Use Hosted Control Planes for ROSA to speed up cluster creation

.github/actions/aurora-delete-database/action.yaml

@@ -18,4 +18,4 @@ runs:
       working-directory: provision/aws/rds
       env:
         AURORA_CLUSTER: ${{ inputs.name }}
-        AWS_REGION: ${{ inputs.region }}
+        AURORA_REGION: ${{ inputs.region }}

.github/workflows/rosa-cluster-create.yml

@@ -8,7 +8,6 @@ on:
         type: string
       computeMachineType:
         description: 'Instance type for the compute nodes'
-        default: m5.xlarge
         type: string
       multiAz:
         description: 'Deploy to multiple availability zones in the region'
@@ -20,7 +19,6 @@ on:
         type: string
       replicas:
         description: 'Number of worker nodes to provision'
-        default: '2'
         type: string
       region:
         description: 'The AWS region to create the cluster in. Defaults to "vars.AWS_DEFAULT_REGION" if omitted.'
@@ -33,23 +31,10 @@ on:
         type: string
       computeMachineType:
         description: 'Instance type for the compute nodes'
-        required: true
-        default: m5.xlarge
-        type: string
-      multiAz:
-        description: 'Deploy to multiple availability zones in the region'
-        required: true
-        default: false
-        type: boolean
-      availabilityZones:
-        description: 'Availability zones to deploy to'
-        required: false
-        default: ''
         type: string
       replicas:
         description: 'Number of worker nodes to provision'
         required: true
-        default: '2'
         type: string
       region:
         description: 'The AWS region to create the cluster in. Defaults to "vars.AWS_DEFAULT_REGION" if omitted.'
@@ -76,16 +61,20 @@ jobs:
           aws-default-region: ${{ vars.AWS_DEFAULT_REGION }}
           rosa-token: ${{ secrets.ROSA_TOKEN }}
 
+      - name: Setup OpenTofu
+        uses: opentofu/setup-opentofu@v1
+        with:
+          tofu_wrapper: false
+
       - name: Create ROSA Cluster
         run: ./rosa_create_cluster.sh
         working-directory: provision/aws
         env:
           VERSION: ${{ env.OPENSHIFT_VERSION }}
           CLUSTER_NAME: ${{ inputs.clusterName || format('gh-{0}', github.repository_owner) }}
           COMPUTE_MACHINE_TYPE: ${{ inputs.computeMachineType }}
-          MULTI_AZ: ${{ inputs.multiAz }}
-          AVAILABILITY_ZONES: ${{ inputs.availabilityZones }}
           REPLICAS: ${{ inputs.replicas }}
+          TF_VAR_rhcs_token: ${{ secrets.ROSA_TOKEN }}
 
       - name: Archive ROSA logs
         uses: actions/upload-artifact@v4

.github/workflows/rosa-cluster-delete.yml

@@ -45,19 +45,27 @@ jobs:
         with:
           clusterName: ${{ inputs.clusterName || format('gh-{0}', github.repository_owner) }}
 
+      - name: Setup OpenTofu
+        uses: opentofu/setup-opentofu@v1
+        with:
+          tofu_wrapper: false
+
       - name: Delete a ROSA Cluster
         if: ${{ inputs.deleteAll == 'no' }}
         shell: bash
         run: ./rosa_delete_cluster.sh
         working-directory: provision/aws
         env:
           CLUSTER_NAME: ${{ inputs.clusterName || format('gh-{0}', github.repository_owner) }}
+          TF_VAR_rhcs_token: ${{ secrets.ROSA_TOKEN }}
 
       - name: Delete all ROSA Clusters
         if: ${{ inputs.deleteAll == 'yes' }}
         shell: bash
         run: ./rosa_cluster_reaper.sh
         working-directory: provision/aws
+        env:
+          TF_VAR_rhcs_token: ${{ secrets.ROSA_TOKEN }}
 
       - name: Archive ROSA logs
         uses: actions/upload-artifact@v4

.github/workflows/rosa-multi-az-cluster-create.yml

@@ -86,8 +86,8 @@ jobs:
 
