<인프런> 대세는 쿠버네티스 [초급] - No.8 Object - Pod 실습

Overview

이번 글에서는 지난 이론 학습에 이어 Kubernetes의 가장 기본이 되는 오브젝트, Pod에 대해 실습을 통해 확인해보는 시간을 가졌다.

여러 컨테이너를 포함한 Pod 생성부터 시작해, 포트 충돌 상황, Auto Healing, Label 기반 서비스 연결, 노드 스케줄링 등
실제 클러스터에서 Pod가 어떤 방식으로 배치되고 서비스와 연결되는지를 자세히 살펴보았다.

2022.08.29 - [교육, 커뮤니티 후기] - <인프런> 대세는 쿠버네티스 [초급] - No.7 Object - Pod

 

 

---

 

 

1. Pod

 

 

실습 1) Pod 생성 

 

아래와 같이 생성해 준다.

$ vi pod-1.yaml
apiVersion: v1
kind: Pod
metadata:
 name: pod-1
spec:
 containers:
 - name: container1
   image: tmkube/p8000
   ports:
   - containerPort: 8000
 - name: container2
   image: tmkube/p8080
   ports:
   - containerPort: 8080

$ kubectl apply -f pod-1.yaml

$ kubectl get po -o wide
NAME    READY   STATUS    RESTARTS   AGE   IP              NODE          NOMINATED NODE   READINESS GATES
pod-1   2/2     Running   0          28s   172.16.12.132   dh-k8s-node   <none>           <none>

 

 

생성 후 Pod IP를 이용해 확인해준다.

$ kubectl describe po pod-1
Name:         pod-1
Namespace:    default
Priority:     0
Node:         dh-k8s-node/192.168.21.113
Start Time:   Tue, 30 Aug 2022 14:32:58 +0900
Labels:       <none>
Annotations:  cni.projectcalico.org/containerID: 9147e1f435033ac92c3a57bad88a9796bd8547057f685034f7f94785bd8aaa73
              cni.projectcalico.org/podIP: 172.16.12.132/32
              cni.projectcalico.org/podIPs: 172.16.12.132/32
Status:       Running
IP:           172.16.12.132
IPs:
  IP:  172.16.12.132

$ curl 172.16.12.132:8000
containerPort : 8000
$ curl 172.16.12.132:8080
containerPort : 8080

 

 

Container1으로 들어가서 Curl 명령어로 확인해보자.

$ kubectl exec -ti pod-1 /bin/bash
kubectl exec [POD] [COMMAND] is DEPRECATED and will be removed in a future version. Use kubectl exec [POD] -- [COMMAND] instead.
Defaulted container "container1" out of: container1, container2

root@pod-1:/#

root@pod-1:/# curl localhost:8080
containerPort : 8080

root@pod-1:/# curl localhost:8000
containerPort : 8000

 

 

 

실습 2) ContainerPort 동일하게 설정 후 Pod 생성

 

containerPort를 동일하게 설정해서 생성해보자.

$ vi pod-2.yaml
apiVersion: v1
kind: Pod
metadata:
 name: pod-2
spec:
 containers:
 - name: container1
   image: tmkube/p8000
   ports:
   - containerPort: 8000
 - name: container2
   image: tmkube/p8000
   ports:
   - containerPort: 8000

$ kubectl apply -f pod-2.yaml

$ kubectl get po -o wide |grep pod-2
pod-2   1/2     Error     1 (4s ago)   10s   172.16.12.134   dh-k8s-node   <none>           <none>

 

 

상태를 보면 error가 발생한다. describe와 logs로 확인해보자.

