Red Hat OpenShift VPC cluster on IBM Cloud module

Use this module to provision an IBM Cloud Red Hat OpenShift cluster on VPC Gen2. The module either creates the required Cloud Object Storage instance or uses an existing instance. The module also supports optionally passing a key management configuration for secret encryption and boot volume encryption.

Optionally, the module supports advanced security group management for the worker nodes, VPE, and load balancer associated with the cluster. This feature allows you to configure security groups for the cluster's worker nodes, VPE, and load balancer.

❗ Important: You can't update Red Hat OpenShift cluster nodes by using this module. The Terraform logic ignores updates to prevent possible destructive changes.

Before you begin

• Ensure that you have an up-to-date version of the IBM Cloud CLI.
• Ensure that you have an up-to-date version of the IBM Cloud Kubernetes service CLI.
• Ensure that you have an up-to-date version of the IBM Cloud VPC Infrastructure service CLI. Only required if providing additional security groups with the var.additional_lb_security_group_ids.
• Ensure that you have an up-to-date version of the jq
• Ensure that you have an up-to-date version of the kubectl

Overview

• terraform-ibm-base-ocp-vpc
• Submodules
  • fscloud
• Examples
  • 2 MZR clusters in same VPC example
  • Advanced example (mzr, auto-scale, kms, taints)
  • Attaching custom security groups
  • Basic single zone cluster with allowed outbound traffic
  • Cluster security group rules example
  • Cross account KMS encryption example
  • Financial Services compliant example
• Contributing

terraform-ibm-base-ocp-vpc

Usage

module "ocp_base" {
  source               = "terraform-ibm-modules/base-ocp-vpc/ibm"
  version              = "X.X.X" # Replace "X.X.X" with a release version to lock into a specific release
  cluster_name         = "example-cluster-name"
  resource_group_id    = "xxXXxxXXxXxXXXXxxXxxxXXXXxXXXXX"
  region               = "us-south"
  force_delete_storage = true
  vpc_id               = "79cxxxx-xxxx-xxxx-xxxx-xxxxxXX8667"
  # obtain the below values from the targeted VPC and adjust to the number of zones, subnets, subnet name, cidr_block, id, zone
  vpc_subnets          = {
    zone-1    = [
        {
            cidr_block = "192.168.32.0/22"
            id         = "0717-afc29fbb-0dbe-493a-a5b9-f3c5899cb8b9"
            zone       = "us-south-1"
        },
        {
            cidr_block = "192.168.36.0/22"
            id         = "0727-d65c1eda-9e38-4200-8452-cb8ff5bb3140"
            zone       = "us-south-2"
        },
        {
            cidr_block = "192.168.40.0/22"
            id         = "0737-9a823cd3-16bf-4ba4-a429-9e1fc7db74b8"
            zone       = "us-south-3"
        }
    ]
    zone-2 = [
        {
            cidr_block = "192.168.0.0/22"
            id         = "0717-846b9490-34ae-4a6c-8288-28112dca1ba3"
            zone       = "us-south-1"
        },
        {
            cidr_block = "192.168.4.0/22"
            id         = "0727-ef8db7f6-ffa5-4d8b-a317-4631741a45ee"
            zone       = "us-south-2"
        },
        {
            cidr_block = "192.168.8.0/22"
            id         = "0737-c9a6d871-d95b-4914-abf5-82c22f4161d1"
            zone       = "us-south-3"
        }
    ]
    zone-3 = [
        {
            cidr_block = "192.168.16.0/22"
            id         = "0717-d46e227c-89d4-4b02-9008-d03907a275b6"
            zone       = "us-south-1"
        },
        {
            cidr_block = "192.168.20.0/22"
            id         = "0727-93b1edcb-966c-4517-a7af-6ac63cd93adf"
            zone       = "us-south-2"
        },
        {
            cidr_block = "192.168.24.0/22"
            id         = "0737-807ec4f1-4d84-484e-b2f4-62dd5e431065"
            zone       = "us-south-3"
        }
    ]
  }
  worker_pools         = [
    {
      subnet_prefix    = "default"
      pool_name        = "default"
      machine_type     = "bx2.4x16"
      workers_per_zone = 2
      operating_system = "REDHAT_8_64"
    }
  ]
}

Secure by default cluster settings

In OCP version 4.15, outbound traffic is disabled by default. Learn more.

