OpenShift 4 UPI on Google Cloud

This terraform implementation will deploy OpenShift 4.x into a Google Cloud Platform (GCP) VPC, with two subnets for controlplane and worker nodes. Traffic to the master nodes is handled via a pair of loadbalancers, one for internal traffic and another for external API traffic. Application loadbalancing is handled by a third loadbalancer that talks to the router pods on the infra or worker nodes. Worker, Infra and Master nodes are deployed across 3 Availability Zones

Prerequisites

1. Enable Service APIs
2. Configure DNS
3. Create GCP Service Account with proper IAM roles

Minimal TFVARS file

gcp_project_id      = "hsbc-261614"
gcp_region          = "us-central1"
cluster_name        = "ocp42"

# From Prereq. Step #2
gcp_public_dns_zone_name = "gcp-ncolon-xyz"
base_domain              = "gcp.ncolon.xyz"

# From Prereq. Step #3
gcp_service_account = "credentials.json"

# Your openshift installion pull secret
openshift_pull_secret = "openshift_pull_secret.json"

Customizable Variables

Variable	Description	Default	Type
gcp_project_id	The target GCP project for the cluster.	-	string
gcp_service_account	Path to JSON file with details for the GCP APIs service account (from Prereq Step #3)	-	string
gcp_region	he target GCP region for the cluster	-	string
gcp_extra_labels	Extra GCP labels to be applied to created resources	{}	map
cluster_name	Cluster Identifier	-	string
openshift_master_count	Number of master nodes to deploy	3	string
openshift_worker_count	Number of worker nodes to deploy	3	string
openshift_infra_count	Number of infra nodes to deploy	0	string
machine_cidr	CIDR for OpenShift VNET	10.0.0.0/16	string
base_domain	DNS name for your deployment	-	string
gcp_public_dns_zone_name	The name of the public DNS zone to use for this cluster (from Prereq Step #2)	-	string
gcp_bootstrap_instance_type	Size of bootstrap VM	n1-standard-4	string
gcp_master_instance_type	Size of master node VMs	n1-standard-4	string
gcp_infra_instance_type	Size of infra node VMs	n1-standard-4	string
gcp_worker_instance_type	Sizs of worker node VMs	n1-standard-4	string
openshift_cluster_network_cidr	CIDR for Kubernetes pods	10.128.0.0/14	string
openshift_cluster_network_host_prefix	Detemines the number of pods a node can host. 23 gives you 510 pods per node.	23	string
openshift_service_network_cidr	CIDR for Kubernetes services	172.30.0.0/16	string
openshift_pull_secret	path to filename that holds your OpenShift pull-secret	-	string
gcp_master_os_disk_size	Size of master node root volume	1024	string
gcp_worker_os_disk_size	Size of worker node root volume	128	string
gcp_infra_os_disk_size	Size of infra node root volume	128	string
gcp_image_uri	URL of the CoreOS image. Can be found here	URL	string
openshift_version	Version of OpenShift to deploy.	latest	string
gcp_bootstrap_enabled	Setting this to false allows the bootstrap resources to be disabled	true	bool
gcp_bootstrap_lb	Setting this to false allows the bootstrap resources to be removed from the cluster load balancers	true	bool
airgapped	Configuration for an AirGapped environment	AirGapped default is false	map

Deploy with Terraform

1. Clone github repository

git clone https://github.com/ibm-cloud-architecture/terraform-openshift4-gcp.git

2. Create your terraform.tfvars file

3. Deploy with terraform

$ terraform init
$ terraform plan
$ terraform apply

4. Destroy bootstrap node

$ TF_VAR_gcp_bootstrap_enabled=false terraform apply

5. To access your cluster

 $ export KUBECONFIG=$PWD/installer-files/auth/kubeconfig
 $ oc get nodes
NAME                                                         STATUS   ROLES          AGE   VERSION
ocp42-a26ek-master-0.us-central1-a.c.hsbc-261614.internal    Ready    master         95m   v1.14.6+31a56cf75
ocp42-a26ek-master-1.us-central1-b.c.hsbc-261614.internal    Ready    master         95m   v1.14.6+31a56cf75
ocp42-a26ek-master-2.us-central1-c.c.hsbc-261614.internal    Ready    master         95m   v1.14.6+31a56cf75
ocp42-a26ek-w-a-xrgvb.us-central1-a.c.hsbc-261614.internal   Ready    worker         91m   v1.14.6+31a56cf75
ocp42-a26ek-w-b-h72ss.us-central1-b.c.hsbc-261614.internal   Ready    worker         90m   v1.14.6+31a56cf75
ocp42-a26ek-w-c-4x64c.us-central1-c.c.hsbc-261614.internal   Ready    worker         90m   v1.14.6+31a56cf75

Infra and Worker Node Deployment

Deployment of Openshift Worker and Infra nodes is handled by the machine-operator-api cluster operator.

$ oc get machineset -n openshift-machine-api
NAME              DESIRED   CURRENT   READY   AVAILABLE   AGE
ocp42-a26ek-w-a   1         1         1       1           91m
ocp42-a26ek-w-b   1         1         1       1           91m
ocp42-a26ek-w-c   1         1         1       1           91m

$ oc get machines -n openshift-machine-api
NAME                    STATE     TYPE            REGION        ZONE            AGE
ocp42-a26ek-w-a-xrgvb   RUNNING   n1-standard-4   us-central1   us-central1-a   91m
ocp42-a26ek-w-b-h72ss   RUNNING   n1-standard-4   us-central1   us-central1-b   91m
ocp42-a26ek-w-c-4x64c   RUNNING   n1-standard-4   us-central1   us-central1-c   91m

If openshift_infra_count > 0 the infra nodes will host the router/ingress pods, all the monitoring infrastrucutre, and the image registry.