Scalability and Cost Management for Azure Red Hat OpenShift
This content is authored by Red Hat experts, but has not yet been tested on every supported configuration.
With Azure Red Hat OpenShift (ARO), you can take advantage of flexible pricing models, including pay-as-you-go and reserved instances, to further optimize your cloud spending. Its auto-scaling capabilities help reduce costs by avoiding over-provisioning, making it a cost-effective solution for organizations seeking to balance performance and expenditure
This guide demonstrates how to implement scheduled scaling in Azure Red Hat OpenShift (ARO), enabling your cluster to automatically adjust its size according to a predefined schedule. By configuring scale-downs during periods of low activity and scale-ups when additional resources are needed, you can ensure both cost efficiency and optimal performance.
Leveraging ARO’s automated scaling capabilities allows for dynamic adjustment of worker node capacity, eliminating wasteful spending on idle infrastructure resources. This approach reduces the need for manual intervention and ensures consistent compute resources for both traditional workloads and AI/ML operations during peak hours.
Prerequisites
The following three CLIs need to be installed.
occli- Access to Azure Red Hat OpenShift cluster
Note: You must log into your ARO cluster via your oc cli before going through the following steps.
Step 1: Create a New project and Service Account
Create a new project
oc new-project worker-scaling
Create the service account
oc create serviceaccount worker-scaler -n worker-scaling
Step 2: Create RBAC Resources
Create the necessary ClusterRole and ClusterRoleBinding to grant permissions:
oc apply -f - <<EOF
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
name: worker-scaler
rules:
- apiGroups: ["machine.openshift.io"]
resources: ["machinesets"]
verbs: ["get", "list", "patch", "update"]
- apiGroups: [""]
resources: ["nodes"]
verbs: ["get", "list"]
EOF
oc apply -f - <<EOF
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRoleBinding
metadata:
name: worker-scaler
roleRef:
apiGroup: rbac.authorization.k8s.io
kind: ClusterRole
name: worker-scaler
subjects:
- kind: ServiceAccount
name: worker-scaler
namespace: worker-scaling
EOF
Step 3: Create the Scaling Script
Environment Variables
DESIRED_REPLICAS: Number of replicas per machineset (default: 3)MACHINESET_LABEL: Label selector for machinesets (default: worker role)
Create a ConfigMap containing the scaling script:
oc apply -f - <<'EOF'
apiVersion: v1
kind: ConfigMap
metadata:
name: scaling-script
namespace: worker-scaling
data:
scale-workers.sh: |
#!/bin/bash
set -e
# Configuration
DESIRED_REPLICAS="${DESIRED_REPLICAS:-3}"
MACHINESET_LABEL="machine.openshift.io/cluster-api-machine-role=worker"
echo "Starting worker node scaling..."
echo "Target replicas: $DESIRED_REPLICAS"
# Get worker machinesets
MACHINESETS=$(oc get machinesets -n openshift-machine-api \
-l "$MACHINESET_LABEL" -o name)
if [ -z "$MACHINESETS" ]; then
echo "No machinesets found with label: $MACHINESET_LABEL"
exit 1
fi
# Scale each machineset
for MACHINESET in $MACHINESETS; do
MACHINESET_NAME=$(echo $MACHINESET | cut -d'/' -f2)
echo "Scaling $MACHINESET_NAME to $DESIRED_REPLICAS replicas"
# Get current replicas
CURRENT_REPLICAS=$(oc get $MACHINESET -n openshift-machine-api \
-o jsonpath='{.spec.replicas}')
echo "Current replicas for $MACHINESET_NAME: $CURRENT_REPLICAS"
if [ "$CURRENT_REPLICAS" != "$DESIRED_REPLICAS" ]; then
# Scale the machineset
oc patch $MACHINESET -n openshift-machine-api \
-p "{\"spec\":{\"replicas\":$DESIRED_REPLICAS}}" --type=merge
echo "Scaled $MACHINESET_NAME from $CURRENT_REPLICAS to" \
"$DESIRED_REPLICAS replicas"
else
echo "$MACHINESET_NAME already has $DESIRED_REPLICAS replicas"
fi
done
echo "Scaling operation completed"
# Wait and report status
echo "Waiting 30 seconds before checking status..."
