쿠버네티스 기본 개념에 대해 알아보려고 한다.
Pod
쿠버네티스 노드는 여러 Pod로 구성될 수 있고, 각 Pod는 여러 Container를 가질 수 있다.
@cloudshell:~ (proud-curve)$ kubectl explain pods
KIND: Pod
VERSION: v1
DESCRIPTION:
Pod is a collection of containers that can run on a host. This resource is
created by clients and scheduled onto hosts.
FIELDS:
apiVersion <string>
APIVersion defines the versioned schema of this representation of an object.
Servers should convert recognized schemas to the latest internal value, and
may reject unrecognized values. More info:
https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources
kind <string>
Kind is a string value representing the REST resource this object
represents. Servers may infer this from the endpoint the client submits
requests to. Cannot be updated. In CamelCase. More info:
https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds
metadata <ObjectMeta>
Standard object's metadata. More info:
https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#metadata
spec <PodSpec>
Specification of the desired behavior of the pod. More info:
https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#spec-and-status
status <PodStatus>
Most recently observed status of the pod. This data may not be up to date.
Populated by the system. Read-only. More info:
https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#spec-and-status
한 Pod에 있는 컨테이너들은 리소스를 공유하고 같은 Pod에 있는 컨테이너들은 내부 IP를 통해 통신한다.
1. get pods를 통해 pod 리스트를 확인할 수 있다.
@cloudshell:~ (proud-curve)$ kubectl get pods
NAME READY STATUS RESTARTS AGE
hello-world-rest-api-8c7449fd-f7lh2 1/1 Running 0 16h
@cloudshell:~ (proud-curve)$ kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
hello-world-rest-api-8c7449fd-f7lh2 1/1 Running 0 16h 10.44.1.5 gke-in28minutes-cluster-default-pool-24f9f7fb-xq70 <none> <none>
2. 특정 pod에 대해 상세 조회를 하려면 1번 단계에서 확인한 pod name을 이용해서 kubectl describe pod를 하면 된다.
@cloudshell:~ (proud-curve)$ kubectl describe pod hello-world-rest-api-8c7449fd-f7lh2
Name: hello-world-rest-api-8c7449fd-f7lh2
Namespace: default
Priority: 0
Service Account: default
Node: gke-in28minutes-cluster-default-pool-24f9f7fb-xq70/10.128.0.7
Start Time: Sun, 16 Feb 2025 03:44:56 +0000
Labels: app=hello-world-rest-api
pod-template-hash=8c7449fd
Annotations: cloud.google.com/cluster_autoscaler_unhelpable_since: 2025-02-15T13:37:43+0000
cloud.google.com/cluster_autoscaler_unhelpable_until: Inf
Status: Running
IP: 10.44.1.5
IPs:
IP: 10.44.1.5
Controlled By: ReplicaSet/hello-world-rest-api-8c7449fd
Containers:
hello-world-rest-api:
Container ID: containerd://a5f84557e8c2e25fe5ba0e04bcabac8f9b77cbfbb491b98846169c71687f5c1e
Image: in28min/hello-world-rest-api:0.0.1.RELEASE
Image ID: docker.io/in28min/hello-world-rest-api@sha256:00469c343814aabe56ad1034427f546d43bafaaa11208a1eb0720993743f72be
Port: <none>
Host Port: <none>
State: Running
Started: Sun, 16 Feb 2025 03:45:15 +0000
Ready: True
Restart Count: 0
Environment: <none>
Mounts:
/var/run/secrets/kubernetes.io/serviceaccount from kube-api-access-d2tdw (ro)
Conditions:
Type Status
PodReadyToStartContainers True
Initialized True
Ready True
ContainersReady True
PodScheduled True
Volumes:
kube-api-access-d2tdw:
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: <none>
여기에서 Labels에 있는 이름이 중요하다. 나중에 리플리카셋이나 서비스 등으로 Pod을 묶을 때 쓰일 수 있다. 그리고 Annotation은 Pod에 대한 메타 정보이다.
Replicaset
리플리카셋은 Pod를 모니터링 하고 있다가 필요한 Pod 수보다 줄어들게 되면 다시 Pod를 생성한다. 특정 수 이상의 Pod이 실행되도록 만들어준다.
