Katacoda在线课：Liveness and Readiness Healthchecks
本系列教程希望能通过交互式学习网站与传统方式结合，更高效、容易的学习知识。 本系列教程将使用 Katacoda在线学习平台 完成学习。
在此场景中，您将了解 Kubernetes 如何使用 Readiness and Liveness Probes 检查容器运行状况。
Readiness Probe 检查应用是否准备好开始处理流量。此探针解决了容器已启动的问题，但该进程仍在预热和配置自身，这意味着它尚未准备好接收流量。
Liveness Probe 确保应用程序健康并能够处理请求。
启动集群
首先，我们需要启动一个 Kubernetes 集群。
执行以下命令启动集群组件并下载 Kubectl CLI
controlplane $ launch.sh
Waiting for Kubernetes to start...
Kubernetes started
集群启动后，使用 kubectl apply -f deploy.yaml 部署演示应用程序。
deploy.yaml
kind: List
apiVersion: v1
items:
- kind: ReplicationController
apiVersion: v1
metadata:
name: frontend
labels:
name: frontend
spec:
replicas: 1
selector:
name: frontend
template:
metadata:
labels:
name: frontend
spec:
containers:
- name: frontend
image: katacoda/docker-http-server:health
readinessProbe:
httpGet:
path: /
port: 80
initialDelaySeconds: 1
timeoutSeconds: 1
livenessProbe:
httpGet:
path: /
port: 80
initialDelaySeconds: 1
timeoutSeconds: 1
- kind: ReplicationController
apiVersion: v1
metadata:
name: bad-frontend
labels:
name: bad-frontend
spec:
replicas: 1
selector:
name: bad-frontend
template:
metadata:
labels:
name: bad-frontend
spec:
containers:
- name: bad-frontend
image: katacoda/docker-http-server:unhealthy
readinessProbe:
httpGet:
path: /
port: 80
initialDelaySeconds: 1
timeoutSeconds: 1
livenessProbe:
httpGet:
path: /
port: 80
initialDelaySeconds: 1
timeoutSeconds: 1
- kind: Service
apiVersion: v1
metadata:
labels:
app: frontend
kubernetes.io/cluster-service: "true"
name: frontend
spec:
type: NodePort
ports:
- port: 80
nodePort: 30080
selector:
app: frontend
controlplane $ kubectl apply -f deploy.yaml
replicationcontroller/frontend created
replicationcontroller/bad-frontend created
service/frontend created
Readiness Probe
在部署集群时，还部署了两个 Pod 来演示健康检查。您可以使用 cat deploy.yaml 查看部署。
livenessProbe:
httpGet:
path: /
port: 80
initialDelaySeconds: 1
timeoutSeconds: 1
可以根据您的应用程序更改设置来调用不同的端点，例如 /ping。
获取状态
第一个 Pod bad-frontend 是一个 HTTP 服务，它总是返回 500 错误，表明它没有正确启动。您可以使用 kubectl get pods --selector="name=bad-frontend" 命令查看 Pod 的状态。
controlplane $ kubectl get pods --selector="name=bad-frontend"
NAME READY STATUS RESTARTS AGE
bad-frontend-jk5z2 0/1 Running 2 78s
Kubectl 将返回使用我们的特定标签部署的 Pod。因为健康检查失败，它会显示没有容器为准备就绪的状态，同时它还将显示容器的重启尝试次数。
要了解有关失败原因的更多详细信息，请对该 Pod 使用描述命令。
controlplane $ pod=$(kubectl get pods --selector="name=bad-frontend" --output=jsonpath={.items..metadata.name})
controlplane $ kubectl describe pod $pod
Name: bad-frontend-jk5z2
Namespace: default
Priority: 0
PriorityClassName: <none>
Node: controlplane/172.17.0.67
Start Time: Fri, 23 Jul 2021 05:54:59 +0000
Labels: name=bad-frontend
Annotations: <none>
Status: Running
IP: 10.32.0.6
Controlled By: ReplicationController/bad-frontend
Containers:
bad-frontend:
Container ID: docker://811c3fa6d76c13f4b5bc7a6b2d6f514292e9673be46572a79b8f2d3d5e36bc62
Image: katacoda/docker-http-server:unhealthy
