成都酒店网站建设,怎样在微信上开发小程序,wordpress+广告联盟,秒玩大型游戏的网站Pod配置以及生命周期
1. Pod结构定义
每个pod中都可以包含一个或多个容器#xff0c;这些容器可以分为两类
用户程序所在的容器#xff0c;数量可多可少Pause容器#xff0c;这是每个Pod都会有的一个根容器#xff0c;它的作用有两个 可以以它为根据#xff0c;评估整个…Pod配置以及生命周期
1. Pod结构定义
每个pod中都可以包含一个或多个容器这些容器可以分为两类
用户程序所在的容器数量可多可少Pause容器这是每个Pod都会有的一个根容器它的作用有两个 可以以它为根据评估整个pod的健康状态可以在根容器上设置IP地址其他容器都以此IPPod IP以实现Pod内部的网络通信
[rootmaster ~]# kubectl get po -o wide --namespacedev
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
nginx 1/1 Running 0 6d14h 10.244.1.6 node1 none none[rootnode1 ~]# docker ps |grep nginx
a26eb43e2a6b 98ebf73aba75 nginx -g daemon of… 6 days ago Up 6 days k8s_pod_nginx_dev_7229a25a-7b87-49cd-a8e7-e0f2ef83924b_0
b46b6ff6da7e registry.aliyuncs.com/google_containers/pause:3.6 /pause 6 days ago Up 6 days k8s_POD_nginx_dev_7229a25a-7b87-49cd-a8e7-e0f2ef83924b_0可以看到创建的nginx pod里面有两个容器
1.1 定义
在k8s中所有资源的一级属性都是一样的主要包含5部分
apiVersion string 版本由k8s内部定义版本号可以用 kubectl api-versions查询到kind string 类型由k8s内部定义可以用 kubectl api-resources查询到metadata Object 元数据主要是资源标识和说明常用的有namenamespacelabels等spec Object 描述这是配置中最重要的一部分里面是对各种资源配置的详细描述status Object 状态信息里面的内容不需要定义由k8s自动生成
在上面的属性中spec是重点
containers []Object 容器列表用于定义容器的详细信息nodeName String 根据nodename的值将pod调度到指定的Node节点上nodeSelector map[] 根据NodeSelector中定义的信息选择将该Pod调度到包含这些label的node上hostNetwork boolean 是否使用主机网络模式默认为false如果设置为true表示使用宿主机网络volumes []Object 存储卷用于定义Pod上面挂载的存储信息restartPolicy string 重启策略表示Pod在遇到故障的时候的处理策略
1.1.1 pod定义
apiVersion: v1 #必选版本号例如v1
kind: Pod #必选资源类型例如 Pod
metadata: #必选元数据name: string #必选Pod名称namespace: string #Pod所属的命名空间,默认为defaultlabels: #自定义标签列表- name: string
spec: #必选Pod中容器的详细定义containers: #必选Pod中容器列表- name: string #必选容器名称image: string #必选容器的镜像名称imagePullPolicy: [ Always|Never|IfNotPresent ] #获取镜像的策略 command: [string] #容器的启动命令列表如不指定使用打包时使用的启动命令args: [string] #容器的启动命令参数列表workingDir: string #容器的工作目录volumeMounts: #挂载到容器内部的存储卷配置- name: string #引用pod定义的共享存储卷的名称需用volumes[]部分定义的的卷名mountPath: string #存储卷在容器内mount的绝对路径应少于512字符readOnly: boolean #是否为只读模式ports: #需要暴露的端口库号列表- name: string #端口的名称containerPort: int #容器需要监听的端口号hostPort: int #容器所在主机需要监听的端口号默认与Container相同protocol: string #端口协议支持TCP和UDP默认TCPenv: #容器运行前需设置的环境变量列表- name: string #环境变量名称value: string #环境变量的值resources: #资源限制和请求的设置limits: #资源限制的设置cpu: string #Cpu的限制单位为core数将用于docker run --cpu-shares参数memory: string #内存限制单位可以为Mib/Gib将用于docker run --memory参数requests: #资源请求的设置cpu: string #Cpu请求容器启动的初始可用数量memory: string #内存请求,容器启动的初始可用数量lifecycle: #生命周期钩子postStart: #容器启动后立即执行此钩子,如果执行失败,会根据重启策略进行重启preStop: #容器终止前执行此钩子,无论结果如何,容器都会终止livenessProbe: #对Pod内各容器健康检查的设置当探测无响应几次后将自动重启该容器exec: #对Pod容器内检查方式设置为exec方式command: [string] #exec方式需要制定的命令或脚本httpGet: #对Pod内个容器健康检查方法设置为HttpGet需要制定Path、portpath: stringport: numberhost: stringscheme: stringHttpHeaders:- name: stringvalue: stringtcpSocket: #对Pod内个容器健康检查方式设置为tcpSocket方式port: numberinitialDelaySeconds: 0 #容器启动完成后首次探测的时间单位为秒timeoutSeconds: 0 #对容器健康检查探测等待响应的超时时间单位秒默认1秒periodSeconds: 0 #对容器监控检查的定期探测时间设置单位秒默认10秒一次successThreshold: 0failureThreshold: 0securityContext:privileged: falserestartPolicy: [Always | Never | OnFailure] #Pod的重启策略nodeName: string #设置NodeName表示将该Pod调度到指定到名称的node节点上nodeSelector: obeject #设置NodeSelector表示将该Pod调度到包含这个label的node上imagePullSecrets: #Pull镜像时使用的secret名称以keysecretkey格式指定- name: stringhostNetwork: false #是否使用主机网络模式默认为false如果设置为true表示使用宿主机网络volumes: #在该pod上定义共享存储卷列表- name: string #共享存储卷名称 volumes类型有很多种emptyDir: {} #类型为emtyDir的存储卷与Pod同生命周期的一个临时目录。为空值hostPath: string #类型为hostPath的存储卷表示挂载Pod所在宿主机的目录path: string #Pod所在宿主机的目录将被用于同期中mount的目录secret: #类型为secret的存储卷挂载集群与定义的secret对象到容器内部scretname: string items: - key: stringpath: stringconfigMap: #类型为configMap的存储卷挂载预定义的configMap对象到容器内部name: stringitems:- key: stringpath: string
可以发现pod的资源非常繁多因此要一个一个记住是不现实的
所以k8s提供了能够查看每种资源的配置项的命令
#kubectl explain 资源类型 #查看某种资源可以配置的一级属性
#kubectl explain 资源类型.属性 #查看属性的子属性查看pod资源的一级属性
[rootmaster ~]# kubectl explain pod
KIND: Pod
VERSION: v1DESCRIPTION:Pod is a collection of containers that can run on a host. This resource iscreated by clients and scheduled onto hosts.FIELDS:apiVersion stringAPIVersion defines the versioned schema of this representation of anobject. Servers should convert recognized schemas to the latest internalvalue, and may reject unrecognized values. More info:https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resourceskind stringKind is a string value representing the REST resource this objectrepresents. Servers may infer this from the endpoint the client submitsrequests to. Cannot be updated. In CamelCase. More info:https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kindsmetadata ObjectStandard objects metadata. More info:https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#metadataspec ObjectSpecification of the desired behavior of the pod. More info:https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#spec-and-statusstatus ObjectMost 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
查看pod资源的metadata的子属性
[rootmaster ~]# kubectl explain pod.metadata
KIND: Pod