@cloudshell:~ (proud-curve)$ kubectl explain replicaset
GROUP: apps
KIND: ReplicaSet
VERSION: v1
DESCRIPTION:
ReplicaSet ensures that a specified number of pod replicas are running at
any given time.
FIELDS:
apiVersion <string>
APIVersion defines the versioned schema of this representation of an object.
Servers should convert recognized schemas to the latest internal value, and
may reject unrecognized values. More info:
https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources
kind <string>
Kind is a string value representing the REST resource this object
represents. Servers may infer this from the endpoint the client submits
requests to. Cannot be updated. In CamelCase. More info:
https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds
metadata <ObjectMeta>
If the Labels of a ReplicaSet are empty, they are defaulted to be the same
as the Pod(s) that the ReplicaSet manages. Standard object's metadata. More
info:
https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#metadata
spec <ReplicaSetSpec>
Spec defines the specification of the desired behavior of the ReplicaSet.
More info:
https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#spec-and-status
status <ReplicaSetStatus>
Status is the most recently observed status of the ReplicaSet. This data may
be out of date by some window of time. Populated by the system. Read-only.
More info:
https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#spec-and-status
1. 리플리카셋 리스트를 여러 명령어로 조회할 수 있다.
@cloudshell:~ (proud-curve)$ kubectl get replicasets
NAME DESIRED CURRENT READY AGE
hello-world-rest-api-8c7449fd 1 1 1 17h
@cloudshell:~ (proud-curve)$ kubectl get replicaset
NAME DESIRED CURRENT READY AGE
hello-world-rest-api-8c7449fd 1 1 1 17h
@cloudshell:~ (proud-curve)$ kubectl get rs
NAME DESIRED CURRENT READY AGE
hello-world-rest-api-8c7449fd 1 1 1 17h
2. Pod을 삭제하면 다른 이름으로 자동으로 Pod이 생성되는 것을 확인할 수 있는데, 이 역할을 하는 것이 Replicaset이라고 할 수 있다.
@cloudshell:~ (proud-curve)$ kubectl get pods
NAME READY STATUS RESTARTS AGE
hello-world-rest-api-8c7449fd-f7lh2 1/1 Running 0 17h
@cloudshell:~ (proud-curve)$ kubectl delete pod hello-world-rest-api-8c7449fd-f7lh2
pod "hello-world-rest-api-8c7449fd-f7lh2" deleted
@cloudshell:~ (proud-curve)$ kubectl get pods
NAME READY STATUS RESTARTS AGE
hello-world-rest-api-8c7449fd-bw2hn 1/1 Running 0 5s
3. pod을 3개로 늘려보자. 인스턴스가 3개로 늘어났다고 볼 수 있다.
@cloudshell:~ (proud-curve)$ kubectl scale deployment hello-world-rest-api --replicas=3
deployment.apps/hello-world-rest-api scaled
@cloudshell:~ (proud-curve)$ kubectl get pods
NAME READY STATUS RESTARTS AGE
hello-world-rest-api-8c7449fd-bw2hn 1/1 Running 0 5m47s
hello-world-rest-api-8c7449fd-fzg7d 0/1 ContainerCreating 0 4s
hello-world-rest-api-8c7449fd-mgg96 0/1 ContainerCreating 0 4s
@cloudshell:~ (proud-curve-451012-r7)$ kubectl get rs
NAME DESIRED CURRENT READY AGE
hello-world-rest-api-8c7449fd 3 3 3 17h
4. Pod이 3개로 늘어나고 나서 hello-world 서비스를 브라우저로 접근하면 Pod가 돌아가면서 동작하여 서비스가 됨을 알 수 있다.
5. kubectl scale deployment를 통해 내부적으로 어떤 동작이 있었는지 보기 위해 event를 확인해보면 pod 2개가 추가로 생성된 것을 확인할 수 있다.
