stackit-cmf-replatform-springboot-k8s

Runnable replatform example for STACKIT: migrate a Spring Boot workload to SKE with DNS and Observability.

What this example creates

• Optional target project creation
• SKE cluster (small default node pool)
• DNS zone and external-dns integration
• STACKIT Observability instance
• Spring Boot Deployment + Service (LoadBalancer)
• Optional Horizontal Pod Autoscaler (HPA) for Spring Boot
• Metrics exporter sidecar on port 9090
• Observability scrape config for Spring Boot metrics
• Grafana dashboard import
• Optional load-generator pod

Prerequisites

• Terraform >= 1.5
• Service account key file
• Required permissions for SKE, DNS, Observability, and optional project creation

1) Configure variables

Copy the example file:

cp env.tfvars.example env.tfvars

Then edit env.tfvars.

Values you must replace (currently xxxx)

• target_project_owner_email
• parent_container_id
• ske_cluster_name
• observability_instance_name
• dns_zone_name
• dns_zone_display_name

Notes for these required values

• target_project_owner_email: service account or owner email used for project creation
• parent_container_id: folder/container ID (for create_project = true)
• ske_cluster_name: max 11 characters
• observability_instance_name: unique readable instance name
• dns_zone_name: delegated DNS zone (for example cmf-1234.runs.onstackit.cloud)
• dns_zone_display_name: display name for the DNS zone resource

Commonly adjusted values

• service_account_key_path
• target_project_name
• region, availability_zone
• node_pool_machine_type
• springboot_image
• enable_springboot_hpa, springboot_hpa_*
• enable_load_generator, load_profile

Kubernetes rightsizing options

For replatformed workloads on SKE, rightsizing can be applied at multiple layers:

• Pod layer: adjust requests/limits and optionally enable HPA.
• Node pool layer: tune node pool min/max and machine type.
• Storage layer: choose suitable storage classes for persistence characteristics.

Enable HPA (optional)

Set in env.tfvars:

enable_springboot_hpa                         = true
springboot_hpa_min_replicas                   = 1
springboot_hpa_max_replicas                   = 5
springboot_hpa_target_cpu_utilization_percentage = 70

Then apply:

terraform apply -var-file=env.tfvars

2) Deploy

terraform init
terraform validate
terraform plan -var-file=env.tfvars
terraform apply -var-file=env.tfvars

3) Verify workload

mkdir -p .tmp
terraform output -raw kubeconfig > .tmp/replatform.kubeconfig
export KUBECONFIG=$PWD/.tmp/replatform.kubeconfig
kubectl get ns
kubectl get deploy,svc -n springboot
kubectl get pods -n springboot

4) Verify metrics path

HOST=$(terraform output -raw springboot_hostname)
curl -sS "http://$HOST:9090/metrics" | head -n 30

Expected: lines beginning with springboot_legacy_counter, springboot_legacy_gauge, springboot_legacy_metric.

5) Verify outputs

terraform output

Important outputs:

• springboot_url
• springboot_hostname
• observability_grafana_url
• project_id

Troubleshooting

• If metrics-exporter restarts, inspect logs:

POD=$(kubectl get pod -n springboot -l app=springboot -o jsonpath='{.items[0].metadata.name}')
kubectl logs -n springboot "$POD" -c metrics-exporter --tail=200
kubectl logs -n springboot "$POD" -c metrics-exporter --previous --tail=200

• If scrape exists but no data arrives, check target health in Prometheus/Grafana datasource (up{job="<scrape-job>"}).
• Scrape interval must be greater than 60s for this environment (61s is set by default).

Destroy

terraform destroy -var-file=env.tfvars