$ kubectl describe po pod-2
Name:         pod-2
Namespace:    default
Priority:     0
Node:         dh-k8s-node/192.168.21.113
Start Time:   Tue, 30 Aug 2022 14:48:32 +0900
Labels:       <none>
Annotations:  cni.projectcalico.org/containerID: d786ae9d806c67397f0352f0b8b04155091b68fe107559cbb557441453745697
              cni.projectcalico.org/podIP: 172.16.12.134/32
              cni.projectcalico.org/podIPs: 172.16.12.134/32
Status:       Running
IP:           172.16.12.134
IPs:
  IP:  172.16.12.134
Containers:
  container1:
    Container ID:   docker://d7be96c370d325ccc7adf5a7fd1d17d23957633b8030e0bbc652499478f27687
    Image:          tmkube/p8000
    Image ID:       docker-pullable://tmkube/p8000@sha256:241f2721ac81d83f8b6513c3ec4f44a04a3173cb4629fff60da45909fc6f773d
    Port:           8000/TCP
    Host Port:      0/TCP
    State:          Running
      Started:      Tue, 30 Aug 2022 14:48:35 +0900
    Ready:          True
    Restart Count:  0
    Environment:    <none>
    Mounts:
      /var/run/secrets/kubernetes.io/serviceaccount from kube-api-access-zf2n8 (ro)
  container2:
    Container ID:   docker://369c612c82e6aa028b8ca55bc14715531bca1dd0476cd4c14e95b1a332415111
    Image:          tmkube/p8000
    Image ID:       docker-pullable://tmkube/p8000@sha256:241f2721ac81d83f8b6513c3ec4f44a04a3173cb4629fff60da45909fc6f773d
    Port:           8000/TCP
    Host Port:      0/TCP
    State:          Waiting
      Reason:       CrashLoopBackOff
    Last State:     Terminated
      Reason:       Error
      Exit Code:    1
      Started:      Tue, 30 Aug 2022 14:51:48 +0900
      Finished:     Tue, 30 Aug 2022 14:51:48 +0900
    Ready:          False
    Restart Count:  5
    Environment:    <none>
    Mounts:
      /var/run/secrets/kubernetes.io/serviceaccount from kube-api-access-zf2n8 (ro)
Conditions:
  Type              Status
  Initialized       True 
  Ready             False 
  ContainersReady   False 
  PodScheduled      True 
Volumes:
  kube-api-access-zf2n8:
    Type:                    Projected (a volume that contains injected data from multiple sources)
    TokenExpirationSeconds:  3607
    ConfigMapName:           kube-root-ca.crt
    ConfigMapOptional:       <nil>
    DownwardAPI:             true
QoS Class:                   BestEffort
Node-Selectors:              <none>
Tolerations:                 node.kubernetes.io/not-ready:NoExecute op=Exists for 300s
                             node.kubernetes.io/unreachable:NoExecute op=Exists for 300s
Events:
  Type     Reason     Age                    From               Message
  ----     ------     ----                   ----               -------
  Normal   Scheduled  3m34s                  default-scheduler  Successfully assigned default/pod-2 to dh-k8s-node
  Normal   Pulling    3m33s                  kubelet            Pulling image "tmkube/p8000"
  Normal   Pulled     3m31s                  kubelet            Successfully pulled image "tmkube/p8000" in 2.622825113s
  Normal   Created    3m31s                  kubelet            Created container container1
  Normal   Started    3m31s                  kubelet            Started container container1
  Normal   Pulled     3m28s                  kubelet            Successfully pulled image "tmkube/p8000" in 2.572180607s
  Normal   Pulled     3m24s                  kubelet            Successfully pulled image "tmkube/p8000" in 2.572726498s
  Normal   Pulled     3m9s                   kubelet            Successfully pulled image "tmkube/p8000" in 2.621793452s
  Normal   Pulling    2m44s (x4 over 3m31s)  kubelet            Pulling image "tmkube/p8000"
  Normal   Created    2m42s (x4 over 3m28s)  kubelet            Created container container2
  Normal   Pulled     2m42s                  kubelet            Successfully pulled image "tmkube/p8000" in 2.53015705s
  Normal   Started    2m41s (x4 over 3m28s)  kubelet            Started container container2
  Warning  BackOff    2m40s (x5 over 3m24s)  kubelet            Back-off restarting failed container

 

 

 

아래와 같이 container2에 에러로그가 발생한다. 현재 8000번 Port가 이미 사용중이라는 error이다.