@cloudshell:~ (proud-curve)$ kubectl get events --sort-by=.metadata.creationTimestamp
LAST SEEN TYPE REASON OBJECT MESSAGE
17m Normal Scheduled pod/hello-world-rest-api-8c7449fd-bw2hn Successfully assigned default/hello-world-rest-api-8c7449fd-bw2hn to gke-in28minutes-cluster-default-pool-24f9f7fb-xq70
17m Normal Killing pod/hello-world-rest-api-8c7449fd-f7lh2 Stopping container hello-world-rest-api
17m Normal SuccessfulCreate replicaset/hello-world-rest-api-8c7449fd Created pod: hello-world-rest-api-8c7449fd-bw2hn
17m Normal Pulled pod/hello-world-rest-api-8c7449fd-bw2hn Container image "in28min/hello-world-rest-api:0.0.1.RELEASE" already present on machine
17m Normal Created pod/hello-world-rest-api-8c7449fd-bw2hn Created container hello-world-rest-api
17m Normal Started pod/hello-world-rest-api-8c7449fd-bw2hn Started container hello-world-rest-api
12m Normal SuccessfulCreate replicaset/hello-world-rest-api-8c7449fd Created pod: hello-world-rest-api-8c7449fd-fzg7d
12m Normal ScalingReplicaSet deployment/hello-world-rest-api Scaled up replica set hello-world-rest-api-8c7449fd to 3 from 1
12m Normal Scheduled pod/hello-world-rest-api-8c7449fd-fzg7d Successfully assigned default/hello-world-rest-api-8c7449fd-fzg7d to gke-in28minutes-cluster-default-pool-24f9f7fb-vzdc
12m Normal Scheduled pod/hello-world-rest-api-8c7449fd-mgg96 Successfully assigned default/hello-world-rest-api-8c7449fd-mgg96 to gke-in28minutes-cluster-default-pool-24f9f7fb-6q96
12m Normal SuccessfulCreate replicaset/hello-world-rest-api-8c7449fd Created pod: hello-world-rest-api-8c7449fd-mgg96
12m Normal Pulling pod/hello-world-rest-api-8c7449fd-fzg7d Pulling image "in28min/hello-world-rest-api:0.0.1.RELEASE"
12m Normal Pulling pod/hello-world-rest-api-8c7449fd-mgg96 Pulling image "in28min/hello-world-rest-api:0.0.1.RELEASE"
12m Normal Created pod/hello-world-rest-api-8c7449fd-fzg7d Created container hello-world-rest-api
12m Normal Started pod/hello-world-rest-api-8c7449fd-mgg96 Started container hello-world-rest-api
12m Normal Created pod/hello-world-rest-api-8c7449fd-mgg96 Created container hello-world-rest-api
12m Normal Pulled pod/hello-world-rest-api-8c7449fd-mgg96 Successfully pulled image "in28min/hello-world-rest-api:0.0.1.RELEASE" in 3.116s (3.116s including waiting). Image size: 88504325 bytes.
12m Normal Started pod/hello-world-rest-api-8c7449fd-fzg7d Started container hello-world-rest-api
12m Normal Pulled pod/hello-world-rest-api-8c7449fd-fzg7d Successfully pulled image "in28min/hello-world-rest-api:0.0.1.RELEASE" in 3.163s (3.163s including waiting). Image size: 88504325 bytes.
Deployment
다운타임 없이 새로운 버전을 배포할 수 있도록 Pod와 Replicaset의 배포를 관리한다.
@cloudshell:~ (proud-curve)$ kubectl explain deployment
GROUP: apps
KIND: Deployment
VERSION: v1
DESCRIPTION:
Deployment enables declarative updates for Pods and ReplicaSets.
FIELDS:
apiVersion <string>
APIVersion defines the versioned schema of this representation of an object.
Servers should convert recognized schemas to the latest internal value, and
may reject unrecognized values. More info:
https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources
kind <string>
Kind is a string value representing the REST resource this object
represents. Servers may infer this from the endpoint the client submits
requests to. Cannot be updated. In CamelCase. More info:
https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds
metadata <ObjectMeta>
Standard object's metadata. More info:
https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#metadata
spec <DeploymentSpec>
Specification of the desired behavior of the Deployment.
status <DeploymentStatus>
Most recently observed status of the Deployment.
1. 애플리케이션 도커 이미지를 변경한다.