VERSION: v1RESOURCE: metadata ObjectDESCRIPTION:Standard objects metadata. More info:https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#metadataObjectMeta is metadata that all persisted resources must have, whichincludes all objects users must create.FIELDS:annotations map[string]stringAnnotations is an unstructured key value map stored with a resource thatmay be set by external tools to store and retrieve arbitrary metadata. Theyare not queryable and should be preserved when modifying objects. Moreinfo: http://kubernetes.io/docs/user-guide/annotationsclusterName stringThe name of the cluster which the object belongs to. This is used todistinguish resources with same name and namespace in different clusters.This field is not set anywhere right now and apiserver is going to ignoreit if set in create or update request.......查看pod.spec.containers属性
[rootmaster ~]# kubectl explain pod.spec.containers
KIND: Pod
VERSION: v1RESOURCE: containers []ObjectDESCRIPTION:List of containers belonging to the pod. Containers cannot currently beadded or removed. There must be at least one container in a Pod. Cannot beupdated.A single application container that you want to run within a pod.FIELDS:args []stringArguments to the entrypoint. The docker images CMD is used if this is notprovided. Variable references $(VAR_NAME) are expanded using thecontainers environment. If a variable cannot be resolved, the reference inthe input string will be unchanged. Double $$ are reduced to a single $,which allows for escaping the $(VAR_NAME) syntax: i.e. $$(VAR_NAME) willproduce the string literal $(VAR_NAME). Escaped references will never beexpanded, regardless of whether the variable exists or not. Cannot beupdated. More info:https://kubernetes.io/docs/tasks/inject-data-application/define-command-argument-container/#running-a-command-in-a-shellcommand []stringEntrypoint array. Not executed within a shell. The docker imagesENTRYPOINT is used if this is not provided. Variable references $(VAR_NAME)are expanded using the containers environment. If a variable cannot beresolved, the reference in the input string will be unchanged. Double $$are reduced to a single $, which allows for escaping the $(VAR_NAME)syntax: i.e. $$(VAR_NAME) will produce the string literal $(VAR_NAME).Escaped references will never be expanded, regardless of whether thevariable exists or not. Cannot be updated. More info:https://kubernetes.io/docs/tasks/inject-data-application/define-command-argument-container/#running-a-command-in-a-shellenv []ObjectList of environment variables to set in the container. Cannot be updated.envFrom []ObjectList of sources to populate environment variables in the container. Thekeys defined within a source must be a C_IDENTIFIER. All invalid keys willbe reported as an event when the container is starting. When a key existsin multiple sources, the value associated with the last source will takeprecedence. Values defined by an Env with a duplicate key will takeprecedence. Cannot be updated.2. 简易使用
2.1 基本配置
创建pod-base.yaml文件内容如下
apiVersion: v1
kind: Pod
metadata: name: pod-basenamespace: devlabels:user: zhangsan
spec:containers:- name: nginximage: nginx:1.17.1- name: busyboximage: busybox:1.30上面定义了一个比较简单的Pod的配置里面有两个容器
nginx用1.17.1版本的nginx镜像创建nginx是一个轻量级web容器busybox用1.30版本的busybox镜像创建busybox是一个小巧的linux命令集合
[rootmaster pod]# kubectl create -f pod-base.yaml
pod/pod-base created
[rootmaster pod]# kubectl get pod -n dev
NAME READY STATUS RESTARTS AGE
nginx 1/1 Running 0 6d23h
pod-base 0/2 ContainerCreating 0 32s
[rootmaster pod]#
[rootmaster pod]# kubectl get pod -n dev
NAME READY STATUS RESTARTS AGE
nginx 1/1 Running 0 6d23h
pod-base 1/2 CrashLoopBackOff 1 (6s ago) 41s
[rootmaster pod]# kubectl get pod -n dev -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
nginx 1/1 Running 0 6d23h 10.244.1.6 node1 none none
pod-base 1/2 CrashLoopBackOff 5 (2m32s ago) 6m4s 10.244.2.3 node2 none none2.2 设置镜像拉取
创建pod-imagepullpolicy.yaml文件内容如下
apiVersion: v1
kind: Pod
metadata: name: pod-imagepullpolicynamespace: devlabels:user: ayanami
spec:containers:- name: nginximage: nginx:1.17.1imagePullPolicy: Always #用于设置镜像拉取策略- name: busyboximage: busybox:1.30imagePullPolicy用于设置镜像拉取策略k8s支持配置三种拉取策略
Always总是从远程仓库拉取镜像一直用远程IfNotPresent本地有则使用本地镜像本地没有则从远程仓库拉取镜像本地有则本地本地没有则远程Never只使用本地镜像从不去远程仓库拉取本地没有就报错一直使用本地
默认值说明
如果镜像TAG为具体版本号默认策略是IfNotPresent如果镜像TAG为latest最终版本默认策略是always
[rootmaster pod]# kubectl create -f pod-imagepullpolicy.yaml
pod/pod-imagepullpolicy created
[rootmaster pod]# kubectl get pod -n dev
NAME READY STATUS RESTARTS AGE
nginx 1/1 Running 0 6d23h
pod-base 1/2 CrashLoopBackOff 6 (4m1s ago) 10m
pod-imagepullpolicy 1/2 CrashLoopBackOff 2 (17s ago) 50s
[rootmaster pod]# kubectl get pod -n dev -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
nginx 1/1 Running 0 6d23h 10.244.1.6 node1 none none
pod-base 1/2 CrashLoopBackOff 6 (4m7s ago) 10m 10.244.2.3 node2 none none
pod-imagepullpolicy 1/2 CrashLoopBackOff 2 (23s ago) 56s 10.244.2.4 node2 none none2.3 启动命令