There is a provision to toggle outbound traffic by using the modules' disable_outbound_traffic_protection input. Refer Managing outbound traffic protection in VPC clusters.

Default Worker Pool management

You can manage the default worker pool using Terraform, and make changes to it through this module. This option is enabled by default. Under the hood, the default worker pool is imported as a ibm_container_vpc_worker_pool resource. Advanced users may opt-out of this option by setting import_default_worker_pool_on_create parameter to false. For most use cases it is recommended to keep this variable to true.

Important Considerations for Terraform and Default Worker Pool

Terraform Destroy

When using the default behavior of handling the default worker pool as a stand-alone ibm_container_vpc_worker_pool, you must manually remove the default worker pool from the Terraform state before running a terraform destroy command on the module. This is due to a known limitation in IBM Cloud.

Terraform CLI Example

For a cluster with 1 or more worker pools, follow these steps:

      $ terraform state list | grep ibm_container_vpc_worker_pool | grep default
        > module.ocp_base.data.ibm_container_vpc_worker_pool.all_pools["default"]
        > module.ocp_base.ibm_container_vpc_worker_pool.pool["default"]

      $ terraform state rm "module.ocp_base.ibm_container_vpc_worker_pool.pool[\"default\"]"

Schematics Example: For a cluster with 1 or more worker pools, follow these steps:

        $ ibmcloud schematics workspace state rm --id <workspace_id> --address "module.ocp_base.ibm_container_vpc_worker_pool.pool[\"default\"]"

Changes Requiring Re-creation of Default Worker Pool

If you need to make changes to the default worker pool that require its re-creation (e.g., changing the worker node operating_system), you must set the allow_default_worker_pool_replacement variable to true, perform the apply, and then set it back to false in the code before the subsequent apply. This is only necessary for changes that require the recreation the entire default pool and is not needed for scenarios that does not require recreating the worker pool such as changing the number of workers in the default worker pool.

This approach is due to a limitation in the Terraform provider that may be lifted in the future.

Advanced security group options

The Terraform module provides options to attach additional security groups to the worker nodes, VPE, and load balancer associated with the cluster.

The custom_sg example demonstrates how to use these capabilities.

See the IBM Cloud documentation on this topic here

Tip: The terraform-ibm-security-groups module can be used to create security groups and rules.

Worker nodes

• Additional security groups can be specified at cluster creation time. These security groups are attached to all worker nodes of the cluster, including additional worker nodes/pools added after the creation of the cluster. See the variable custom_security_group_ids.
• Additional security groups can be specified for specific worker pools. These security groups only apply to the worker pool. See the field additional_security_group_ids in the variable worker_pools.

In all cases, note that:

• The default VPC security is no longer attached to the worker nodes.
• You can opt-out of attaching the IBM-managed cluster security group (named kube-) through the flag attach_ibm_managed_security_group.
• It is impossible to change the security groups associated with a cluster after the creation of that cluster.

VPEs (Virtual Private Endpoints)

• The IBM Cloud OCP stack creates VPEs by default. Prior to version 4.14, a VPE to the master is created. From version 4.14, VPEs to the master, container registry, and IBM Cloud kube APIs are created.
• You can attach additional security groups through the additional_vpe_security_group_ids variable.
• The default IBM-managed security group is attached to those VPEs in all cases.

Load balancers

• The IBM Cloud OCP stack manages the lifecycle of VPC Loadbalancers for your cluster. See the LoadBalancer section in the Understanding options for exposing apps.
• By default, one load balancer is created at cluster creation for the default cluster ingress.
• You can attach additional security groups using the additional_lb_security_group_ids variable. This set of security groups is attached to all loadbalancers managed by the cluster.
• Important: If additional load balancers are added after creating the cluster, for example, by exposing a Kubernetes service of type LoadBalancer, update the number_of_lbs variable and re-run this module to attach the extra security groups to the newly created load balancer.
• The default IBM-managed security group is attached to the LBs in all cases.

Troubleshooting

New kube_version message

• When you run a terraform plan command, you might get a message about a new version of Kubernetes, as in the following example:

  kube_version = "4.12.16_openshift" -> "4.12.20_openshift"
  
  Unless you have made equivalent changes to your configuration, or ignored the relevant attributes using ignore_changes, the following plan may include actions to undo or respond to these changes.