@cloudshell:~ (proud-curve)$ kubectl get rs -o wide
NAME DESIRED CURRENT READY AGE CONTAINERS IMAGES SELECTOR
hello-world-rest-api-8c7449fd 3 3 3 17h hello-world-rest-api in28min/hello-world-rest-api:0.0.1.RELEASE app=hello-world-rest-api,pod-template-hash=8c7449fd
@cloudshell:~ (proud-curve)$ kubectl set image deployment hello-world-rest-api hello-world-rest-api=in28min/hello-world-rest-api:0.0.2.RELEASE
deployment.apps/hello-world-rest-api image updated
2. 이미지 변경을 하게 되면 애플리케이션 버전2로 deployment 된 것인데, 이 과정에서 replicaset은 새로운 이름으로 생성이 되고 pod도 새 버전의 애플리케이션을 위한 pod이 구성된다.
@cloudshell:~ (proud-curve)$ kubectl get rs -o wide
NAME DESIRED CURRENT READY AGE CONTAINERS IMAGES SELECTOR
hello-world-rest-api-89b8d4c97 3 3 3 25s hello-world-rest-api in28min/hello-world-rest-api:0.0.2.RELEASE app=hello-world-rest-api,pod-template-hash=89b8d4c97
hello-world-rest-api-8c7449fd 0 0 0 18h hello-world-rest-api in28min/hello-world-rest-api:0.0.1.RELEASE app=hello-world-rest-api,pod-template-hash=8c7449fd
@cloudshell:~ (proud-curve)$ kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
hello-world-rest-api-89b8d4c97-6l564 1/1 Running 0 44s 10.44.1.9 gke-in28minutes-cluster-default-pool-24f9f7fb-xq70 <none> <none>
hello-world-rest-api-89b8d4c97-6w9w5 1/1 Running 0 41s 10.44.2.8 gke-in28minutes-cluster-default-pool-24f9f7fb-6q96 <none> <none>
hello-world-rest-api-89b8d4c97-m88z8 1/1 Running 0 39s 10.44.0.8 gke-in28minutes-cluster-default-pool-24f9f7fb-vzdc <none> <none>
3. 내부적으로 일어난 동작을 보면 새로운 relicaset을 먼저 생성한 후 pod 1개를 먼저 생성하여 image를 내려받아 컨테이너를 실행하고 기존 replicaset의 개수를 2로 줄이면서 기존 container를 중지하고 pod을 삭제한다. 이 작업을 replicas 개수만큼 3번만큼 진행하여 기존 relicaset에서 실행되던 pod을 순차적으로 신규 relicaset으로 이관되게 된다. 롤링 업데이트 전략이 적용된 예시이다.
@cloudshell:~ (proud-curve)$ kubectl get events --sort-by=.metadata.creationTimestamp
LAST SEEN TYPE REASON OBJECT MESSAGE
...
6m36s Normal SuccessfulCreate replicaset/hello-world-rest-api-89b8d4c97 Created pod: hello-world-rest-api-89b8d4c97-6l564
6m35s Normal Scheduled pod/hello-world-rest-api-89b8d4c97-6l564 Successfully assigned default/hello-world-rest-api-89b8d4c97-6l564 to gke-in28minutes-cluster-default-pool-24f9f7fb-xq70
6m36s Normal ScalingReplicaSet deployment/hello-world-rest-api Scaled up replica set hello-world-rest-api-89b8d4c97 to 1
6m35s Normal Pulling pod/hello-world-rest-api-89b8d4c97-6l564 Pulling image "in28min/hello-world-rest-api:0.0.2.RELEASE"
6m34s Normal Started pod/hello-world-rest-api-89b8d4c97-6l564 Started container hello-world-rest-api
6m34s Normal Created pod/hello-world-rest-api-89b8d4c97-6l564 Created container hello-world-rest-api
6m34s Normal Pulled pod/hello-world-rest-api-89b8d4c97-6l564 Successfully pulled image "in28min/hello-world-rest-api:0.0.2.RELEASE" in 1.45s (1.45s including waiting). Image size: 88504316 bytes.