在前面的案例中一直有一个问题没有解决就是busybox容器一直没有成功运行那么到底是什么原因导致这个容器的故障呢
这是因为busybox并不是一个程序而是类似于一个工具类的集合k8s集群启动管理后它会自动关闭。解决方法就是让其一直在运行这就用到了command配置
创建pod-command.yaml文件内容如下
apiVersion: v1
kind: Pod
metadata: name: pod-commandnamespace: dev
spec:containers:- name: nginximage: nginx:1.17.1imagePullPolicy: Always #用于设置镜像拉取策略- name: busyboximage: busybox:1.30command: [/bin/sh,-c,touch /tmp/hello.txt;while true;do /bin/echo $(date %T) /tmp/hello.txt;sleep 3;done;]对上面命令的解释
/bin/sh,-c 使用sh执行命令
touch /tmp/hello.txt; 创建一个/tmp/hello.txt文件
while true;do /bin/echo $(data %T) /tmp/hello.txt;sleep 3;done; 每隔三秒向文件中写入当前时间
[rootmaster pod]# kubectl create -f pod-command.yaml
pod/pod-command created
[rootmaster pod]# kubectl get pod -n dev
NAME READY STATUS RESTARTS AGE
nginx 1/1 Running 0 6d23h
pod-base 1/2 CrashLoopBackOff 7 (2m16s ago) 13m
pod-command 2/2 Running 0 36s
pod-imagepullpolicy 1/2 CrashLoopBackOff 5 (49s ago) 4m9s
[rootmaster pod]# kubectl get pod -n dev -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
nginx 1/1 Running 0 6d23h 10.244.1.6 node1 none none
pod-base 1/2 CrashLoopBackOff 7 (2m19s ago) 13m 10.244.2.3 node2 none none
pod-command 2/2 Running 0 39s 10.244.1.8 node1 none none
pod-imagepullpolicy 1/2 CrashLoopBackOff 5 (52s ago) 4m12s 10.244.2.4 node2 none none进入容器查看文件:
[rootmaster pod]# kubectl exec pod-command -n dev -it -c busybox -- /bin/sh
/ # tail -f /tmp/hello.txt
13:41:43
13:41:46
13:41:49
13:41:52
13:41:55
13:41:58
13:42:01
13:42:04
13:42:07
13:42:10
13:42:13
13:42:16特别说明
通过上面发现command已经可以完成启动命令和传递参数的功能为什么这里还要提供一个args选项用于传递参数呢
这其实跟docker有关系k8s中的commandarg两项其实是实现覆盖DockerFile中的ENTRYPOINT的功能
如果command和args均没有写那么用DockerFile的配置如果command写了但args没有写那么DockerFile默认的配置会被忽略执行输入的command如果command没写但args写了那么DockerFile中配置的ENTRYPOINT的命令会被执行使用当前args的参数如果command和args都写了那么DockerFile的配置被忽略执行command并追加上args参数
2.4 环境变量
创建pod-env.yaml文件内容如下
apiVersion: v1
kind: Pod
metadata: name: pod-envnamespace: dev
spec:containers:- name: busyboximage: busybox:1.30command: [/bin/sh,-c,while true;do /bin/echo $(date %T) sleep 60;done;]env: #设置环境变量列表- name: usernamevalue: admin- name: passwordvalue: 123456[rootmaster pod]# kubectl create -f pod-env.yaml
pod/pod-env created
[rootmaster pod]# kubectl get pod -n dev
NAME READY STATUS RESTARTS AGE
nginx 1/1 Running 0 6d23h
pod-base 1/2 CrashLoopBackOff 8 (76s ago) 17m
pod-command 2/2 Running 0 4m47s
pod-env 1/1 Running 0 5s
pod-imagepullpolicy 1/2 CrashLoopBackOff 6 (2m8s ago) 8m20s进入容器
[rootmaster ~]# kubectl exec -it pod-env -n dev -c busybox -- /bin/sh
/ # echo $username
admin
/ # echo $password
123456
/ # exit2.5 端口设置
查看端口资源
[rootmaster pod]# kubectl explain pod.spec.containers.ports
KIND: Pod
VERSION: v1RESOURCE: ports []ObjectDESCRIPTION:List of ports to expose from the container. Exposing a port here gives thesystem additional information about the network connections a containeruses, but is primarily informational. Not specifying a port here DOES NOTprevent that port from being exposed. Any port which is listening on thedefault 0.0.0.0 address inside a container will be accessible from thenetwork. Cannot be updated.ContainerPort represents a network port in a single container.FIELDS:containerPort integer -required-Number of port to expose on the pods IP address. This must be a valid portnumber, 0 x 65536.hostIP stringWhat host IP to bind the external port to.hostPort integerNumber of port to expose on the host. If specified, this must be a validport number, 0 x 65536. If HostNetwork is specified, this must matchContainerPort. Most containers do not need this.name stringIf specified, this must be an IANA_SVC_NAME and unique within the pod. Eachnamed port in a pod must have a unique name. Name for the port that can bereferred to by services.protocol stringProtocol for port. Must be UDP, TCP, or SCTP. Defaults to TCP.接下来创建pod-ports.yaml
apiVersion: v1
kind: Pod
metadata: name: pod-portsnamespace: dev
spec:containers:- name: nginximage: nginx:1.17.1ports: - name: nginx-portcontainerPort: 80protocol: TCP[rootmaster pod]# kubectl create -f pod-ports.yaml
pod/pod-ports created
[rootmaster pod]# kubectl get pod -n dev
NAME READY STATUS RESTARTS AGE
nginx 1/1 Running 0 6d23h
pod-base 1/2 CrashLoopBackOff 9 (109s ago) 23m
pod-command 2/2 Running 0 10m
pod-env 1/1 Running 0 5m49s
pod-imagepullpolicy 1/2 CrashLoopBackOff 7 (2m41s ago) 14m
pod-ports 1/1 Running 0 18s[rootmaster pod]# kubectl get pod pod-ports -n dev -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
pod-ports 1/1 Running 0 37s 10.244.2.6 node2 none none查看详情
[rootmaster pod]# kubectl get pod pod-ports -n dev -o yaml
apiVersion: v1
kind: Pod
metadata:creationTimestamp: 2022-08-12T14:04:57Zname: pod-portsnamespace: devresourceVersion: 125110uid: 676d55e3-71e2-422d-93f5-4537b45ff543
spec:containers:- image: nginx:1.17.1imagePullPolicy: IfNotPresentname: nginxports:- containerPort: 80name: nginx-portprotocol: TCP2.6 资源配额
容器中的程序要运行肯定是要占用一定资源的比如cpu和内存等如果不对某个容器的资源做限制那么它就可能吃掉大量资源导致其他容器无法运行。
