kubernetes
[따배쿠] Controller - StatefulSet
bbiyak2da
2024. 12. 5. 16:25
StatefulSet
- Pod의 상태를 유지해주는 Controller
- Pod 이름
- Pod의 볼륨 (스토리지)
- StatefulSet으로 Pod 생성 시, 0번부터 생성된다. ex) ns-nginx-0, ns-nginx-1, ns-nginx-2
StatefulSet Definition
StatefulSet은 serviceName이라는 Field를 가지고 있다.
예시
# statefulset-exam.yaml 파일 작성
root@master:~/Getting-Start-Kubernetes/6# cat statefulset-exam.yaml
apiVersion: apps/v1
kind: StatefulSet
metadata:
name: sf-nginx
spec:
replicas: 3
serviceName: sf-service
# podManagementPolicy: OrderedReady
podManagementPolicy: Parallel
selector:
matchLabels:
app: webui
template:
metadata:
name: nginx-pod
labels:
app: webui
spec:
containers:
- name: nginx-container
*podManagedPolicy : OrderedReady (Default 값)
pod 생성 시, 순차적으로 0번 만드는게 성공하면 1번 생성, 1번 만드는게 성공하면 2번 생성 …
*podManagedPolicy : Parallel
pod 생성 시, 0번 1번 2번을 동시에 생성
# 생성 및 확인
root@master:~/Getting-Start-Kubernetes/6# kubectl create -f statefulset-exam.yaml
statefulset.apps/sf-nginx created
root@master:~/Getting-Start-Kubernetes/6# kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
sf-nginx-0 1/1 Running 0 3s 192.168.104.8 node2 <none> <none>
sf-nginx-1 1/1 Running 0 3s 192.168.104.7 node2 <none> <none>
sf-nginx-2 1/1 Running 0 3s 192.168.166.144 node1 <none> <none>
# 삭제 후 확인
root@master:~/Getting-Start-Kubernetes/6# kubectl delete pod sf-nginx-0
pod "sf-nginx-0" deleted
root@master:~/Getting-Start-Kubernetes/6# ^C
root@master:~/Getting-Start-Kubernetes/6# kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
sf-nginx-0 1/1 Running 0 1s 192.168.104.9 node2 <none> <none>
sf-nginx-1 1/1 Running 0 85s 192.168.104.7 node2 <none> <none>
sf-nginx-2 1/1 Running 0 85s 192.168.166.144 node1 <none> <none>
sf-nginx-0 pod를 삭제해도, 똑같은 이름으로 다시 생성된다.
왜냐면 StatefulSet의 속성으로 pod의 이름을 보장해주기 때문이다. ex) sf-nginx-XX
즉 0번이 지워지면 0번이 다시 생성된다.
DaemonSet과 다른 점은, StatefulSet은 node당 한 개의 pod만 보장하지 않는다는 점이다.
위의 예시와 같이 한 node에 두 개의 pod가 생성될 수 있다.
다른 Controller와 다른 점은, pod name을 지정할 수 있다는 점이다.
예를 들면 ReplicationController로 생성시 pod name이 random hash 값으로 붙는다. ex) sf-nginx-skdjfd
예시2 (daemonSet scaling)
# DaemonSet scale in
root@master:~/Getting-Start-Kubernetes/6# kubectl scale statefulset sf-nginx --replicas=4
statefulset.apps/sf-nginx scaled
replicas 수를 3에서 4로 바꿔보자
# 확인
root@master:~/Getting-Start-Kubernetes/6# kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
sf-nginx-0 1/1 Running 0 7m2s 192.168.104.9 node2 <none> <none>
sf-nginx-1 1/1 Running 0 8m26s 192.168.104.7 node2 <none> <none>
sf-nginx-2 1/1 Running 0 8m26s 192.168.166.144 node1 <none> <none>
sf-nginx-3 1/1 Running 0 98s 192.168.166.147 node1 <none> <none>
# DaemonSet scale out
root@master:~/Getting-Start-Kubernetes/6# kubectl scale statefulset sf-nginx --replicas=2
statefulset.apps/sf-nginx scaled
# 확인
root@master:~/Getting-Start-Kubernetes/6# kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
sf-nginx-0 1/1 Running 0 8m5s 192.168.104.9 node2 <none> <none>
sf-nginx-1 1/1 Running 0 9m29s 192.168.104.7 node2 <none> <none>
pod 2, 3번이 삭제되었다.
예시2 (daemonSet Rolling update)
