Preparing to update a cluster with manually maintained credentials | Updating clusters

Update requirements for clusters with manually maintained credentials
Configuring the Cloud Credential Operator utility for a cluster update
Updating cloud provider resources with the Cloud Credential Operator utility
Updating cloud provider resources with manually maintained credentials
Indicating that the cluster is ready to upgrade

The Cloud Credential Operator (CCO) Upgradable status for a cluster with manually maintained credentials is False by default.

For minor releases, for example, from 4.12 to 4.13, this status prevents you from updating until you have addressed any updated permissions and annotated the CloudCredential resource to indicate that the permissions are updated as needed for the next version. This annotation changes the Upgradable status to True.
For z-stream releases, for example, from 4.13.0 to 4.13.1, no permissions are added or changed, so the update is not blocked.

Before updating a cluster with manually maintained credentials, you must accommodate any new or changed credentials in the release image for the version of OKD you are updating to.

Update requirements for clusters with manually maintained credentials

Before you update a cluster that uses manually maintained credentials with the Cloud Credential Operator (CCO), you must update the cloud provider resources for the new release.

If the cloud credential management for your cluster was configured using the CCO utility (ccoctl), use the ccoctl utility to update the resources. Clusters that were configured to use manual mode without the ccoctl utility require manual updates for the resources.

After updating the cloud provider resources, you must update the upgradeable-to annotation for the cluster to indicate that it is ready to update.

The process to update the cloud provider resources and the upgradeable-to annotation can only be completed by using command line tools.

Cloud credential configuration options and update requirements by platform type

Some platforms only support using the CCO in one mode. For clusters that are installed on those platforms, the platform type determines the credentials update requirements.

For platforms that support using the CCO in multiple modes, you must determine which mode the cluster is configured to use and take the required actions for that configuration.

Decision tree showing the possible update paths for your cluster depending on the configured CCO credentials mode.

Figure 1. Credentials update requirements by platform type

OpenStack, oVirt, and VMware vSphere

These platforms do not support using the CCO in manual mode. Clusters on these platforms handle changes in cloud provider resources automatically and do not require an update to the upgradeable-to annotation.

Administrators of clusters on these platforms should skip the manually maintained credentials section of the update process.

IBM Cloud and Nutanix

Clusters installed on these platforms are configured using the ccoctl utility.

Administrators of clusters on these platforms must take the following actions:

Configure the ccoctl utility for the new release.
Use the ccoctl utility to update the cloud provider resources.
Indicate that the cluster is ready to update with the upgradeable-to annotation.

Microsoft Azure Stack Hub

These clusters use manual mode with long-lived credentials and do not use the ccoctl utility.

Administrators of clusters on these platforms must take the following actions:

Manually update the cloud provider resources for the new release.
Indicate that the cluster is ready to update with the upgradeable-to annotation.

Amazon Web Services (AWS), global Microsoft Azure, and Google Cloud Platform (GCP)

Clusters installed on these platforms support multiple CCO modes.

The required update process depends on the mode that the cluster is configured to use. If you are not sure what mode the CCO is configured to use on your cluster, you can use the web console or the CLI to determine this information.

Additional resources

Determining the Cloud Credential Operator mode by using the web console

You can determine what mode the Cloud Credential Operator (CCO) is configured to use by using the web console.

Only Amazon Web Services (AWS), global Microsoft Azure, and Google Cloud Platform (GCP) clusters support multiple CCO modes.

Prerequisites

You have access to an OKD account with cluster administrator permissions.

Procedure

Log in to the OKD web console as a user with the cluster-admin role.
Navigate to Administration → Cluster Settings.
On the Cluster Settings page, select the Configuration tab.
Under Configuration resource, select CloudCredential.
On the CloudCredential details page, select the YAML tab.