针对这种情况k8s提供了对内存和cpu的资源进行配额的机制这种机制主要通过resources选项实现它有两个子选项
limits用于限制运行时容器的最大占用资源当容器占用资源超过limits时会被终止并进行重启requests用于设置容器需要的最小资源如果环境资源不够容器将无法启动
创建pod-resources.yaml
apiVersion: v1
kind: Pod
metadata: name: pod-resourcesnamespace: dev
spec:containers:- name: nginximage: nginx:1.17.1resources: #资源配额limits: #限制资源(上限)cpu: 2 #cpu限制memory: 10Gi #内存限制requests: #请求资源(下限)cpu: 1memory: 10Mi #内存限制[rootmaster pod]# kubectl create -f pod-resources.yaml
pod/pod-resources created
[rootmaster pod]# kubectl get pod pod-resources -n dev
NAME READY STATUS RESTARTS AGE
pod-resources 1/1 Running 0 8s
[rootmaster pod]# kubectl describe pod pod-resources -n dev
Name: pod-resources
Namespace: dev
Priority: 0
Node: node1/192.168.200.102
Start Time: Fri, 12 Aug 2022 22:07:19 0800
Labels: none
Annotations: none
Status: Running
IP: 10.244.1.10
IPs:IP: 10.244.1.10
Containers:nginx:Container ID: docker://fa7247baf122da83cfd677186bf1d45f53ab7a36e6d40c7ce8fcdc3c3b0131eeImage: nginx:1.17.1Image ID: docker-pullable://nginxsha256:b4b9b3eee194703fc2fa8afa5b7510c77ae70cfba567af1376a573a967c03dbbPort: noneHost Port: noneState: RunningStarted: Fri, 12 Aug 2022 22:07:21 0800Ready: TrueRestart Count: 0Limits:cpu: 2memory: 10GiRequests:cpu: 1memory: 10MiEnvironment: noneMounts:/var/run/secrets/kubernetes.io/serviceaccount from kube-api-access-9vhq6 (ro)
Conditions:Type StatusInitialized True Ready True ContainersReady True PodScheduled True
Volumes:kube-api-access-9vhq6:Type: Projected (a volume that contains injected data from multiple sources)TokenExpirationSeconds: 3607ConfigMapName: kube-root-ca.crtConfigMapOptional: nilDownwardAPI: true
QoS Class: Burstable
Node-Selectors: none
Tolerations: node.kubernetes.io/not-ready:NoExecute opExists for 300snode.kubernetes.io/unreachable:NoExecute opExists for 300s
Events:Type Reason Age From Message---- ------ ---- ---- -------Normal Scheduled 16s default-scheduler Successfully assigned dev/pod-resources to node1Normal Pulled 2s kubelet Container image nginx:1.17.1 already present on machineNormal Created 2s kubelet Created container nginxNormal Started 2s kubelet Started container nginx3. 生命周期
我们一般将Pod对象从创建到终止的这段时间范围称为Pod的生命周期它主要包含下面的过程
pod创建过程运行初始化容器init container过程运行主容器main container 容器启动后钩子post start、容器终止前钩子pre stop容器的存活性检测liveness probe、就绪性检测readiness probe pod终止过程 我们首先了解下Pod的状态值我们可以通过kubectl explain pod.status命令来了解关于Pod的一些信息Pod的状态定义在PodStatus对象中其中有一个phase字段下面是phase的可能取值
挂起Pending:Pod信息已经提交给了集群但它尚未被调度完成或仍处于下载镜像的过程中。运行中Running:Pod已经被调度到某节点并且所有容器已经被Kubectl创建完成。成功Succeeded:Pod中的所有容器都已经成功终止并且不会被重启。失败Failed:所有容器都已经终止并且至少有一个容器是因为失败终止即容器返回了非0值的退出状态或被系统终止。未知Unknown:API Server无法正常获取到Pod对象的状态信息通常由于网络通信失败所导致。
除此之外PodStatus对象中还包含一个PodCondition的数组里面包含的属性有
lastProbeTime最后一次探测 Pod Condition 的时间戳。lastTransitionTime上次 Condition 从一种状态转换到另一种状态的时间。message上次 Condition 状态转换的详细描述。reasonCondition 最后一次转换的原因。statusCondition 状态类型可以为 “True”, “False”, and “Unknown”.typeCondition 类型包括以下方面 PodScheduledPod 已经被调度到其他 node 里ReadyPod 能够提供服务请求可以被添加到所有可匹配服务的负载平衡池中Initialized所有的init containers已经启动成功Unschedulable调度程序现在无法调度 Pod例如由于缺乏资源或其他限制ContainersReadyPod 里的所有容器都是 ready 状态
3.1 pod创建过程 用户通过kubectl或其他api客户端提交需要创建的pod信息给apiserverapiserver开始生成pod对象的信息并将信息存入etcd然后返回确认信息至客户端apiserver开始反映etcd中pod对象的变化其他组件使用watch机制来跟踪检查apiserver上的变动scheduler发现有新的pod对象要创建开始为pod分配主机并将结果信息更新至apiservernode节点上的kubelet发现有pod调度过来尝试调用docker启动容器并将结果返回送至apiserverapiserver将接收到的pod状态信息存入etcd中
3.2 终止过程
用户向apiserver发送删除pod对象的命令apiserver中的pod对象信息会随着时间的推移而更新在宽限期内默认30spod被视为dead将pod标记为terminating状态kubelet在监控到pod对象转为terminating状态的同时启动pod关闭进程端点控制器监控到pod对象的关闭行为时将其从所有匹配到此端点的service资源的端点列表移除如果当前pod对象定义了prestop钩子处理器则在其标记为terminating后即会以同步的方式启动执行pod对象中的容器进程收到停止信号宽限期结束后若pod中还存在仍在运行的进程那么pod对象会收到立即终止的信号kubelet请求apiserver将此pod资源的宽限期设置为0从而完成删除操作此时pod对于用户已不可见
3.3 初始化容器
初始化容器Init Container是在Pod的主容器启动之前要运行的容器主要是做一些主容器的前置工作可以是一个 它具有两大特征
初始化容器必须运行完成直至结束如果某个初始化容器运行失败那么Kubernetes需要重启它直至成功完成。初始化容器必须按照定义的顺序执行当且仅当前一个成功之后后面一个才能运行。
初始化容器有很多应用场景下面列出的是最常见的几种
提供主容器镜像中不具备的工具程序或自定义代码。初始化容器要先于应用容器串行启动并运行完成因此可用于延后应用容器的启动直至其依赖的条件得到满足。
案例
假设要以主容器来运行Nginx,但是要求在运行Nginx之前要能够连接上MySQL和Redis所在的服务器。
为了简化测试提前规定好MySQL和Redis所在的IP地址分别为192.168.111.101,192.168.111.102注意这两个IP都不能ping通
创建pod-initcontainer.yaml文件
apiVersion: v1
kind: Pod
metadata:name: pod-initcontainernamespace: devlabels:user: zhangsan
spec:containers:- name: nginx # 容器名称image: nginx:1.17.1 # 容器需要的镜像地址imagePullPolicy: IfNotPresent # 设置镜像的拉取策略ports:- name: nginx-port # 端口名称如果执行必须保证name在Pod中是唯一的containerPort: 80 # 容器要监听的端口protocol: TCP # 端口协议resources: # 资源配额limits: # 限制资源的上限cpu: 2 # CPU限制单位是core数memory: 10Gi # 内存限制requests: # 限制资源的下限cpu: 1memory: 10MiinitContainers: # 初始化容器配置- name: test-mysqlimage: busybox:1.30command: [sh,-c,until ping 192.168.111.101 -c 1;do echo waiting for mysql ...;sleep 2;done;]securityContext:privileged: true # 使用特权模式运行- name: test-redisimage: busybox:1.30command: [sh,-c,until ping 192.168.111.102 -c 1;do echo waiting for redis ...;sleep 2; done;]
接下来启动并动态查看状态
[rootmaster pod]# kubectl create -f pod-initcontainer.yaml