이처럼 한 Pod내의 Container는 같은  Port를 사용할 수 없다.

[root@dh-k8s-master ~]# kubectl logs pod-2 -c container1
[root@dh-k8s-master ~]# kubectl logs pod-2 -c container2
events.js:177
      throw er; // Unhandled 'error' event
      ^

Error: listen EADDRINUSE: address already in use :::8000
    at Server.setupListenHandle [as _listen2] (net.js:1226:14)
    at listenInCluster (net.js:1274:12)
    at Server.listen (net.js:1362:7)
    at Object.<anonymous> (/port8000.js:7:3)
    at Module._compile (internal/modules/cjs/loader.js:774:30)
    at Object.Module._extensions..js (internal/modules/cjs/loader.js:785:10)
    at Module.load (internal/modules/cjs/loader.js:641:32)
    at Function.Module._load (internal/modules/cjs/loader.js:556:12)
    at Function.Module.runMain (internal/modules/cjs/loader.js:837:10)
    at internal/main/run_main_module.js:17:11
Emitted 'error' event at:
    at emitErrorNT (net.js:1253:8)
    at processTicksAndRejections (internal/process/task_queues.js:84:9) {
  code: 'EADDRINUSE',
  errno: 'EADDRINUSE',
  syscall: 'listen',
  address: '::',
  port: 8000
}

 

 

 

 

실습 3) Pod가 Cluster에서 관리가 되었을 될 때 Pod 삭제 후 Auto Healing

 

ReplicationController 를 만들어보자.

$ vi replication-1.yaml

apiVersion: v1
kind: ReplicationController
metadata:
  name: replication-1
spec:
  replicas: 1
  selector:
    app: rc
  template:
    metadata:
      name: pod-1
      labels:
        app: rc
    spec:
      containers:
      - name: container
        image: tmkube/init

$ kubectl apply -f replication-1.yaml 
replicationcontroller/replication-1 created

$ kubectl get po -o wide
NAME                  READY   STATUS    RESTARTS   AGE    IP              NODE          NOMINATED NODE   READINESS GATES
pod-1                 2/2     Running   0          35m    172.16.12.132   dh-k8s-node   <none>           <none>
replication-1-t764t   1/1     Running   0          101s   172.16.12.135   dh-k8s-node   <none>           <none>

 

 

pod 삭제를 해보자. 

$ kubectl delete po replication-1-t764t
pod "replication-1-t764t" deleted

$ kubectl get po -o wide
NAME                  READY   STATUS    RESTARTS   AGE   IP              NODE          NOMINATED NODE   READINESS GATES
pod-1                 2/2     Running   0          39m   172.16.12.132   dh-k8s-node   <none>           <none>
replication-1-g56xt   1/1     Running   0          34s   172.16.12.136   dh-k8s-node   <none>           <none>

• IP가 바뀌었다.

 

 

 

 

 

---

 

 

 

2. Label

 

 

실습 1) Label을 넣어 Pod 생성

$ vi pod-1.yaml

apiVersion: v1
kind: Pod
metadata:
  name: pod-1
  labels:
    type: web
    lo: dev
spec:
  containers:
  - name: container
    image: tmkube/init


$ vi pod-2.yaml

apiVersion: v1
kind: Pod
metadata:
  name: pod-2
  labels:
    type: db
    lo: dev
spec:
  containers:
  - name: container
    image: tmkube/init


$ vi pod-3.yaml

apiVersion: v1
kind: Pod
metadata:
  name: pod-3
  labels:
    type: server
    lo: dev
spec:
  containers:
  - name: container
    image: tmkube/init


$ vi pod-4.yaml

apiVersion: v1
kind: Pod
metadata:
  name: pod-4
  labels:
    type: web
    lo: prod
spec:
  containers:
  - name: container
    image: tmkube/init