# kubectl edit 으로 이미지 버전 수정
root@master:~/Getting-Start-Kubernetes/6# kubectl edit statefulset.apps sf-nginx
# Please edit the object below. Lines beginning with a '#' will be ignored,
# and an empty file will abort the edit. If an error occurs while saving this file will be
# reopened with the relevant failures.
#
apiVersion: apps/v1
kind: StatefulSet
metadata:
creationTimestamp: "2024-12-05T07:53:49Z"
generation: 4
name: sf-nginx
namespace: default
resourceVersion: "245455"
uid: ef624b46-e2dc-402f-b614-c742c1667d3a
spec:
persistentVolumeClaimRetentionPolicy:
whenDeleted: Retain
whenScaled: Retain
podManagementPolicy: Parallel
replicas: 2
revisionHistoryLimit: 10
selector:
matchLabels:
app: webui
serviceName: sf-service
template:
metadata:
creationTimestamp: null
labels:
app: webui
name: nginx-pod
spec:
containers:
- image: nginx:1.15
imagePullPolicy: IfNotPresent
name: nginx-container
resources: {}
terminationMessagePath: /dev/termination-log
terminationMessagePolicy: File
dnsPolicy: ClusterFirst
restartPolicy: Always
schedulerName: default-scheduler
securityContext: {}
terminationGracePeriodSeconds: 30
updateStrategy:
rollingUpdate:
partition: 0
type: RollingUpdate
status:
availableReplicas: 2
collisionCount: 0
currentReplicas: 2
currentRevision: sf-nginx-7c9755b47
observedGeneration: 4
readyReplicas: 2
replicas: 2
updateRevision: sf-nginx-7c9755b47
updatedReplicas: 2
# 확인
root@master:~/Getting-Start-Kubernetes/6# kubectl get pods
NAME READY STATUS RESTARTS AGE
sf-nginx-0 1/1 Running 0 7s
sf-nginx-1 1/1 Running 0 10sroot@master:~/Getting-Start-Kubernetes/6# kubectl describe pod sf-nginx-0
Name: sf-nginx-0
Namespace: default
Priority: 0
Service Account: default
Node: node2/10.100.0.102
Start Time: Thu, 05 Dec 2024 08:07:13 +0000
Labels: app=webui
apps.kubernetes.io/pod-index=0
controller-revision-hash=sf-nginx-7c9755b47
statefulset.kubernetes.io/pod-name=sf-nginx-0
Annotations: cni.projectcalico.org/containerID: 7eedc38f202d27206458bf789687d1346cbb9b617247a95f24c30f4cb1c3f171
cni.projectcalico.org/podIP: 192.168.104.10/32
cni.projectcalico.org/podIPs: 192.168.104.10/32
Status: Running
IP: 192.168.104.10
IPs:
IP: 192.168.104.10
Controlled By: StatefulSet/sf-nginx
Containers:
nginx-container:
Container ID: containerd://d00eb9a5f3515070ee685172a052b02529c9b52814365b98e674ead1d9846e2a
Image: nginx:1.15
Image ID: docker.io/library/nginx@sha256:23b4dcdf0d34d4a129755fc6f52e1c6e23bb34ea011b315d87e193033bcd1b68
Port: <none>
Host Port: <none>
State: Running
Started: Thu, 05 Dec 2024 08:07:13 +0000
Ready: True
Restart Count: 0
Environment: <none>
Mounts:
/var/run/secrets/kubernetes.io/serviceaccount from kube-api-access-mbjzp (ro)
Conditions:
Type Status
PodReadyToStartContainers True
Initialized True
Ready True
ContainersReady True
PodScheduled True
Volumes:
kube-api-access-mbjzp:
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 26s default-scheduler Successfully assigned default/sf-nginx-0 to node2
Normal Pulled 26s kubelet Container image "nginx:1.15" already present on machine
Normal Created 26s kubelet Created container nginx-container
Normal Started 26s kubelet Started container nginx-container
daemonSet도 Rolling update가 가능하다.
# rollback
root@master:~/Getting-Start-Kubernetes/6# kubectl rollout undo statefulset sf-nginx
statefulset.apps/sf-nginx rolled backroot@master:~/Getting-Start-Kubernetes/6# kubectl get pods
NAME READY STATUS RESTARTS AGE
sf-nginx-0 1/1 Running 0 40s
sf-nginx-1 1/1 Running 0 42sroot@master:~/Getting-Start-Kubernetes/6# kubectl describe pod sf-nginx-0
Name: sf-nginx-0
Namespace: default
Priority: 0
Service Account: default
Node: node2/10.100.0.102
Start Time: Thu, 05 Dec 2024 08:10:07 +0000
Labels: app=webui
apps.kubernetes.io/pod-index=0
controller-revision-hash=sf-nginx-66f56bb7cb
statefulset.kubernetes.io/pod-name=sf-nginx-0
Annotations: cni.projectcalico.org/containerID: 73b20e501a203566ff88ebe715b5f4a26c3949fa97aca2e87f4b044e4b86074e
cni.projectcalico.org/podIP: 192.168.104.11/32
cni.projectcalico.org/podIPs: 192.168.104.11/32
Status: Running
IP: 192.168.104.11
IPs:
IP: 192.168.104.11
Controlled By: StatefulSet/sf-nginx
Containers:
nginx-container:
Container ID: containerd://7f4f2d1055da19e23fff29bea7cc41032a25823fb676aa073d683757cecb12dc
Image: nginx:1.14
Image ID: docker.io/library/nginx@sha256:f7988fb6c02e0ce69257d9bd9cf37ae20a60f1df7563c3a2a6abe24160306b8d
Port: <none>
Host Port: <none>
State: Running
Started: Thu, 05 Dec 2024 08:10:08 +0000
Ready: True
Restart Count: 0
Environment: <none>
Mounts:
/var/run/secrets/kubernetes.io/serviceaccount from kube-api-access-p7h99 (ro)
Conditions:
Type Status
PodReadyToStartContainers True
Initialized True
Ready True
ContainersReady True
PodScheduled True
Volumes:
kube-api-access-p7h99:
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 44s default-scheduler Successfully assigned default/sf-nginx-0 to node2
Normal Pulled 43s kubelet Container image "nginx:1.14" already present on machine
Normal Created 43s kubelet Created container nginx-container
Normal Started 43s kubelet Started container nginx-container
nginx:1.14로 rollback이 된 것을 확인 가능하다.
참고 영상
https://www.youtube.com/watch?v=Mx3y9un1KeI