pod/pod-initcontainer created
[rootmaster pod]# kubectl get pods pod-initcontainer -n dev -w
NAME READY STATUS RESTARTS AGE
pod-initcontainer 0/1 Init:0/2 0 31s我们发现Pod没有准备就绪状态是Init。
#查看具体的运行节点
[rootmaster pod]# kubectl get pods pod-initcontainer -n dev -w -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
pod-initcontainer 0/1 Init:0/2 0 13m 10.244.2.9 node2 none none我们给对应的节点的网卡添加IP:
ifconfig ens33:1 192.168.111.101 netmask 255.255.255.0 up
ifconfig ens33:2 192.168.111.102 netmask 255.255.255.0 up[rootmaster pod]# kubectl get pods pod-initcontainer -n dev -w
NAME READY STATUS RESTARTS AGE
pod-initcontainer 0/1 Init:0/2 0 13mpod-initcontainer 0/1 Init:1/2 0 14m
pod-initcontainer 0/1 PodInitializing 0 14m
pod-initcontainer 1/1 Running 0 14m3.4 Pod Hook
钩子函数能够感知自身生命周期中的事件并在相应的时刻到来时运行用户指定的程序代码。
k8s在主容器的启动之后和停止之前提供了两个钩子函数
post start容器创建之后执行如果失败了会重启容器主要用于资源部署、环境准备等。不过需要注意的是如果钩子花费太长时间以至于不能运行或者挂起容器将不能达到 running 状态。pre stop容器终止之前执行执行完成之后容器将成功终止在其完成之前会阻塞删除容器的操作主要用于优雅关闭应用程序、通知其他系统等。如果钩子在执行期间挂起Pod 阶段将停留在 running 状态并且永不会达到 failed 状态。
我们有三种方式来实现上面的钩子函数 Exec命令在容器内执行一次命令不过要注意的是该命令消耗的资源会被计入容器 ......lifecycle:postStart:exec:command:- cat- /tmp/healthy
...... TCPSocket在当前容器尝试访问指定的socket ......lifecycle:postStart:tcpSocket:port:8080
...... HttpGet在当前容器中向某url发起http请求 ......lifecycle:postStart:httpGet:path:/ # url地址port:80 # 端口号host:192.168.209.128 # 主机地址scheme: HTTP # 支持的协议http或https
......
示例
创建pod-hook-exec.yaml文件内容如下
apiVersion: v1
kind: Pod
metadata: name: pod-hook-execnamespace: dev
spec:containers:- name: main-containerimage: nginx:1.17.1ports: - name: nginx-portcontainerPort: 80lifecycle:postStart:exec: #在容器启动的时候执行一个命令修改掉nginx的默认首页内容command: [/bin/sh,-c,echo postStart... /usr/share/nginx/html/index.html]preStop: #在容器停止之前停止nginx服务exec:command: [/usr/sbin/nginx,-s,quit][rootmaster pod]# kubectl create -f pod-hook-exec.yaml
pod/pod-hook-exec created
[rootmaster pod]# kubectl get pod pod-hook-exec -n dev -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
pod-hook-exec 1/1 Running 0 8s 10.244.1.11 node1 none nonenode1节点
[rootnode1 ~]# curl 10.244.1.11
postStart...3.5 容器探测
容器探测用于检测容器中的应用实例是否能正常工作是保障业务可用性的一种传统机制。如果经过探测实例的状态不符合预期
那么K8S就会把该问题实例“摘除”不承担业务流量k8s提供了两种探针来实现容器探测分别是
liveness probes存活性探针用于检测应用实例当前是否处于正常运行状态如果不是k8s会重启容器readiness probes就绪性探针用于检测应用实例当前是否可以接受请求如果不能k8s不会转发流量
即livenessProbe决定是否重启容器readinesProbe决定是否将请求转发给容器 k8s在1.16版本之后新增了startupProbe探针用于判断容器内应用程序是否已经启动。如果配置了startupProbe探针就会先禁止其他的探针直到startupProbe探针成功为止一旦成功将不再进行探测。 上面两种探针目前均支持三种探测方式 Exec命令在容器内执行一次命令如果命令执行的退出码为0则认为程序正常否则不正常 ......livenessProbe:postStart:exec:command:- cat- /tmp/healthy
...... TCPSocket将会尝试访问同一个用户容器的端口如果能够建立这条连接则认为程序正常否则不正常 ......livenessProbe:tcpSocket:port: 8080
......HTTPGet调用容器内Web应用的URL如果返回的状态码在200和399之间则认为程序正常否则不正常 ......lifecycle:postStart:httpGet:path: #uri地址port:host: scheme: HTTP #支持的协议http或者https
......下面以liveness probes为例
3.5.1 EXEC
创建pod-liveness-exec.yaml
apiVersion: v1
kind: Pod
metadata: name: pod-liveness-execnamespace: dev
spec:containers:- name: main-containerimage: nginx:1.17.1ports: - name: nginx-portcontainerPort: 80livenessProbe:exec:command: [/bin/cat,/tmp/hello.txt] #执行一个查看文件的命令[rootnode1 ~]# kubectl create -f pod-liveness-exec.yaml
pod/pod-liveness-exec created
[rootnode1 ~]# kubectl get pod pod-liveness-exec -n dev
NAME READY STATUS RESTARTS AGE
pod-liveness-exec 1/1 Running 0 7s[rootmaster pod]# kubectl describe pod pod-liveness-exec -n dev
Events:Type Reason Age From Message---- ------ ---- ---- -------Normal Scheduled 62s default-scheduler Successfully assigned dev/pod-liveness-exec to node2Normal Pulled 1s (x3 over 61s) kubelet Container image nginx:1.17.1 already present on machineNormal Created 1s (x3 over 61s) kubelet Created container main-containerNormal Started 1s (x3 over 61s) kubelet Started container main-containerWarning Unhealthy 1s (x6 over 51s) kubelet Liveness probe failed: /bin/cat: /tmp/hello.txt: No such file or directoryNormal Killing 1s (x2 over 31s) kubelet Container main-container failed liveness probe, will be restarted由于没有此文件报错导致容器重启
修改其文件内容
apiVersion: v1
kind: Pod
metadata: name: pod-liveness-execnamespace: dev
spec:containers:- name: main-containerimage: nginx:1.17.1ports: - name: nginx-portcontainerPort: 80livenessProbe:exec:command: [/bin/ls,/tmp/] #执行一个查看文件的命令[rootmaster pod]# kubectl delete -f pod-liveness-exec.yaml
pod pod-liveness-exec deleted
[rootmaster pod]# kubectl create -f pod-liveness-exec.yaml
pod/pod-liveness-exec created
[rootmaster pod]# kubectl get pod pod-liveness-exec -n dev
NAME READY STATUS RESTARTS AGE
pod-liveness-exec 1/1 Running 0 7s
[rootmaster pod]# kubectl describe pod pod-liveness-exec -n dev
Name: pod-liveness-exec
Namespace: dev
Priority: 0
Node: node2/192.168.200.103
Start Time: Fri, 12 Aug 2022 23:49:20 0800
Labels: none
Annotations: none
Status: Running
IP: 10.244.2.11
IPs:IP: 10.244.2.11