In the YAML block, check the value of spec.credentialsMode. The following values are possible, though not all are supported on all platforms:

'': The CCO is operating in the default mode. In this configuration, the CCO operates in mint or passthrough mode, depending on the credentials provided during installation.
Mint: The CCO is operating in mint mode.
Passthrough: The CCO is operating in passthrough mode.
Manual: The CCO is operating in manual mode.

To determine the specific configuration of an AWS or GCP cluster that has a spec.credentialsMode of '', Mint, or Manual, you must investigate further.

AWS and GCP clusters support using mint mode with the root secret deleted. If the cluster is specifically configured to use mint mode or uses mint mode by default, you must determine if the root secret is present on the cluster before updating.

An AWS or GCP cluster that uses manual mode might be configured to create and manage cloud credentials from outside of the cluster using the AWS Security Token Service (STS) or GCP Workload Identity. You can determine whether your cluster uses this strategy by examining the cluster Authentication object.

AWS or GCP clusters that use mint mode only: To determine whether the cluster is operating without the root secret, navigate to Workloads → secrets and look for the root secret for your cloud provider.

Ensure that the Project dropdown is set to All Projects.

Platform secret name

AWS

aws-creds

GCP

gcp-credentials
- If you see one of these values, your cluster is using mint or passthrough mode with the root secret present.
- If you do not see these values, your cluster is using the CCO in mint mode with the root secret removed.
AWS or GCP clusters that use manual mode only: To determine whether the cluster is configured to create and manage cloud credentials from outside of the cluster, you must check the cluster Authentication object YAML values.
1. Navigate to Administration → Cluster Settings.
2. On the Cluster Settings page, select the Configuration tab.
3. Under Configuration resource, select Authentication.
4. On the Authentication details page, select the YAML tab.
5. In the YAML block, check the value of the .spec.serviceAccountIssuer parameter.
  A value that contains a URL that is associated with your cloud provider indicates that the CCO is using manual mode with AWS STS or GCP Workload Identity to create and manage cloud credentials from outside of the cluster. These clusters are configured using the ccoctl utility.
  
  An empty value ('') indicates that the cluster is using the CCO in manual mode but was not configured using the ccoctl utility.

Platform	secret name
AWS	`aws-creds`
GCP	`gcp-credentials`

Next steps

If you are updating a cluster that has the CCO operating in mint or passthrough mode and the root secret is present, you do not need to update any cloud provider resources and can continue to the next part of the update process.
If your cluster is using the CCO in mint mode with the root secret removed, you must reinstate the credential secret with the administrator-level credential before continuing to the next part of the update process.
If your cluster was configured using the CCO utility (ccoctl), you must take the following actions:
1. Configure the ccoctl utility for the new release and use it to update the cloud provider resources.
2. Update the upgradeable-to annotation to indicate that the cluster is ready to update.
If your cluster is using the CCO in manual mode but was not configured using the ccoctl utility, you must take the following actions:
1. Manually update the cloud provider resources for the new release.
2. Update the upgradeable-to annotation to indicate that the cluster is ready to update.

Additional resources

Determining the Cloud Credential Operator mode by using the CLI

You can determine what mode the Cloud Credential Operator (CCO) is configured to use by using the CLI.

Only Amazon Web Services (AWS), global Microsoft Azure, and Google Cloud Platform (GCP) clusters support multiple CCO modes.

Prerequisites

You have access to an OKD account with cluster administrator permissions.
You have installed the OpenShift CLI (oc).

Procedure

To determine the mode that the CCO is configured to use, enter the following command:

$ oc get cloudcredentials cluster \
  -o=jsonpath={.spec.credentialsMode}

The following output values are possible, though not all are supported on all platforms:

'': The CCO is operating in the default mode. In this configuration, the CCO operates in mint or passthrough mode, depending on the credentials provided during installation.
Mint: The CCO is operating in mint mode.
Passthrough: The CCO is operating in passthrough mode.
Manual: The CCO is operating in manual mode.