sleep 30
echo "Current machineset status:"
oc get machinesets -n openshift-machine-api -l "$MACHINESET_LABEL"
echo "Current node count:"
oc get nodes --no-headers | wc -l
EOF
Step 4: Create the CronJob
For testing you can adjust accordingly. For example
"0 8 * * *"- Daily at 8:00 AM"0 8 * * 1-5"- Weekdays at 8:00 AM"0 8,20 * * *"- Daily at 8:00 AM and 8:00 PM"*/30 * * * *"- Every 30 minutes
Create a CronJob that will execute the scaling script:
oc apply -f - <<EOF
apiVersion: batch/v1
kind: CronJob
metadata:
name: worker-scaler
namespace: worker-scaling
spec:
# Schedule: Run every day at 8:00 AM (adjust as needed)
schedule: "0 8 * * *"
jobTemplate:
spec:
template:
spec:
serviceAccountName: worker-scaler
restartPolicy: OnFailure
containers:
- name: worker-scaler
image: quay.io/openshift/origin-cli:latest
command: ["/bin/bash"]
args: ["/scripts/scale-workers.sh"]
env:
# Set desired number of replicas per machineset
- name: DESIRED_REPLICAS
value: "3"
# Optional: Specify machineset label selector
- name: MACHINESET_LABEL
value: "machine.openshift.io/cluster-api-machine-role=worker"
volumeMounts:
- name: scaling-script
mountPath: /scripts
resources:
requests:
memory: "64Mi"
cpu: "50m"
limits:
memory: "128Mi"
cpu: "100m"
volumes:
- name: scaling-script
configMap:
name: scaling-script
defaultMode: 0755
# Keep last 3 successful jobs and 1 failed job
successfulJobsHistoryLimit: 3
failedJobsHistoryLimit: 1
EOF
Step 5: Verify the Setup
Check that all resources are created correctly:
Verify service account
oc get serviceaccount worker-scaler -n worker-scaling
Example output:
aro-cluster$ oc get serviceaccount worker-scaler -n worker-scaling
NAME SECRETS AGE
worker-scaler 1 101m
Verify RBAC
oc get clusterrole worker-scaler
oc get clusterrolebinding worker-scaler
Example output:
aro-cluster$ oc get clusterrole worker-scaler
NAME CREATED AT
worker-scaler 2025-06-16T18:35:12Z
aro-cluster$ oc get clusterrolebinding worker-scaler
NAME ROLE AGE
worker-scaler ClusterRole/worker-scaler 106m
Verify ConfigMap
oc get configmap scaling-script -n worker-scaling
Example output:
aro-cluster$ oc get configmap scaling-script -n worker-scaling
NAME DATA AGE
scaling-script 1 107m
Verify CronJob
oc get cronjob worker-scaler -n worker-scaling
Example output:
aro-cluster$ oc get cronjob worker-scaler -n worker-scaling
NAME SCHEDULE SUSPEND ACTIVE LAST SCHEDULE AGE
worker-scaler 0 8 * * * False 0 <none> 6s
Step 6: Test the CronJob
You can manually trigger the CronJob to test it:
Create a manual job from the CronJob
oc create job --from=cronjob/worker-scaler manual-test-1 -n worker-scaling
Check the job status
oc get jobs -n worker-scaling
Check the pod logs
oc logs -f job/manual-test-1 -n worker-scaling
Example output:
aro-cluster$ oc get jobs -n worker-scaling
NAME COMPLETIONS DURATION AGE
manual-test-1 0/1 17s 17s
Step 7: Monitor and Manage
Monitor the CronJob execution:
Check CronJob status
oc get cronjob worker-scaler -n worker-scaling
Example output:
aro-cluster$ oc get cronjob worker-scaler -n worker-scaling
NAME SCHEDULE SUSPEND ACTIVE LAST SCHEDULE AGE
worker-scaler 0 8 * * * False 0 <none> 3m15s
View recent jobs
oc get jobs -n worker-scaling
Example output:
aro-cluster$ oc get jobs -n worker-scaling
NAME COMPLETIONS DURATION AGE
manual-test-1 1/1 55s 2m51s
Creating a Scale-Down CronJob
To create a complementary scale-down job:
oc apply -f - <<EOF
apiVersion: batch/v1
kind: CronJob
metadata:
name: worker-scaler-down
namespace: worker-scaling
spec:
# Schedule: Run every day at 6:00 PM
schedule: "0 18 * * *"
jobTemplate:
spec:
template:
spec:
serviceAccountName: worker-scaler
restartPolicy: OnFailure
containers:
- name: worker-scaler
image: quay.io/openshift/origin-cli:latest
command: ["/bin/bash"]
args: ["/scripts/scale-workers.sh"]
env:
- name: DESIRED_REPLICAS
value: "1" # Scale down to 1 replica
volumeMounts:
- name: scaling-script
mountPath: /scripts
volumes:
- name: scaling-script
configMap:
name: scaling-script
defaultMode: 0755
EOF
Finally sit back and watch the machinesets scale on the schedule you configured.
Check machinesets
oc get machinesets -n openshift-machine-api
Check machines
oc get machines -n openshift-machine-api
Check nodes
oc get nodes