6m33s Normal Pulling pod/hello-world-rest-api-89b8d4c97-6w9w5 Pulling image "in28min/hello-world-rest-api:0.0.2.RELEASE"
6m33s Normal ScalingReplicaSet deployment/hello-world-rest-api Scaled down replica set hello-world-rest-api-8c7449fd to 2 from 3
6m33s Normal Killing pod/hello-world-rest-api-8c7449fd-bw2hn Stopping container hello-world-rest-api
6m33s Normal SuccessfulDelete replicaset/hello-world-rest-api-8c7449fd Deleted pod: hello-world-rest-api-8c7449fd-bw2hn
6m33s Normal Scheduled pod/hello-world-rest-api-89b8d4c97-6w9w5 Successfully assigned default/hello-world-rest-api-89b8d4c97-6w9w5 to gke-in28minutes-cluster-default-pool-24f9f7fb-6q96
6m33s Normal ScalingReplicaSet deployment/hello-world-rest-api Scaled up replica set hello-world-rest-api-89b8d4c97 to 2 from 1
6m33s Normal SuccessfulCreate replicaset/hello-world-rest-api-89b8d4c97 Created pod: hello-world-rest-api-89b8d4c97-6w9w5
6m32s Normal Started pod/hello-world-rest-api-89b8d4c97-6w9w5 Started container hello-world-rest-api
6m32s Normal Created pod/hello-world-rest-api-89b8d4c97-6w9w5 Created container hello-world-rest-api
6m32s Normal Pulled pod/hello-world-rest-api-89b8d4c97-6w9w5 Successfully pulled image "in28min/hello-world-rest-api:0.0.2.RELEASE" in 879ms (879ms including waiting). Image size: 88504316 bytes.
6m31s Normal Scheduled pod/hello-world-rest-api-89b8d4c97-m88z8 Successfully assigned default/hello-world-rest-api-89b8d4c97-m88z8 to gke-in28minutes-cluster-default-pool-24f9f7fb-vzdc
6m31s Normal Pulling pod/hello-world-rest-api-89b8d4c97-m88z8 Pulling image "in28min/hello-world-rest-api:0.0.2.RELEASE"
6m31s Normal SuccessfulDelete replicaset/hello-world-rest-api-8c7449fd Deleted pod: hello-world-rest-api-8c7449fd-mgg96
6m31s Normal Killing pod/hello-world-rest-api-8c7449fd-mgg96 Stopping container hello-world-rest-api
6m31s Normal SuccessfulCreate replicaset/hello-world-rest-api-89b8d4c97 Created pod: hello-world-rest-api-89b8d4c97-m88z8
6m31s Normal ScalingReplicaSet deployment/hello-world-rest-api Scaled down replica set hello-world-rest-api-8c7449fd to 1 from 2
6m31s Normal ScalingReplicaSet deployment/hello-world-rest-api Scaled up replica set hello-world-rest-api-89b8d4c97 to 3 from 2
6m30s Normal Started pod/hello-world-rest-api-89b8d4c97-m88z8 Started container hello-world-rest-api
6m30s Normal Pulled pod/hello-world-rest-api-89b8d4c97-m88z8 Successfully pulled image "in28min/hello-world-rest-api:0.0.2.RELEASE" in 856ms (856ms including waiting). Image size: 88504316 bytes.
6m30s Normal Created pod/hello-world-rest-api-89b8d4c97-m88z8 Created container hello-world-rest-api
6m29s Normal Killing pod/hello-world-rest-api-8c7449fd-fzg7d Stopping container hello-world-rest-api
6m29s Normal SuccessfulDelete replicaset/hello-world-rest-api-8c7449fd Deleted pod: hello-world-rest-api-8c7449fd-fzg7d
6m29s Normal ScalingReplicaSet deployment/hello-world-rest-api Scaled down replica set hello-world-rest-api-8c7449fd to 0 from 1'Kubernetes' 카테고리의 다른 글
| GKE 구글 클라우드 쉘을 이용하여 애플리케이션을 쿠버네티스 클러스터에 배포하기 (0) | 2025.02.16 |
|---|---|
| GKE(구글 쿠버네티스 엔진) 클러스터 생성 (1) | 2025.02.16 |