To determine the specific configuration of an AWS or GCP cluster that has a spec.credentialsMode of '', Mint, or Manual, you must investigate further.

AWS or GCP clusters that use mint mode only: To determine whether the cluster is operating without the root secret, run the following command:
```
$ oc get secret <secret_name> \
  -n=kube-system
```
where <secret_name> is aws-creds for AWS or gcp-credentials for GCP.

If the root secret is present, the output of this command returns information about the secret. An error indicates that the root secret is not present on the cluster.
AWS or GCP clusters that use manual mode only: To determine whether the cluster is configured to create and manage cloud credentials from outside of the cluster, run the following command:
```
$ oc get authentication cluster \
  -o jsonpath \
  --template='{ .spec.serviceAccountIssuer }'
```
This command displays the value of the .spec.serviceAccountIssuer parameter in the cluster Authentication object.
- An output of a URL that is associated with your cloud provider indicates that the CCO is using manual mode with AWS STS or GCP Workload Identity to create and manage cloud credentials from outside of the cluster. These clusters are configured using the ccoctl utility.
- An empty output indicates that the cluster is using the CCO in manual mode but was not configured using the ccoctl utility.

Next steps

If you are updating a cluster that has the CCO operating in mint or passthrough mode and the root secret is present, you do not need to update any cloud provider resources and can continue to the next part of the update process.
If your cluster is using the CCO in mint mode with the root secret removed, you must reinstate the credential secret with the administrator-level credential before continuing to the next part of the update process.
If your cluster was configured using the CCO utility (ccoctl), you must take the following actions:
1. Configure the ccoctl utility for the new release and use it to update the cloud provider resources.
2. Update the upgradeable-to annotation to indicate that the cluster is ready to update.
If your cluster is using the CCO in manual mode but was not configured using the ccoctl utility, you must take the following actions:
1. Manually update the cloud provider resources for the new release.
2. Update the upgradeable-to annotation to indicate that the cluster is ready to update.

Additional resources

Configuring the Cloud Credential Operator utility for a cluster update

To upgrade a cluster that uses the Cloud Credential Operator (CCO) in manual mode to create and manage cloud credentials from outside of the cluster, extract and prepare the CCO utility (ccoctl) binary.

The ccoctl utility is a Linux binary that must run in a Linux environment.

Prerequisites

You have access to an OKD account with cluster administrator access.
You have installed the OpenShift CLI (oc).

Your cluster was configured using the ccoctl utility to create and manage cloud credentials from outside of the cluster.

Procedure

Obtain the OKD release image by running the following command:

$ RELEASE_IMAGE=$(./openshift-install version | awk '/release image/ {print $3}')

Obtain the CCO container image from the OKD release image by running the following command:
```
$ CCO_IMAGE=$(oc adm release info --image-for='cloud-credential-operator' $RELEASE_IMAGE -a ~/.pull-secret)
```
Ensure that the architecture of the $RELEASE_IMAGE matches the architecture of the environment in which you will use the ccoctl tool.
Extract the ccoctl binary from the CCO container image within the OKD release image by running the following command:
```
$ oc image extract $CCO_IMAGE --file="/usr/bin/ccoctl" -a ~/.pull-secret
```
Change the permissions to make ccoctl executable by running the following command:
```
$ chmod 775 ccoctl
```

Verification

To verify that ccoctl is ready to use, display the help file by running the following command:

$ ccoctl --help

Output of ccoctl --help

OpenShift credentials provisioning tool

Usage:
  ccoctl [command]

Available Commands:
  alibabacloud Manage credentials objects for alibaba cloud
  aws          Manage credentials objects for AWS cloud
  gcp          Manage credentials objects for Google cloud
  help         Help about any command
  ibmcloud     Manage credentials objects for IBM Cloud
  nutanix      Manage credentials objects for Nutanix

Flags:
  -h, --help   help for ccoctl

Use "ccoctl [command] --help" for more information about a command.

