Samples outside of its inventory are ignored, as are skipped samples.

Modifications to any samples that are managed by the Operator are allowed as long as the version annotation is not modified or deleted. However, on an upgrade, as the version annotation will change, those modifications can get replaced as the sample will be updated with the newer version. The Jenkins images are part of the image payload from the installation and are tagged into the image streams directly.

The samples resource includes a finalizer, which cleans up the following upon its deletion:

Upon deletion of the samples resource, the Cluster Samples Operator recreates the resource using the default configuration.

cluster-samples-operator

The Cluster Storage Operator sets OKD cluster-wide storage defaults. It ensures a default storage class exists for OKD clusters.

cluster-storage-operator

No configuration is required.

Cluster Operators manage specific areas of cluster functionality. The Cluster Version Operator (CVO) manages the lifecycle of cluster Operators, many of which are installed in OKD by default.

The CVO also checks with the OpenShift Update Service to see the valid updates and update paths based on current component versions and information in the graph.

cluster-version-operator

The Console Operator installs and maintains the OKD web console on a cluster.

console-operator

The DNS Operator deploys and manages CoreDNS to provide a name resolution service to pods that enables DNS-based Kubernetes Service discovery in OKD.

The Operator creates a working default deployment based on the cluster’s configuration.

The DNS Operator manages CoreDNS as a Kubernetes daemon set exposed as a service with a static IP. CoreDNS runs on all nodes in the cluster.

cluster-dns-operator

The etcd cluster Operator automates etcd cluster scaling, enables etcd monitoring and metrics, and simplifies disaster recovery procedures.

cluster-etcd-operator

The Ingress Operator configures and manages the OKD router.

openshift-ingress-operator

The Ingress Operator sets up the router in the openshift-ingress project and creates the deployment for the router:

The Ingress Operator uses the clusterNetwork[].cidr from the network/cluster status to determine what mode (IPv4, IPv6, or dual stack) the managed ingress controller (router) should operate in. For example, if clusterNetwork contains only a v6 cidr, then the ingress controller operate in IPv6-only mode.

In the following example, ingress controllers managed by the Ingress Operator will run in IPv4-only mode because only one cluster network exists and the network is an IPv4 cidr:

The Kubernetes API Server Operator manages and updates the Kubernetes API server deployed on top of OKD. The Operator is based on the OpenShift library-go framework and it is installed using the Cluster Version Operator (CVO).

openshift-kube-apiserver-operator

The Kubernetes Controller Manager Operator manages and updates the Kubernetes Controller Manager deployed on top of OKD. The Operator is based on OpenShift library-go framework and it is installed via the Cluster Version Operator (CVO).

It contains the following components:

By default, the Operator exposes Prometheus metrics through the metrics service.

cluster-kube-controller-manager-operator

The Kubernetes Scheduler Operator manages and updates the Kubernetes Scheduler deployed on top of OKD. The Operator is based on the OKD library-go framework and it is installed with the Cluster Version Operator (CVO).

The Kubernetes Scheduler Operator contains the following components:

By default, the Operator exposes Prometheus metrics through the metrics service.

cluster-kube-scheduler-operator

The configuration for the Kubernetes Scheduler is the result of merging:

All of these are sparse configurations, invalidated JSON snippets which are merged in order to form a valid configuration at the end.

The Machine API Operator manages the lifecycle of specific purpose custom resource definitions (CRD), controllers, and RBAC objects that extend the Kubernetes API. This declares the desired state of machines in a cluster.

machine-api-operator

The Machine Config Operator manages and applies configuration and updates of the base operating system and container runtime, including everything between the kernel and kubelet.

There are four components:

openshift-machine-config-operator

The Marketplace Operator is a conduit to bring off-cluster Operators to your cluster.

operator-marketplace

The Node Tuning Operator helps you manage node-level tuning by orchestrating the Tuned daemon. The majority of high-performance applications require some level of kernel tuning. The Node Tuning Operator provides a unified management interface to users of node-level sysctls and more flexibility to add custom tuning specified by user needs.

The Operator manages the containerized Tuned daemon for OKD as a Kubernetes daemon set. It ensures the custom tuning specification is passed to all containerized Tuned daemons running in the cluster in the format that the daemons understand. The daemons run on all nodes in the cluster, one per node.

Node-level settings applied by the containerized Tuned daemon are rolled back on an event that triggers a profile change or when the containerized Tuned daemon is terminated gracefully by receiving and handling a termination signal.

The Node Tuning Operator is part of a standard OKD installation in version 4.1 and later.

cluster-node-tuning-operator

Operator Lifecycle Manager (OLM) helps users install, update, and manage the lifecycle of Kubernetes native applications (Operators) and their associated services running across their OKD clusters. It is part of the Operator Framework, an open source toolkit designed to manage Operators in an effective, automated, and scalable way.

OLM runs by default in OKD 4.6, which aids cluster administrators in installing, upgrading, and granting access to Operators running on their cluster. The OKD web console provides management screens for cluster administrators to install Operators, as well as grant specific projects access to use the catalog of Operators available on the cluster.

For developers, a self-service experience allows provisioning and configuring instances of databases, monitoring, and big data services without having to be subject matter experts, because the Operator has that knowledge baked into it.

Operator Lifecycle Manager (OLM) is composed of two Operators: the OLM Operator and the Catalog Operator.

Each of these Operators is responsible for managing the custom resource definitions (CRDs) that are the basis for the OLM framework:

Each of these Operators is also responsible for creating the following resources:

The OLM Operator is responsible for deploying applications defined by CSV resources after the required resources specified in the CSV are present in the cluster.

The OLM Operator is not concerned with the creation of the required resources; you can choose to manually create these resources using the CLI or using the Catalog Operator. This separation of concern allows users incremental buy-in in terms of how much of the OLM framework they choose to leverage for their application.

The OLM Operator uses the following workflow:

The Catalog Operator is responsible for resolving and installing cluster service versions (CSVs) and the required resources they specify. It is also responsible for watching catalog sources for updates to packages in channels and upgrading them, automatically if desired, to the latest available versions.

To track a package in a channel, you can create a Subscription object configuring the desired package, channel, and the CatalogSource object you want to use for pulling updates. When updates are found, an appropriate InstallPlan object is written into the namespace on behalf of the user.

The Catalog Operator uses the following workflow:

The Catalog Registry stores CSVs and CRDs for creation in a cluster and stores metadata about packages and channels.

A package manifest is an entry in the Catalog Registry that associates a package identity with sets of CSVs. Within a package, channels point to a particular CSV. Because CSVs explicitly reference the CSV that they replace, a package manifest provides the Catalog Operator with all of the information that is required to update a CSV to the latest version in a channel, stepping through each intermediate version.

The OpenShift API Server Operator installs and maintains the openshift-apiserver on a cluster.

openshift-apiserver-operator

The Prometheus Operator for Kubernetes provides easy monitoring definitions for Kubernetes services and deployment and management of Prometheus instances.

Once installed, the Prometheus Operator provides the following features:

prometheus-operator

The Windows Machine Config Operator (WMCO) orchestrates the process of deploying and managing Windows workloads on a cluster. The WMCO configures Windows machines into compute nodes, enabling Windows container workloads to run in OKD clusters. This is done by creating a machine set that uses a Windows image with the Docker-formatted container runtime installed. The WMCO completes all necessary steps to configure the underlying Windows VM so that it can join the cluster as a compute node.

windows-machine-config-operator