$ vi pod-5.yaml

apiVersion: v1
kind: Pod
metadata:
  name: pod-5
  labels:
    type: db
    lo: prod
spec:
  containers:
  - name: container
    image: tmkube/init

$ vi pod-6.yaml

apiVersion: v1
kind: Pod
metadata:
  name: pod-6
  labels:
    type: server
    lo: prod
spec:
  containers:
  - name: container
    image: tmkube/init

$ kubectl apply -f pod-1.yaml
pod/pod-1 created
$ kubectl apply -f pod-2.yaml
pod/pod-2 created
$ kubectl apply -f pod-3.yaml
pod/pod-3 created

$ kubectl apply -f pod-4.yaml
pod/pod-4 created
$ kubectl apply -f pod-5.yaml
pod/pod-5 created
$ kubectl apply -f pod-6.yaml
pod/pod-6 created

$ kubectl get po -o wide
NAME    READY   STATUS    RESTARTS   AGE     IP              NODE          NOMINATED NODE   READINESS GATES
pod-1   1/1     Running   0          4m9s    172.16.12.138   dh-k8s-node   <none>           <none>
pod-2   1/1     Running   0          4m6s    172.16.12.139   dh-k8s-node   <none>           <none>
pod-3   1/1     Running   0          2m55s   172.16.12.140   dh-k8s-node   <none>           <none>
pod-4   1/1     Running   0          36s     172.16.12.141   dh-k8s-node   <none>           <none>
pod-5   1/1     Running   0          34s     172.16.12.142   dh-k8s-node   <none>           <none>
pod-6   1/1     Running   0          31s     172.16.12.143   dh-k8s-node   <none>           <none>

 

 

 

이제 서비스를 설정해 원하는 Pod를 선택해 보겠다.

$ vi svc-web.yaml

apiVersion: v1
kind: Service
metadata:
  name: svc-for-web
spec:
  selector:
    type: web
  ports:
  - port: 8080
  
$ kubectl apply -f svc-web.yaml
service/svc-for-web created

$ kubectl get svc -o wide
NAME          TYPE        CLUSTER-IP      EXTERNAL-IP   PORT(S)    AGE    SELECTOR
kubernetes    ClusterIP   10.96.0.1       <none>        443/TCP    21h    <none>
svc-for-web   ClusterIP   10.111.41.178   <none>        8080/TCP   118s   type=web

$ kubectl get po -o wide
NAME    READY   STATUS    RESTARTS   AGE   IP              NODE          NOMINATED NODE   READINESS GATES
pod-1   1/1     Running   0          14m   172.16.12.138   dh-k8s-node   <none>           <none>
pod-2   1/1     Running   0          53s   172.16.12.144   dh-k8s-node   <none>           <none>
pod-3   1/1     Running   0          13m   172.16.12.140   dh-k8s-node   <none>           <none>
pod-4   1/1     Running   0          11m   172.16.12.141   dh-k8s-node   <none>           <none>
pod-5   1/1     Running   0          11m   172.16.12.142   dh-k8s-node   <none>           <none>
pod-6   1/1     Running   0          10m   172.16.12.143   dh-k8s-node   <none>           <none>


$ kubectl describe svc svc-for-web
Name:              svc-for-web
Namespace:         default
Labels:            <none>
Annotations:       <none>
Selector:          type=web
Type:              ClusterIP
IP Family Policy:  SingleStack
IP Families:       IPv4
IP:                10.107.92.106
IPs:               10.107.92.106
Port:              <unset>  8080/TCP
TargetPort:        8080/TCP
Endpoints:         172.16.12.138:8080,172.16.12.141:8080
Session Affinity:  None
Events:            <none>

• label이 type=web 인 pod-1,4 가 Service로 할당됬다.

 

 

 

 

이번엔 label이 type=prod 인 pod만 Service로 할당되게 해보자.