Updating cloud provider resources with the Cloud Credential Operator utility

The process for upgrading an OKD cluster that was configured using the CCO utility (ccoctl) is similar to creating the cloud provider resources during installation.

By default, ccoctl creates objects in the directory in which the commands are run. To create the objects in a different directory, use the --output-dir flag. This procedure uses <path_to_ccoctl_output_dir> to refer to this directory.

On AWS clusters, some ccoctl commands make AWS API calls to create or modify AWS resources. You can use the --dry-run flag to avoid making API calls. Using this flag creates JSON files on the local file system instead. You can review and modify the JSON files and then apply them with the AWS CLI tool using the --cli-input-json parameters.

Prerequisites

Obtain the OKD release image for the version that you are upgrading to.
Extract and prepare the ccoctl binary from the release image.

Procedure

Extract the list of CredentialsRequest custom resources (CRs) from the OKD release image by running the following command:
```
$ oc adm release extract --credentials-requests \
  --cloud=<provider_type> \
  --to=<path_to_directory_with_list_of_credentials_requests>/credrequests \
  quay.io/<path_to>/ocp-release:<version>
```
where:
- <provider_type> is the value for your cloud provider. Valid values are alibabacloud, aws, gcp, ibmcloud, and nutanix.
- credrequests is the directory where the list of CredentialsRequest objects is stored. This command creates the directory if it does not exist.

For each CredentialsRequest CR in the release image, ensure that a namespace that matches the text in the spec.secretRef.namespace field exists in the cluster. This field is where the generated secrets that hold the credentials configuration are stored.

Sample AWS CredentialsRequest object

apiVersion: cloudcredential.openshift.io/v1
kind: CredentialsRequest
metadata:
  name: cloud-credential-operator-iam-ro
  namespace: openshift-cloud-credential-operator
spec:
  providerSpec:
    apiVersion: cloudcredential.openshift.io/v1
    kind: AWSProviderSpec
    statementEntries:
    - effect: Allow
      action:
      - iam:GetUser
      - iam:GetUserPolicy
      - iam:ListAccessKeys
      resource: "*"
  secretRef:
    name: cloud-credential-operator-iam-ro-creds
    namespace: openshift-cloud-credential-operator (1)

1	This field indicates the namespace which needs to exist to hold the generated secret.

The CredentialsRequest CRs for other platforms have a similar format with different platform-specific values.

For any CredentialsRequest CR for which the cluster does not already have a namespace with the name specified in spec.secretRef.namespace, create the namespace by running the following command:
```
$ oc create namespace <component_namespace>
```
Use the ccoctl tool to process all CredentialsRequest objects in the credrequests directory by running the command for your cloud provider. The following commands process CredentialsRequest objects:
- Alibaba Cloud: ccoctl alibabacloud create-ram-users
- Amazon Web Services (AWS): ccoctl aws create-iam-roles
- Google Cloud Platform (GCP): ccoctl gcp create-all
- IBM Cloud: ccoctl ibmcloud create-service-id
- Nutanix: ccoctl nutanix create-shared-secrets
Refer to the ccoctl utility instructions in the installation content for your cloud provider for important platform-specific details about the required arguments and special considerations.

For each CredentialsRequest object, ccoctl creates the required provider resources and a permissions policy as defined in each CredentialsRequest object from the OKD release image.

Apply the secrets to your cluster by running the following command:

$ ls <path_to_ccoctl_output_dir>/manifests/*-credentials.yaml | xargs -I{} oc apply -f {}

Verification

You can verify that the required provider resources and permissions policies are created by querying the cloud provider. For more information, refer to your cloud provider documentation on listing roles or service accounts.

Next steps

Update the upgradeable-to annotation to indicate that the cluster is ready to upgrade.

