Services

Services provide stable network endpoints for accessing pods. Since pods are ephemeral and can be created/destroyed frequently, services give you consistent DNS names and IP addresses for reliable communication.

Why Services are important:

Pods can come and go, so clients cannot reliably connect to them directly. Services:

• Provide stable networking: IP and DNS names remain same even if pods change.
• Offer load balancing: Automatically distribute requests across healthy pods
• Enable service discovery: Other components can reach the service by name
• Provide pod abstraction: Clients don’t need to know individual pod IPs
• Handle automatic updates: Adjust endpoints as pods are created or destroyed

In this lab, you'll create a service for the catalog component of our retail store and explore how services enable communication between pods.

Service Types

Kubernetes provides different service types for various use cases:

Type	Purpose	Access
ClusterIP	Internal cluster communication	Cluster-only
NodePort	External access via node ports	External
LoadBalancer	External access via cloud load balancer	External
ExternalName	Map to external DNS name	External

備考

A dedicated lab on LoadBalancer services is available later in this workshop. You will learn how to expose services externally using a cloud load balancer there.

Creating a Service

Let's examine the UI service from our retail store:

~/environment/eks-workshop/base-application/ui/service.yaml

apiVersion: v1
kind: Service
metadata:
  name: ui
  labels:
    app.kubernetes.io/name: ui
    app.kubernetes.io/instance: ui
    app.kubernetes.io/component: service
    app.kubernetes.io/created-by: eks-workshop
spec:
  type: ClusterIP
  ports:
    - port: 80
      targetPort: http
      protocol: TCP
      name: http
  selector:
    app.kubernetes.io/name: ui
    app.kubernetes.io/instance: ui
    app.kubernetes.io/component: service

A

kind: Service: Creates a Service resource

B

metadata.name: Name of the service (ui)

C

spec.type: Service type (ClusterIP for internal access)

D

spec.ports: Port mapping from service to pods

E

spec.selector: Selects which pods receive traffic

Deploy the service:

~$kubectl apply -k ~/environment/eks-workshop/modules/introduction/basics/services/

How Services Connect to Pods

Services don't directly know about specific pods. Instead, they use label selectors to dynamically find pods that should receive traffic. This creates a flexible, loosely-coupled relationship.

Here's how it works:

1. Pods have labels - Key-value pairs that describe the pod
2. Services have selectors - Criteria that match pod labels
3. Kubernetes automatically connects them - Any pod matching the selector becomes an endpoint

Let's see this in action with our UI service:

~$kubectl get service -n ui ui -o jsonpath='{.spec.selector}' | jq

{

  "app.kubernetes.io/component": "service",

  "app.kubernetes.io/instance": "ui",

  "app.kubernetes.io/name": "ui"

}

Now check which pods have matching labels:

~$kubectl get pod -n ui -l app.kubernetes.io/component=service -o jsonpath='{.items[0].metadata.labels}{"\n"}' | jq

{

  "app.kubernetes.io/component": "service",

  "app.kubernetes.io/created-by": "eks-workshop",

  "app.kubernetes.io/instance": "ui",

  "app.kubernetes.io/name": "ui",

  "pod-template-hash": "5989474687"

}

You'll see the UI pods have labels that match the service selector. This is how the service knows which pods to send traffic to.

The relationship is dynamic:

• When new pods start with matching labels, they automatically become service endpoints
• When pods are deleted, they're automatically removed from the service
• If you change a pod's labels, it can be added or removed from services

This label-based system means:

• Services work with any workload controller (Deployments, StatefulSets, etc.)
• Pods can belong to multiple services if they match different selectors
• Services automatically adapt as pods scale up or down

Exploring Your Service

Check service status:

~$kubectl get service -n ui

NAME   TYPE        CLUSTER-IP      EXTERNAL-IP   PORT(S)   AGE

ui     ClusterIP   172.20.83.84    <none>        80/TCP    15m

View service endpoints (the actual pod IPs):

~$kubectl get endpoints -n ui ui

NAME   ENDPOINTS           AGE

ui     10.42.1.15:8080     15m

This shows which pods receive traffic

Get detailed service information:

~$kubectl describe service -n ui ui

Name:                     ui

Namespace:                ui

Labels:                   app.kubernetes.io/component=service

                          app.kubernetes.io/created-by=eks-workshop

                          app.kubernetes.io/instance=ui

                          app.kubernetes.io/name=ui

Annotations:              <none>

Selector:                 app.kubernetes.io/component=service,app.kubernetes.io/instance=ui,app.kubernetes.io/name=ui

Type:                     ClusterIP

IP Family Policy:         SingleStack

IP Families:              IPv4

IP:                       172.16.88.252

IPs:                      172.16.88.252

Port:                     http  80/TCP

TargetPort:               http/TCP

Endpoints:                10.42.129.33:8080

Session Affinity:         None

Internal Traffic Policy:  Cluster

Events:                   <none>

Service Discovery

Services enable automatic service discovery through DNS names:

Full DNS name format:

<service-name>.<namespace>.svc.cluster.local

Examples from our retail store:

• ui.ui.svc.cluster.local
• catalog.catalog.svc.cluster.local
• carts.carts.svc.cluster.local

Short names within the same namespace:

# From a pod in the ui namespace
curl http://ui:80

# From a different namespace, use the full name
curl http://ui.ui.svc.cluster.local:80

Testing Service Communication

Let's test service discovery and communication by creating a test pod:

~$kubectl run test-pod --image=curlimages/curl --restart=Never -- sleep 3600

~$kubectl wait --for=condition=ready pod/test-pod --timeout=60s

~$kubectl exec test-pod -- nslookup ui.ui.svc.cluster.local

Server:         172.16.0.10

Address:        172.16.0.10:53

Name:   ui.ui.svc.cluster.local

Address: 172.16.88.252

~$kubectl exec test-pod -- curl -s http://ui.ui.svc.cluster.local/actuator/info | jq

{

  "pod": {

    "name": "ui-6db5f6bd84-cx4mg"

  }

}

Load Balancing

Services automatically distribute traffic across all healthy pods that match their selector:

Scale the UI deployment to see load balancing:

~$kubectl scale deployment -n ui ui --replicas=3

Watch how the service endpoints update:

~$kubectl get endpoints -n ui ui

NAME   ENDPOINTS                                               AGE

ui     10.42.117.212:8080,10.42.129.33:8080,10.42.174.4:8080   11m

You'll now see multiple pod IPs listed as endpoints - the service automatically discovered the new pods because they have matching labels.

Test load balancing:

~$for i in $(seq 1 5); do printf "Request %d:" "$i"; kubectl exec test-pod -- curl -s http://ui.ui.svc.cluster.local/actuator/info; echo; sleep 1; done

Request 1:{"pod":{"name":"ui-6db5f6bd84-xgpf4"}}

Request 2:{"pod":{"name":"ui-6db5f6bd84-cx4mg"}}

Request 3:{"pod":{"name":"ui-6db5f6bd84-7bq8w"}}

Request 4:{"pod":{"name":"ui-6db5f6bd84-7bq8w"}}

Request 5:{"pod":{"name":"ui-6db5f6bd84-cx4mg"}}

You'll see requests distributed across different pod hostnames, demonstrating how the service load balances across all matching pods.

~$kubectl delete pod test-pod

Key Points to Remember

• Services provide stable network endpoints for ephemeral pods
• ClusterIP services enable internal cluster communication
• Services use label selectors to find target pods
• DNS names follow the pattern: service.namespace.svc.cluster.local
• Services automatically load balance traffic across healthy pods
• Use port forwarding to test services locally