$ vi svc-prod.yaml

apiVersion: v1
kind: Service
metadata:
  name: svc-for-prod
spec:
  selector:
    lo: prod
  ports:
  - port: 8080

$ kubectl apply -f svc-prod.yaml
service/svc-for-prod created


$ kubectl describe svc svc-for-prod
Name:              svc-for-prod
Namespace:         default
Labels:            <none>
Annotations:       <none>
Selector:          lo=prod
Type:              ClusterIP
IP Family Policy:  SingleStack
IP Families:       IPv4
IP:                10.102.25.208
IPs:               10.102.25.208
Port:              <unset>  8080/TCP
TargetPort:        8080/TCP
Endpoints:         172.16.12.141:8080,172.16.12.142:8080,172.16.12.143:8080
Session Affinity:  None
Events:            <none>

# kubectl get po -o wide
NAME    READY   STATUS    RESTARTS   AGE   IP              NODE          NOMINATED NODE   READINESS GATES
pod-1   1/1     Running   0          83m   172.16.12.138   dh-k8s-node   <none>           <none>
pod-2   1/1     Running   0          70m   172.16.12.144   dh-k8s-node   <none>           <none>
pod-3   1/1     Running   0          82m   172.16.12.140   dh-k8s-node   <none>           <none>
pod-4   1/1     Running   0          80m   172.16.12.141   dh-k8s-node   <none>           <none>
pod-5   1/1     Running   0          80m   172.16.12.142   dh-k8s-node   <none>           <none>
pod-6   1/1     Running   0          80m   172.16.12.143   dh-k8s-node   <none>           <none>

•  label이 type=prod 인 pod4,5,6이 Service로 할당되었다.

 

 

 

3. Node Schedule

 

 

 

실습 1) nodeSelector을 넣어 Pod 생성

 

앞서 생성한 pod와 서비스를 전부 삭제해준다.

$ kubectl delete -f svc-prod.yaml 
$ kubectl delete -f svc-web.yaml

# running 중인 pod를 모두 삭제해준다.
$ kubectl get po |grep Running | awk '{print $1}' | xargs kubectl delete po
pod "pod-1" deleted
pod "pod-2" deleted
pod "pod-3" deleted
pod "pod-4" deleted
pod "pod-5" deleted
pod "pod-6" deleted

 

$ vi pod-3.yaml

apiVersion: v1
kind: Pod
metadata:
  name: pod-3
spec:
  nodeSelector:
    kubernetes.io/hostname: dh-k8s-node
  containers:
  - name: container
    image: tmkube/init

$ kubectl apply -f pod-3.yaml 
pod/pod-3 created
$ kubectl get po -o wide
NAME    READY   STATUS    RESTARTS   AGE   IP              NODE          NOMINATED NODE   READINESS GATES
pod-3   1/1     Running   0          5s    172.16.12.145   dh-k8s-node   <none>           <none>

 

 

 

---

 

 

 

마무리

이번 실습을 통해 Kubernetes Pod가 단순한 컨테이너 집합이 아닌,

네트워크 포트, 레이블, 노드 배치 전략 등 다양한 요소와 연계되어 작동한다는 점을 직접 체감할 수 있었다.

• 여러 컨테이너가 포함된 Pod 생성 및 IP 접근 테스트
• 포트 충돌 시 에러 발생 확인 → 한 Pod 내의 컨테이너는 동일한 포트를 사용할 수 없음
• ReplicationController를 통한 Auto Healing → Pod 장애 시 자동 재생성
• Label을 통한 Pod 그룹핑 및 서비스 연결
• nodeSelector를 통한 특정 노드 배치 제어

Pod는 쿠버네티스의 시작점이자 모든 오브젝트의 기반이 되는 만큼,

그 구조와 동작 원리를 잘 이해해두면 상위 개념인 ReplicaSet, Deployment, StatefulSet 등도
더 수월하게 학습할 수 있다.

실습을 통해 이론이 머릿속에 자리 잡는 기분이다.
다음에는 ReplicaSet이나 Deployment로 이어지는 고가용성 및 롤링 업데이트 개념도 직접 실습해보고 싶다!