  A new version is detected because the Kubernetes master node is updated outside of Terraform, and the Terraform state is out of date with that version.

  The Kubernetes version is ignored in the module code, so the infrastructure will not be modified. The message identifies that drift exists in the versions, and after running the terraform apply command, the state will be refreshed.

Required IAM access policies

You need the following permissions to run this module.

• Account Management
  • All Identity and Access Enabled service
    • Viewer platform access
  • All Resource Groups service
    • Viewer platform access
• IAM Services
  • Cloud Object Storage service
    • Editor platform access
    • Manager service access
  • Kubernetes service
    • Administrator platform access
    • Manager service access
  • VPC Infrastructure service
    • Administrator platform access
    • Manager service access

Optionally, you need the following permissions to attach Access Management tags to resources in this module.

• IAM Services
  • Tagging service
    • Administrator platform access

Note

• One worker pool should always be named as default. Refer issue 2849 for further details.

Requirements

Name	Version
terraform	>= 1.3.0
ibm	>= 1.70.0, < 2.0.0
kubernetes	>= 2.16.1, < 3.0.0
null	>= 3.2.1, < 4.0.0

Modules

Name	Source	Version
attach_sg_to_api_vpe	terraform-ibm-modules/security-group/ibm	2.6.2
attach_sg_to_lb	terraform-ibm-modules/security-group/ibm	2.6.2
attach_sg_to_master_vpe	terraform-ibm-modules/security-group/ibm	2.6.2
attach_sg_to_registry_vpe	terraform-ibm-modules/security-group/ibm	2.6.2
cbr_rule	terraform-ibm-modules/cbr/ibm//modules/cbr-rule-module	1.29.0
cos_instance	terraform-ibm-modules/cos/ibm	8.15.10

Resources

Name	Type
ibm_container_addons.addons	resource
ibm_container_vpc_cluster.autoscaling_cluster	resource
ibm_container_vpc_cluster.cluster	resource
ibm_container_vpc_worker_pool.autoscaling_pool	resource
ibm_container_vpc_worker_pool.pool	resource
ibm_resource_tag.cluster_access_tag	resource
ibm_resource_tag.cos_access_tag	resource
kubernetes_config_map_v1_data.set_autoscaling	resource
null_resource.config_map_status	resource
null_resource.confirm_lb_active	resource
null_resource.confirm_network_healthy	resource
null_resource.reset_api_key	resource
ibm_container_addons.existing_addons	data source
ibm_container_cluster_config.cluster_config	data source
ibm_container_cluster_versions.cluster_versions	data source
ibm_container_vpc_worker_pool.all_pools	data source
ibm_iam_account_settings.iam_account_settings	data source
ibm_iam_auth_token.reset_api_key_tokendata	data source
ibm_iam_auth_token.tokendata	data source
ibm_is_lbs.all_lbs	data source
ibm_is_virtual_endpoint_gateway.api_vpe	data source
ibm_is_virtual_endpoint_gateway.master_vpe	data source
ibm_is_virtual_endpoint_gateway.registry_vpe	data source