Containers:main-container:Container ID: docker://a3a7c02ff4a6f391184e23d3929d8b0bddb7937638fa955be7e69d1f448b7443Image: nginx:1.17.1Image ID: docker-pullable://nginxsha256:b4b9b3eee194703fc2fa8afa5b7510c77ae70cfba567af1376a573a967c03dbbPort: 80/TCPHost Port: 0/TCPState: RunningStarted: Fri, 12 Aug 2022 23:49:20 0800Ready: TrueRestart Count: 0Liveness: exec [/bin/ls /tmp/] delay0s timeout1s period10s #success1 #failure3Environment: noneMounts:/var/run/secrets/kubernetes.io/serviceaccount from kube-api-access-s42xs (ro)
Conditions:Type StatusInitialized True Ready True ContainersReady True PodScheduled True
Volumes:kube-api-access-s42xs:Type: Projected (a volume that contains injected data from multiple sources)TokenExpirationSeconds: 3607ConfigMapName: kube-root-ca.crtConfigMapOptional: nilDownwardAPI: true
QoS Class: BestEffort
Node-Selectors: none
Tolerations: node.kubernetes.io/not-ready:NoExecute opExists for 300snode.kubernetes.io/unreachable:NoExecute opExists for 300s
Events:Type Reason Age From Message---- ------ ---- ---- -------Normal Scheduled 24s default-scheduler Successfully assigned dev/pod-liveness-exec to node2Normal Pulled 24s kubelet Container image nginx:1.17.1 already present on machineNormal Created 24s kubelet Created container main-containerNormal Started 24s kubelet Started container main-container3.5.2 TCPSocket
创建pod-liveness-tcpsocket.yaml
apiVersion: v1
kind: Pod
metadata: name: pod-liveness-tcpsocketnamespace: dev
spec:containers:- name: main-containerimage: nginx:1.17.1ports: - name: nginx-portcontainerPort: 80livenessProbe:tcpSocket:port: 8080 #尝试访问8080端口[rootmaster pod]# kubectl create -f pod-liveness-tcpsocket.yaml
pod/pod-liveness-tcpsocket created
[rootmaster pod]# kubectl get pod pod-liveness-tcpsocket -n dev
NAME READY STATUS RESTARTS AGE
pod-liveness-tcpsocket 1/1 Running 0 6s
[rootmaster pod]# kubectl describe pod pod-liveness-tcpsocket -n dev
Name: pod-liveness-tcpsocket
Namespace: dev
Priority: 0
Node: node1/192.168.200.102
Start Time: Fri, 12 Aug 2022 23:52:33 0800
Labels: none
Annotations: none
Status: Running
IP: 10.244.1.12
IPs:IP: 10.244.1.12
Containers:main-container:Container ID: docker://8e9e6ca1f11f06f80fe7d7b3e7a8f18c11bddc5568dff33041b77ad1e9c72088Image: nginx:1.17.1Image ID: docker-pullable://nginxsha256:b4b9b3eee194703fc2fa8afa5b7510c77ae70cfba567af1376a573a967c03dbbPort: 80/TCPHost Port: 0/TCPState: RunningStarted: Fri, 12 Aug 2022 23:52:34 0800Ready: TrueRestart Count: 0Liveness: tcp-socket :8080 delay0s timeout1s period10s #success1 #failure3Environment: noneMounts:/var/run/secrets/kubernetes.io/serviceaccount from kube-api-access-8glq7 (ro)
Conditions:Type StatusInitialized True Ready True ContainersReady True PodScheduled True
Volumes:kube-api-access-8glq7:Type: Projected (a volume that contains injected data from multiple sources)TokenExpirationSeconds: 3607ConfigMapName: kube-root-ca.crtConfigMapOptional: nilDownwardAPI: true
QoS Class: BestEffort
Node-Selectors: none
Tolerations: node.kubernetes.io/not-ready:NoExecute opExists for 300snode.kubernetes.io/unreachable:NoExecute opExists for 300s
Events:Type Reason Age From Message---- ------ ---- ---- -------Normal Scheduled 60s default-scheduler Successfully assigned dev/pod-liveness-tcpsocket to node1Normal Pulled 17s (x2 over 46s) kubelet Container image nginx:1.17.1 already present on machineNormal Created 17s (x2 over 46s) kubelet Created container main-containerNormal Started 17s (x2 over 46s) kubelet Started container main-containerNormal Killing 17s kubelet Container main-container failed liveness probe, will be restartedWarning Unhealthy invalid (x5 over 37s) kubelet Liveness probe failed: dial tcp 10.244.1.12:8080: connect: connection refused更改文件内容
apiVersion: v1
kind: Pod
metadata: name: pod-liveness-tcpsocketnamespace: dev
spec:containers:- name: main-containerimage: nginx:1.17.1ports: - name: nginx-portcontainerPort: 80livenessProbe:tcpSocket:port: 80 #尝试访问80端口[rootmaster pod]# kubectl delete -f pod-liveness-tcpsocket.yaml
pod pod-liveness-tcpsocket deleted
[rootmaster pod]# kubectl create -f pod-liveness-tcpsocket.yaml
pod/pod-liveness-tcpsocket created
[rootmaster pod]# kubectl get pod pod-liveness-tcpsocket -n dev
NAME READY STATUS RESTARTS AGE
pod-liveness-tcpsocket 1/1 Running 0 5s
[rootmaster pod]# kubectl describe pod pod-liveness-tcpsocket -n dev
Name: pod-liveness-tcpsocket
Namespace: dev
Priority: 0
Node: node1/192.168.200.102
Start Time: Fri, 12 Aug 2022 23:54:26 0800
Labels: none
Annotations: none
Status: Running
IP: 10.244.1.13
IPs:IP: 10.244.1.13
Containers:main-container:Container ID: docker://15e1a720e720432817081d3dd0e8a1b42b925e38514df53f2782de7d278a62c0Image: nginx:1.17.1Image ID: docker-pullable://nginxsha256:b4b9b3eee194703fc2fa8afa5b7510c77ae70cfba567af1376a573a967c03dbbPort: 80/TCPHost Port: 0/TCPState: RunningStarted: Fri, 12 Aug 2022 23:54:27 0800Ready: TrueRestart Count: 0Liveness: tcp-socket :80 delay0s timeout1s period10s #success1 #failure3Environment: noneMounts:/var/run/secrets/kubernetes.io/serviceaccount from kube-api-access-tcfk7 (ro)