       - name: Scale ROSA clusters
         run: |
-          rosa edit machinepool -c ${{ env.CLUSTER_PREFIX }}-a --min-replicas 3 scaling
-          rosa edit machinepool -c ${{ env.CLUSTER_PREFIX }}-b --min-replicas 3 scaling
+          rosa edit machinepool -c ${{ env.CLUSTER_PREFIX }}-a --min-replicas 3 --max-replicas 10 scaling
+          rosa edit machinepool -c ${{ env.CLUSTER_PREFIX }}-b --min-replicas 3 --max-replicas 10 scaling
 
       - name: Setup Go Task
         uses: ./.github/actions/task-setup

.gitignore

@@ -96,3 +96,7 @@ quarkus/data/*.db
 # Horreum #
 ###########
 provision/environment_data.json
+
+# OpenTofu / Terraform
+**/*.tfstate*
+**/*.terraform*

doc/kubernetes/modules/ROOT/pages/prerequisite/prerequisite-rosa.adoc

@@ -14,6 +14,7 @@ See <<aws-efs-as-readwritemany-storage>> for more information.
 == Prerequisites
 
 . xref:prerequisite/prerequisite-awscli.adoc[]
+. [Install OpenTofu](https://opentofu.org/docs/intro/install/)
 . Perform the steps outlined in the https://console.redhat.com/openshift/create/rosa/getstarted[ROSA installation guide]:
 .. Enable ROSA Service in AWS account
 .. Download and install the ROSA command line tool
@@ -47,8 +48,6 @@ If no `ADMIN_PASSWORD` is provided in the configuration, it reads it from the AW
 `VERSION`:: OpenShift cluster version.
 `REGION`:: AWS region where the cluster should run.
 `COMPUTE_MACHINE_TYPE`:: https://aws.amazon.com/ec2/instance-types/[AWS instance type] for the default OpenShift worker machine pool.
-`MULTI_AZ`:: Boolean parameter to indicate whether the OpenShift cluster should span many Availability Zones within the selected region.
-`AVAILABILITY_ZONES`:: Comma separated list of Availability Zones to use for the cluster. For example, `eu-central-1a,eu-central-1b`.
 `REPLICAS`:: Number of worker nodes.
 If multi-AZ installation is selected, then this needs to be a multiple of the number of AZs available in the region.
 For example, if the region has 3 AZs, then replicas need to be set to some multiple of 3.

provision/aws/efs/.gitignore

@@ -1,2 +1,3 @@
 manifests
 ccoctl
+iam-trust.json

provision/aws/efs/iam-policy.json

@@ -0,0 +1,44 @@
+{
+  "Version": "2012-10-17",
+  "Statement": [
+    {
+      "Effect": "Allow",
+      "Action": [
+        "elasticfilesystem:DescribeMountTargets",
+        "elasticfilesystem:DescribeAccessPoints",
+        "elasticfilesystem:DescribeFileSystems",
+        "elasticfilesystem:ClientMount",
+        "elasticfilesystem:ClientWrite",
+        "elasticfilesystem:CreateTags",
+        "elasticfilesystem:CreateMountTarget",
+        "elasticfilesystem:DeleteMountTarget",
+        "elasticfilesystem:DeleteTags",
+        "elasticfilesystem:TagResource",
+        "elasticfilesystem:UntagResource"
+      ],
+      "Resource": "*"
+    },
+    {
+      "Effect": "Allow",
+      "Action": [
+        "elasticfilesystem:CreateAccessPoint"
+      ],
+      "Resource": "*",
+      "Condition": {
+        "StringLike": {
+          "aws:RequestTag/efs.csi.aws.com/cluster": "true"
+        }
+      }
+    },
+    {
+      "Effect": "Allow",
+      "Action": "elasticfilesystem:DeleteAccessPoint",
+      "Resource": "*",
+      "Condition": {
+        "StringEquals": {
+          "aws:ResourceTag/efs.csi.aws.com/cluster": "true"
+        }
+      }
+    }
+  ]
+}

provision/aws/rds/aurora_create_peering_connection.sh

@@ -64,7 +64,7 @@ aws ec2 accept-vpc-peering-connection \
 
 # Update the ROSA Cluster VPC's Route Table
 ROSA_PUBLIC_ROUTE_TABLE_ID=$(aws ec2 describe-route-tables \
-  --filters "Name=vpc-id,Values=${ROSA_VPC}" "Name=association.main,Values=true" \
+  --filters "Name=vpc-id,Values=${ROSA_VPC}" "Name=tag:Name,Values=*public*" \
   --query "RouteTables[*].RouteTableId" \
   --output text
 )

provision/aws/rosa_create_cluster.sh

@@ -1,100 +1,74 @@
 #!/usr/bin/env bash
 set -e
 
+if [[ "$RUNNER_DEBUG" == "1" ]]; then
+  set -x
+fi
+
 if [ -f ./.env ]; then
   source ./.env
 fi
 