Inputs

Name	Description	Type	Default	Required
access_tags	A list of access tags to apply to the resources created by the module, see https://cloud.ibm.com/docs/account?topic=account-access-tags-tutorial for more details	list(string)	[]	no
additional_lb_security_group_ids	Additional security groups to add to the load balancers associated with the cluster. Ensure that the number_of_lbs is set to the number of LBs associated with the cluster. This comes in addition to the IBM maintained security group.	list(string)	[]	no
additional_vpe_security_group_ids	Additional security groups to add to all existing load balancers. This comes in addition to the IBM maintained security group.	object({ master = optional(list(string), []) registry = optional(list(string), []) api = optional(list(string), []) })	{}	no
addons	Map of OCP cluster add-on versions to install (NOTE: The 'vpc-block-csi-driver' add-on is installed by default for VPC clusters and 'ibm-storage-operator' is installed by default in OCP 4.15 and later, however you can explicitly specify it here if you wish to choose a later version than the default one). For full list of all supported add-ons and versions, see https://cloud.ibm.com/docs/containers?topic=containers-supported-cluster-addon-versions	object({ debug-tool = optional(string) image-key-synchronizer = optional(string) openshift-data-foundation = optional(string) vpc-file-csi-driver = optional(string) static-route = optional(string) cluster-autoscaler = optional(string) vpc-block-csi-driver = optional(string) ibm-storage-operator = optional(string) })	{}	no
allow_default_worker_pool_replacement	(Advanced users) Set to true to allow the module to recreate a default worker pool. Only use in the case where you are getting an error indicating that the default worker pool cannot be replaced on apply. Once the default worker pool is handled as a stand-alone ibm_container_vpc_worker_pool, if you wish to make any change to the default worker pool which requires the re-creation of the default pool set this variable to true.	bool	false	no
attach_ibm_managed_security_group	Specify whether to attach the IBM-defined default security group (whose name is kube-) to all worker nodes. Only applicable if custom_security_group_ids is set.	bool	true	no
cbr_rules	The list of context-based restriction rules to create.	list(object({ description = string account_id = string rule_contexts = list(object({ attributes = optional(list(object({ name = string value = string }))) })) enforcement_mode = string tags = optional(list(object({ name = string value = string })), []) operations = optional(list(object({ api_types = list(object({ api_type_id = string })) }))) }))	[]	no
cluster_config_endpoint_type	Specify which type of endpoint to use for for cluster config access: 'default', 'private', 'vpe', 'link'. 'default' value will use the default endpoint of the cluster.	string	"default"	no
cluster_name	The name that will be assigned to the provisioned cluster	string	n/a	yes
cluster_ready_when	The cluster is ready when one of the following: MasterNodeReady (not recommended), OneWorkerNodeReady, Normal, IngressReady	string	"IngressReady"	no
cos_name	Name of the COS instance to provision for OpenShift internal registry storage. New instance only provisioned if 'enable_registry_storage' is true and 'use_existing_cos' is false. Default: '<cluster_name>_cos'	string	null	no
custom_security_group_ids	Security groups to add to all worker nodes. This comes in addition to the IBM maintained security group if attach_ibm_managed_security_group is set to true. If this variable is set, the default VPC security group is NOT assigned to the worker nodes.	list(string)	null	no
disable_outbound_traffic_protection	Whether to allow public outbound access from the cluster workers. This is only applicable for ocp_version 4.15	bool	false	no
disable_public_endpoint	Whether access to the public service endpoint is disabled when the cluster is created. Does not affect existing clusters. You can't disable a public endpoint on an existing cluster, so you can't convert a public cluster to a private cluster. To change a public endpoint to private, create another cluster with this input set to true.	bool	false	no
enable_registry_storage	Set to true to enable IBM Cloud Object Storage for the Red Hat OpenShift internal image registry. Set to false only for new cluster deployments in an account that is allowlisted for this feature.	bool	true	no
existing_cos_id	The COS id of an already existing COS instance to use for OpenShift internal registry storage. Only required if 'enable_registry_storage' and 'use_existing_cos' are true	string	null	no
force_delete_storage	Flag indicating whether or not to delete attached storage when destroying the cluster - Default: false	bool	false	no
ignore_worker_pool_size_changes	Enable if using worker autoscaling. Stops Terraform managing worker count	bool	false	no
import_default_worker_pool_on_create	(Advanced users) Whether to handle the default worker pool as a stand-alone ibm_container_vpc_worker_pool resource on cluster creation. Only set to false if you understand the implications of managing the default worker pool as part of the cluster resource. Set to true to import the default worker pool as a separate resource. Set to false to manage the default worker pool as part of the cluster resource.	bool	true	no
kms_config	Use to attach a KMS instance to the cluster. If account_id is not provided, defaults to the account in use.	object({ crk_id = string instance_id = string private_endpoint = optional(bool, true) # defaults to true account_id = optional(string) # To attach KMS instance from another account wait_for_apply = optional(bool, true) # defaults to true so terraform will wait until the KMS is applied to the master, ready and deployed })	null	no
manage_all_addons	Instructs Terraform to manage all cluster addons, even if addons were installed outside of the module. If set to 'true' this module will destroy any addons that were installed by other sources.	bool	false	no
number_of_lbs	The number of LBs to associated the additional_lb_security_group_names security group with.	number	1	no
ocp_entitlement	Value that is applied to the entitlements for OCP cluster provisioning	string	null	no
ocp_version	The version of the OpenShift cluster that should be provisioned (format 4.x). If no value is specified, the current default version is used. You can also specify default. This input is used only during initial cluster provisioning and is ignored for updates. To prevent possible destructive changes, update the cluster version outside of Terraform.	string	null	no
pod_subnet_cidr	Specify a custom subnet CIDR to provide private IP addresses for pods. The subnet must have a CIDR of at least /23 or larger. Default value is 172.30.0.0/16 when the variable is set to null.	string	null	no
region	The IBM Cloud region where the cluster will be provisioned.	string	n/a	yes
resource_group_id	The Id of an existing IBM Cloud resource group where the cluster will be grouped.	string	n/a	yes
service_subnet_cidr	Specify a custom subnet CIDR to provide private IP addresses for services. The subnet must be at least /24 or larger. Default value is 172.21.0.0/16 when the variable is set to null.	string	null	no
tags	Metadata labels describing this cluster deployment, i.e. test	list(string)	[]	no
use_existing_cos	Flag indicating whether or not to use an existing COS instance for OpenShift internal registry storage. Only applicable if 'enable_registry_storage' is true	bool	false	no
use_private_endpoint	Set this to true to force all api calls to use the IBM Cloud private endpoints.	bool	false	no
verify_worker_network_readiness	By setting this to true, a script will run kubectl commands to verify that all worker nodes can communicate successfully with the master. If the runtime does not have access to the kube cluster to run kubectl commands, this should be set to false.	bool	true	no
vpc_id	Id of the VPC instance where this cluster will be provisioned	string	n/a	yes
vpc_subnets	Metadata that describes the VPC's subnets. Obtain this information from the VPC where this cluster will be created	map(list(object({ id = string zone = string cidr_block = string })))	n/a	yes
worker_pools	List of worker pools	list(object({ subnet_prefix = optional(string) vpc_subnets = optional(list(object({ id = string zone = string cidr_block = string }))) pool_name = string machine_type = string workers_per_zone = number resource_group_id = optional(string) operating_system = string labels = optional(map(string)) minSize = optional(number) secondary_storage = optional(string) maxSize = optional(number) enableAutoscaling = optional(bool) boot_volume_encryption_kms_config = optional(object({ crk = string kms_instance_id = string kms_account_id = optional(string) })) additional_security_group_ids = optional(list(string)) }))	n/a	yes
worker_pools_taints	Optional, Map of lists containing node taints by node-pool name	map(list(object({ key = string, value = string, effect = string })))	null	no