Conditions:Type StatusInitialized True Ready True ContainersReady True PodScheduled True
Volumes:kube-api-access-tcfk7:Type: Projected (a volume that contains injected data from multiple sources)TokenExpirationSeconds: 3607ConfigMapName: kube-root-ca.crtConfigMapOptional: nilDownwardAPI: true
QoS Class: BestEffort
Node-Selectors: none
Tolerations: node.kubernetes.io/not-ready:NoExecute opExists for 300snode.kubernetes.io/unreachable:NoExecute opExists for 300s
Events:Type Reason Age From Message---- ------ ---- ---- -------Normal Scheduled 10s default-scheduler Successfully assigned dev/pod-liveness-tcpsocket to node1Normal Pulled invalid kubelet Container image nginx:1.17.1 already present on machineNormal Created invalid kubelet Created container main-containerNormal Started invalid kubelet Started container main-container3.5.3 HTTPGet
创建pod-liveness-httpget.yaml
apiVersion: v1
kind: Pod
metadata: name: pod-liveness-httpgetnamespace: dev
spec:containers:- name: main-containerimage: nginx:1.17.1ports: - name: nginx-portcontainerPort: 80livenessProbe:httpGet: #其实就是访问http://127.0.0.1:80/helloscheme: HTTP #支持的协议http或httpsport: 80path: /hello #uri地址[rootmaster pod]# kubectl create -f pod-liveness-httpget.yaml
pod/pod-liveness-httpget created
[rootmaster pod]# kubectl get pod pod-liveness-httpget -n dev
NAME READY STATUS RESTARTS AGE
pod-liveness-httpget 1/1 Running 0 4s
[rootmaster pod]# kubectl describe pod pod-liveness-httpget -n dev
Events:Type Reason Age From Message---- ------ ---- ---- -------Normal Scheduled 30s default-scheduler Successfully assigned dev/pod-liveness-httpget to node2Normal Pulled 0s (x2 over 29s) kubelet Container image nginx:1.17.1 already present on machineNormal Created 0s (x2 over 29s) kubelet Created container main-containerNormal Started 0s (x2 over 29s) kubelet Started container main-containerWarning Unhealthy 0s (x3 over 20s) kubelet Liveness probe failed: HTTP probe failed with statuscode: 404Normal Killing 0s kubelet Container main-container failed liveness probe, will be restarted可以看见pod在重启详细描述说明没找到这个网址
修改配置文件
apiVersion: v1
kind: Pod
metadata: name: pod-liveness-httpgetnamespace: dev
spec:containers:- name: main-containerimage: nginx:1.17.1ports: - name: nginx-portcontainerPort: 80livenessProbe:httpGet: #其实就是访问http://127.0.0.1:80/scheme: HTTP #支持的协议http或httpsport: 80path: / #uri地址[rootmaster pod]# kubectl create -f pod-liveness-httpget.yaml
pod/pod-liveness-httpget created
[rootmaster pod]# kubectl get pod pod-liveness-httpget -n dev
NAME READY STATUS RESTARTS AGE
pod-liveness-httpget 1/1 Running 0 6s
[rootmaster pod]# kubectl describe pod pod-liveness-httpget -n dev
Name: pod-liveness-httpget
Namespace: dev
Priority: 0
Node: node2/192.168.200.103
Start Time: Fri, 12 Aug 2022 23:57:29 0800
Labels: none
Annotations: none
Status: Running
IP: 10.244.2.13
IPs:IP: 10.244.2.13
Containers:main-container:Container ID: docker://77688fa183b12141506d94bcfb270bf258e692b12664dc6851821705fb549f8fImage: nginx:1.17.1Image ID: docker-pullable://nginxsha256:b4b9b3eee194703fc2fa8afa5b7510c77ae70cfba567af1376a573a967c03dbbPort: 80/TCPHost Port: 0/TCPState: RunningStarted: Fri, 12 Aug 2022 23:57:30 0800Ready: TrueRestart Count: 0Liveness: http-get http://:80/ delay0s timeout1s period10s #success1 #failure3Environment: noneMounts:/var/run/secrets/kubernetes.io/serviceaccount from kube-api-access-plf5d (ro)
Conditions:Type StatusInitialized True Ready True ContainersReady True PodScheduled True
Volumes:kube-api-access-plf5d:Type: Projected (a volume that contains injected data from multiple sources)TokenExpirationSeconds: 3607ConfigMapName: kube-root-ca.crtConfigMapOptional: nilDownwardAPI: true
QoS Class: BestEffort
Node-Selectors: none
Tolerations: node.kubernetes.io/not-ready:NoExecute opExists for 300snode.kubernetes.io/unreachable:NoExecute opExists for 300s
Events:Type Reason Age From Message---- ------ ---- ---- -------Normal Scheduled 21s default-scheduler Successfully assigned dev/pod-liveness-httpget to node2Normal Pulled 20s kubelet Container image nginx:1.17.1 already present on machineNormal Created 20s kubelet Created container main-containerNormal Started 20s kubelet Started container main-container3.5.4 其他
上面已经使用了livenessProbe演示了三种探测方式但是查看livenessProbe的子属性会发现除了这三种方式还有一些其他的配置。
[rootmaster pod]# kubectl explain pod.spec.containers.livenessProbe
KIND: Pod