+function requiredEnv() {
+  for ENV in $@; do
+      if [ -z "${!ENV}" ]; then
+        echo "${ENV} variable must be set"
+        exit 1
+      fi
+  done
+}
+
+SCRIPT_DIR=$( cd -- "$( dirname -- "${BASH_SOURCE[0]}" )" &> /dev/null && pwd )
+
 AWS_ACCOUNT=${AWS_ACCOUNT:-$(aws sts get-caller-identity --query "Account" --output text)}
-if [ -z "$AWS_ACCOUNT" ]; then echo "Variable AWS_ACCOUNT needs to be set."; exit 1; fi
-
-if [ -z "$VERSION" ]; then echo "Variable VERSION needs to be set."; exit 1; fi
-CLUSTER_NAME=${CLUSTER_NAME:-$(whoami)}
-if [ -z "$CLUSTER_NAME" ]; then echo "Variable CLUSTER_NAME needs to be set."; exit 1; fi
-if [ -z "$REGION" ]; then echo "Variable REGION needs to be set."; exit 1; fi
-if [ -z "$COMPUTE_MACHINE_TYPE" ]; then echo "Variable COMPUTE_MACHINE_TYPE needs to be set."; exit 1; fi
-
-if [ "$MULTI_AZ" = "true" ]; then MULTI_AZ_PARAM="--multi-az"; else MULTI_AZ_PARAM=""; fi
-if [ -z "$AVAILABILITY_ZONES" ]; then AVAILABILITY_ZONES_PARAM=""; else AVAILABILITY_ZONES_PARAM="--availability-zones $AVAILABILITY_ZONES"; fi
-if [ -z "$REPLICAS" ]; then echo "Variable REPLICAS needs to be set."; exit 1; fi
-
-echo "Checking if cluster ${CLUSTER_NAME} already exists."
-if rosa describe cluster --cluster="${CLUSTER_NAME}"; then
-  echo "Cluster ${CLUSTER_NAME} already exists."
-else
-  echo "Verifying ROSA prerequisites."
-  echo "Check if AWS CLI is installed."; aws --version
-  echo "Check if ROSA CLI is installed."; rosa version
-  echo "Check if ELB service role is enabled."
-  if ! aws iam get-role --role-name "AWSServiceRoleForElasticLoadBalancing" --no-cli-pager; then
-    aws iam create-service-linked-role --aws-service-name "elasticloadbalancing.amazonaws.com"
-  fi
-  rosa whoami
-  rosa verify quota
-
-  echo "Installing ROSA cluster ${CLUSTER_NAME}"
-
-  MACHINE_CIDR=$(./rosa_machine_cidr.sh)
-
-  ROSA_CMD="rosa create cluster \
-  --sts \
-  --cluster-name ${CLUSTER_NAME} \
-  --version ${VERSION} \
-  --role-arn arn:aws:iam::${AWS_ACCOUNT}:role/ManagedOpenShift-Installer-Role \
-  --support-role-arn arn:aws:iam::${AWS_ACCOUNT}:role/ManagedOpenShift-Support-Role \
-  --controlplane-iam-role arn:aws:iam::${AWS_ACCOUNT}:role/ManagedOpenShift-ControlPlane-Role \
-  --worker-iam-role arn:aws:iam::${AWS_ACCOUNT}:role/ManagedOpenShift-Worker-Role \
-  --operator-roles-prefix ${CLUSTER_NAME} \
-  --region ${REGION} ${MULTI_AZ_PARAM} ${AVAILABILITY_ZONES_PARAM} \
-  --replicas ${REPLICAS} \
-  --compute-machine-type ${COMPUTE_MACHINE_TYPE} \
-  --machine-cidr ${MACHINE_CIDR} \
-  --service-cidr 172.30.0.0/16 \
-  --pod-cidr 10.128.0.0/14 \
-  --host-prefix 23"
-
-  echo $ROSA_CMD
-  $ROSA_CMD
+
+requiredEnv AWS_ACCOUNT CLUSTER_NAME REGION
+
+export CLUSTER_NAME=${CLUSTER_NAME:-$(whoami)}
+
+echo "Verifying ROSA prerequisites."
+echo "Check if AWS CLI is installed."; aws --version
+echo "Check if ROSA CLI is installed."; rosa version
+echo "Check if ELB service role is enabled."
+if ! aws iam get-role --role-name "AWSServiceRoleForElasticLoadBalancing" --no-cli-pager; then
+  aws iam create-service-linked-role --aws-service-name "elasticloadbalancing.amazonaws.com"
 fi
+rosa whoami
+rosa verify quota
 