Outputs

Name	Description
api_vpe	Info about the api VPE, if it exists. For more info about schema, see https://registry.terraform.io/providers/IBM-Cloud/ibm/latest/docs/data-sources/is_virtual_endpoint_gateway
cluster_crn	CRN for the created cluster
cluster_id	ID of cluster created
cluster_name	Name of the created cluster
cos_crn	CRN of the COS instance
ingress_hostname	The hostname that was assigned to your Ingress subdomain.
kms_config	KMS configuration details
master_status	The status of the Kubernetes master.
master_url	The URL of the Kubernetes master.
master_vpe	Info about the master, or default, VPE. For more info about schema, see https://registry.terraform.io/providers/IBM-Cloud/ibm/latest/docs/data-sources/is_virtual_endpoint_gateway
ocp_version	Openshift Version of the cluster
operating_system	The operating system of the workers in the default worker pool.
private_service_endpoint_url	Private service endpoint URL
public_service_endpoint_url	Public service endpoint URL
region	Region cluster is deployed in
registry_vpe	Info about the registry VPE, if it exists. For more info about schema, see https://registry.terraform.io/providers/IBM-Cloud/ibm/latest/docs/data-sources/is_virtual_endpoint_gateway
resource_group_id	Resource group ID the cluster is deployed in
vpc_id	ID of the clusters VPC
vpe_url	The virtual private endpoint URL of the Kubernetes cluster.
workerpools	Worker pools created

Contributing

You can report issues and request features for this module in GitHub issues in the module repo. See Report an issue or request a feature.

To set up your local development environment, see Local development setup in the project documentation.