VERSION: v1RESOURCE: livenessProbe ObjectDESCRIPTION:Periodic probe of container liveness. Container will be restarted if theprobe fails. Cannot be updated. More info:https://kubernetes.io/docs/concepts/workloads/pods/pod-lifecycle#container-probesProbe describes a health check to be performed against a container todetermine whether it is alive or ready to receive traffic.FIELDS:exec ObjectExec specifies the action to take.failureThreshold integerMinimum consecutive failures for the probe to be considered failed afterhaving succeeded. Defaults to 3. Minimum value is 1.grpc ObjectGRPC specifies an action involving a GRPC port. This is an alpha field andrequires enabling GRPCContainerProbe feature gate.httpGet ObjectHTTPGet specifies the http request to perform.initialDelaySeconds integerNumber of seconds after the container has started before liveness probesare initiated. More info:https://kubernetes.io/docs/concepts/workloads/pods/pod-lifecycle#container-probesperiodSeconds integerHow often (in seconds) to perform the probe. Default to 10 seconds. Minimumvalue is 1.successThreshold integerMinimum consecutive successes for the probe to be considered successfulafter having failed. Defaults to 1. Must be 1 for liveness and startup.Minimum value is 1.tcpSocket ObjectTCPSocket specifies an action involving a TCP port.terminationGracePeriodSeconds integerOptional duration in seconds the pod needs to terminate gracefully uponprobe failure. The grace period is the duration in seconds after theprocesses running in the pod are sent a termination signal and the timewhen the processes are forcibly halted with a kill signal. Set this valuelonger than the expected cleanup time for your process. If this value isnil, the pods terminationGracePeriodSeconds will be used. Otherwise, thisvalue overrides the value provided by the pod spec. Value must benon-negative integer. The value zero indicates stop immediately via thekill signal (no opportunity to shut down). This is a beta field andrequires enabling ProbeTerminationGracePeriod feature gate. Minimum valueis 1. spec.terminationGracePeriodSeconds is used if unset.timeoutSeconds integerNumber of seconds after which the probe times out. Defaults to 1 second.Minimum value is 1. More info:https://kubernetes.io/docs/concepts/workloads/pods/pod-lifecycle#container-probesinitialDelaySeconds 容器启动后等待多少秒执行第一次探测 timeoutSeconds 探测超时时间。默认1秒最小1秒periodSeconds 执行探测的频率。默认是10秒最小1秒failureThreshold 连续探测失败多少次才被认定为失败。默认是3。最小值是1successThreshold 连续探测成功多少次才被认定为成功。默认是1
initialDelaySeconds可以用来配置第一次执行探针的等待时间对于启动非常慢的应用这个参数非常有用比如 Jenkins、Gitlab 这类应用但是如何设置一个合适的初始延迟时间呢这个就和应用具体的环境有关系了所以这个值往往不是通用的这样的话可能就会导致一个问题我们的资源清单在别的环境下可能就会健康检查失败了为解决这个问题在 Kubernetes v1.16 版本官方特地新增了一个 startupProbe启动探针该探针将推迟所有其他探针直到 Pod 完成启动为止使用方法和存活探针一样
......startupProbe:httpGet:path: /healthzport: 8080failureThreshold: 30 # 尽量设置大点periodSeconds: 10
......
有的时候应用程序可能暂时无法对外提供服务例如应用程序可能需要在启动期间加载大量数据或配置文件。在这种情况下你不想杀死应用程序也不想对外提供服务。那么这个时候我们就可以使用readiness probe来检测和减轻这些情况。
Pod 中的容器可以报告自己还没有准备不能处理 Kubernetes 服务发送过来的流量。
readiness probe的配置跟liveness probe基本上一致的。唯一的不同是使用readinessProbe而不是livenessProbe。两者如果同时使用的话就可以确保流量不会到达还未准备好的容器准备好过后如果应用程序出现了错误则会重新启动容器。
3.6 重启策略
在容器探测中一旦容器探测出现了问题Kubernetes就会对容器所在的Pod进行重启其实这是由Pod的重启策略决定的Pod的重启策略有3种分别如下 Always容器失效时自动重启该容器默认值。 OnFailure容器终止运行且退出码不为0时重启。 Never不论状态如何都不重启该容器。
重启策略适用于Pod对象中的所有容器首次需要重启的容器将在其需要的时候立即进行重启随后再次重启的操作将由kubelet延迟一段时间后进行且反复的重启操作的延迟时长以此为10s、20s、40s、80s、160s和300s300s是最大的延迟时长。
创建pod-restart.yaml
apiVersion: v1
kind: Pod
metadata:name: pod-restartnamespace: devlabels:user: zhangsan
spec:containers:- name: nginx # 容器名称image: nginx:1.17.1 # 容器需要的镜像地址imagePullPolicy: IfNotPresent # 设置镜像的拉取策略ports:- name: nginx-port # 端口名称如果执行必须保证name在Pod中是唯一的containerPort: 80 # 容器要监听的端口protocol: TCP # 端口协议livenessProbe: # 声明周期配置httpGet:port: 80scheme: HTTPpath: /hellohost: 127.0.0.1restartPolicy: Never # 重启策略
[rootmaster pod]# kubectl create -f pod-restart.yaml
pod/pod-restart-policy created
[rootmaster pod]# kubectl describe pod pod-restart -n dev
Events:Type Reason Age From Message---- ------ ---- ---- -------Normal Scheduled 35s default-scheduler Successfully assigned dev/pod-restart-policy to node1Normal Pulled 22s kubelet Container image nginx:1.17.1 already present on machineNormal Created 22s kubelet Created container nginxNormal Started 22s kubelet Started container nginxWarning Unhealthy invalid (x3 over 12s) kubelet Liveness probe failed: Get http://127.0.0.1:80/hello: dial tcp 127.0.0.1:80: connect: connection refusedNormal Killing invalid kubelet Stopping container nginxestart namespace: dev labels: user: zhangsan spec: containers: - name: nginx # 容器名称 image: nginx:1.17.1 # 容器需要的镜像地址 imagePullPolicy: IfNotPresent # 设置镜像的拉取策略 ports:- name: nginx-port # 端口名称如果执行必须保证name在Pod中是唯一的containerPort: 80 # 容器要监听的端口protocol: TCP # 端口协议livenessProbe: # 声明周期配置httpGet:port: 80scheme: HTTPpath: /hellohost: 127.0.0.1restartPolicy: Never # 重启策略 ~~~shell
[rootmaster pod]# kubectl create -f pod-restart.yaml
pod/pod-restart-policy created
[rootmaster pod]# kubectl describe pod pod-restart -n dev
Events:Type Reason Age From Message---- ------ ---- ---- -------Normal Scheduled 35s default-scheduler Successfully assigned dev/pod-restart-policy to node1Normal Pulled 22s kubelet Container image nginx:1.17.1 already present on machineNormal Created 22s kubelet Created container nginxNormal Started 22s kubelet Started container nginxWarning Unhealthy invalid (x3 over 12s) kubelet Liveness probe failed: Get http://127.0.0.1:80/hello: dial tcp 127.0.0.1:80: connect: connection refusedNormal Killing invalid kubelet Stopping container nginx我们发现容器探测失败后直接停止容器并没有选择重启。