Additional resources

Updating cloud provider resources with manually maintained credentials

Before upgrading a cluster with manually maintained credentials, you must create any new credentials for the release image that you are upgrading to. You must also review the required permissions for existing credentials and accommodate any new permissions requirements in the new release for those components.

Procedure

Extract and examine the CredentialsRequest custom resource for the new release.

The "Manually creating IAM" section of the installation content for your cloud provider explains how to obtain and use the credentials required for your cloud.
Update the manually maintained credentials on your cluster:
- Create new secrets for any CredentialsRequest custom resources that are added by the new release image.
- If the CredentialsRequest custom resources for any existing credentials that are stored in secrets have changed permissions requirements, update the permissions as required.

If your cluster uses cluster capabilities to disable one or more optional components, delete the CredentialsRequest custom resources for any disabled components.

Example credrequests directory contents for OKD 4.12 on AWS

0000_30_machine-api-operator_00_credentials-request.yaml (1)
0000_50_cloud-credential-operator_05-iam-ro-credentialsrequest.yaml (2)
0000_50_cluster-image-registry-operator_01-registry-credentials-request.yaml (3)
0000_50_cluster-ingress-operator_00-ingress-credentials-request.yaml (4)
0000_50_cluster-network-operator_02-cncc-credentials.yaml (5)
0000_50_cluster-storage-operator_03_credentials_request_aws.yaml (6)

1	The Machine API Operator CR is required.
2	The Cloud Credential Operator CR is required.
3	The Image Registry Operator CR is required.
4	The Ingress Operator CR is required.
5	The Network Operator CR is required.
6	The Storage Operator CR is an optional component and might be disabled in your cluster.

Example credrequests directory contents for OKD 4.12 on GCP

0000_26_cloud-controller-manager-operator_16_credentialsrequest-gcp.yaml (1)
0000_30_machine-api-operator_00_credentials-request.yaml (2)
0000_50_cloud-credential-operator_05-gcp-ro-credentialsrequest.yaml (3)
0000_50_cluster-image-registry-operator_01-registry-credentials-request-gcs.yaml (4)
0000_50_cluster-ingress-operator_00-ingress-credentials-request.yaml (5)
0000_50_cluster-network-operator_02-cncc-credentials.yaml (6)
0000_50_cluster-storage-operator_03_credentials_request_gcp.yaml (7)

1	The Cloud Controller Manager Operator CR is required.
2	The Machine API Operator CR is required.
3	The Cloud Credential Operator CR is required.
4	The Image Registry Operator CR is required.
5	The Ingress Operator CR is required.
6	The Network Operator CR is required.
7	The Storage Operator CR is an optional component and might be disabled in your cluster.

Next steps

Update the upgradeable-to annotation to indicate that the cluster is ready to upgrade.

Additional resources

Indicating that the cluster is ready to upgrade

The Cloud Credential Operator (CCO) Upgradable status for a cluster with manually maintained credentials is False by default.

Prerequisites

For the release image that you are upgrading to, you have processed any new credentials manually or by using the Cloud Credential Operator utility (ccoctl).
You have installed the OpenShift CLI (oc).

Procedure

Log in to oc on the cluster as a user with the cluster-admin role.
Edit the CloudCredential resource to add an upgradeable-to annotation within the metadata field by running the following command:
```
$ oc edit cloudcredential cluster
```
Text to add
```
...
  metadata:
    annotations:
      cloudcredential.openshift.io/upgradeable-to: <version_number>
...
```
Where <version_number> is the version that you are upgrading to, in the format x.y.z. For example, use 4.12.2 for OKD 4.12.2.

It may take several minutes after adding the annotation for the upgradeable status to change.

Verification

In the Administrator perspective of the web console, navigate to Administration → Cluster Settings.
To view the CCO status details, click cloud-credential in the Cluster Operators list.
- If the Upgradeable status in the Conditions section is False, verify that the upgradeable-to annotation is free of typographical errors.
When the Upgradeable status in the Conditions section is True, begin the OKD upgrade.