Image ID: docker-pullable://katacoda/docker-http-server@sha256:bea95c69c299c690103c39ebb3159c39c5061fee1dad13aa1b0625e0c6b52f22
Port: <none>
Host Port: <none>
State: Waiting
Reason: CrashLoopBackOff
Last State: Terminated
Reason: Error
Exit Code: 2
Started: Fri, 23 Jul 2021 05:57:07 +0000
Finished: Fri, 23 Jul 2021 05:57:37 +0000
Ready: False
Restart Count: 4
Liveness: http-get http://:80/ delay=1s timeout=1s period=10s #success=1 #failure=3
Readiness: http-get http://:80/ delay=1s timeout=1s period=10s #success=1 #failure=3
Environment: <none>
Mounts:
/var/run/secrets/kubernetes.io/serviceaccount from default-token-5n24z (ro)
Conditions:
Type Status
Initialized True
Ready False
ContainersReady False
PodScheduled True
Volumes:
default-token-5n24z:
Type: Secret (a volume populated by a Secret)
SecretName: default-token-5n24z
Optional: false
QoS Class: BestEffort
Node-Selectors: <none>
Tolerations: node.kubernetes.io/not-ready:NoExecute for 300s
node.kubernetes.io/unreachable:NoExecute for 300s
Events:
Type Reason Age From Message
---- ------ ---- ---- -------
Normal Scheduled 2m53s default-scheduler Successfully assigned default/bad-frontend-jk5z2 to controlplane
Normal Pulling 2m52s kubelet, controlplane Pulling image "katacoda/docker-http-server:unhealthy"
Normal Pulled 2m45s kubelet, controlplane Successfully pulled image "katacoda/docker-http-server:unhealthy"
Normal Created 105s (x3 over 2m45s) kubelet, controlplane Created container bad-frontend
Normal Started 105s (x3 over 2m45s) kubelet, controlplane Started container bad-frontend
Warning Unhealthy 105s (x6 over 2m35s) kubelet, controlplane Liveness probe failed: HTTP probe failed with statuscode: 500
Normal Killing 105s (x2 over 2m15s) kubelet, controlplane Container bad-frontend failed liveness probe, will be restarted
Normal Pulled 105s (x2 over 2m15s) kubelet, controlplane Container image "katacoda/docker-http-server:unhealthy" already present on machine
Warning Unhealthy 103s (x7 over 2m43s) kubelet, controlplane Readiness probe failed: HTTP probe failed with statuscode: 500
Readiness
我们的第二个 Pod，frontend，在启动时返回 OK 状态。
controlplane $ kubectl get pods --selector="name=frontend"
NAME READY STATUS RESTARTS AGE
frontend-54czd 1/1 Running 0 4m9s
Liveness Probe
由于我们的第二个 Pod 当前处于健康状态，我们可以模拟发生的故障。
目前，该 Pod 应该没有发生崩溃。
controlplane $ kubectl get pods --selector="name=frontend"
NAME READY STATUS RESTARTS AGE
frontend-54czd 1/1 Running 0 7m17s
让服务崩溃
HTTP 服务器有一个额外的端点，这将导致它返回 500 错误。使用 kubectl exec 可以调用端点。
controlplane $ pod=$(kubectl get pods --selector="name=frontend" --output=jsonpath={.items..metadata.name})
controlplane $ kubectl exec $pod -- /usr/bin/curl -s localhost/unhealthy
Liveness
Kubernetes 将根据配置执行 Liveness Probe。如果探测器失败，Kubernetes 将销毁并重新创建失败的容器。执行上面的命令使服务崩溃并观察 Kubernetes 自动恢复它。
controlplane $ kubectl get pods --selector="name=frontend"
NAME READY STATUS RESTARTS AGE
frontend-54czd 1/1 Running 1 9m42s
检查可能需要一些时间才能检测到。