-mkdir -p "logs/${CLUSTER_NAME}"
+echo "Installing ROSA cluster ${CLUSTER_NAME}"
 
-function custom_date() {
-    date '+%Y%m%d-%H%M%S'
-}
+cd ${SCRIPT_DIR}/../opentofu/modules/rosa/hcp
+tofu init
+tofu workspace new ${CLUSTER_NAME} || true
+export TF_WORKSPACE=${CLUSTER_NAME}
+
+TOFU_CMD="tofu apply -auto-approve \
+  -var cluster_name=${CLUSTER_NAME} \
+  -var region=${REGION}"
+
+if [ -n "${COMPUTE_MACHINE_TYPE}" ]; then
+  TOFU_CMD+=" -var instance_type=${COMPUTE_MACHINE_TYPE}"
+fi
+
+if [ -n "${VERSION}" ]; then
+  TOFU_CMD+=" -var openshift_version=${VERSION}"
+fi
+
+if [ -n "${REPLICAS}" ]; then
+  TOFU_CMD+=" -var replicas=${REPLICAS}"
+fi
 
-echo "Creating operator roles."
-rosa create operator-roles --cluster "${CLUSTER_NAME}" --mode auto --yes > "logs/${CLUSTER_NAME}/$(custom_date)_create-operator-roles.log"
-
-echo "Creating OIDC provider."
-rosa create oidc-provider --cluster "${CLUSTER_NAME}" --mode auto --yes > "logs/${CLUSTER_NAME}/$(custom_date)_create-oidc-provider.log"
-
-echo "Waiting for cluster installation to finish."
-# There have been failures with 'ERR: Failed to watch logs for cluster ... connection reset by peer' probably because services in the cluster were restarting during the cluster initialization.
-# Those errors don't show an installation problem, and installation will continue asynchronously. Therefore, retry.
-TIMEOUT=$(($(date +%s) + 3600))
-while true ; do
-  if (rosa logs install --cluster "${CLUSTER_NAME}" --watch --tail=1000000 >> "logs/${CLUSTER_NAME}/$(custom_date)_create-cluster.log"); then
-    break
-  fi
-  if (( TIMEOUT < $(date +%s))); then
-    echo "Timeout exceeded"
-    exit 1
-  fi
-  echo "retrying watching logs after failure"
-  sleep 1
-done
-
-echo "Cluster installation complete."
-echo
-
-./rosa_recreate_admin.sh
+echo ${TOFU_CMD}
+${TOFU_CMD}
 
 SCALING_MACHINE_POOL=$(rosa list machinepools -c "${CLUSTER_NAME}" -o json | jq -r '.[] | select(.id == "scaling") | .id')
 if [[ "${SCALING_MACHINE_POOL}" != "scaling" ]]; then
-    rosa create machinepool -c "${CLUSTER_NAME}" --instance-type m5.4xlarge --max-replicas 10 --min-replicas 0 --name scaling --enable-autoscaling
+    rosa create machinepool -c "${CLUSTER_NAME}" --instance-type m5.4xlarge --max-replicas 10 --min-replicas 1 --name scaling --enable-autoscaling
 fi
 
+cd ${SCRIPT_DIR}
+./rosa_oc_login.sh
 ./rosa_efs_create.sh
 ../infinispan/install_operator.sh