kubernetes 1.3管中窺豹- RS（Replica Sets）：the next-generation Replication Controller

時間 2019-11-08

標籤 kubernetes 1.3 管中窺豹 replica sets generation replication controller 简体版

原文原文鏈接

前言

kubernates 1.3出了幾個新的概念，其中包括deployments，Replica Sets，而且官網稱之爲是the next-generation Replication Controller。因爲NCE項目立刻就要使用其中包括deployments以及RS這個方式來管理pod，所以有必要了解下它的優越性。nginx

回顧老版RC概念

說RC以前先要提一個container和pod，container就是docker中的容器，pod能夠理解爲一個或者一組container。pod中的container能夠進行共享資源。 RC-Replication Controller是一種資源對象，它能夠經過json模板來建立，經過模板中定義的pod模板（Templet）來建立相應的帶Lable的pod。RC經過包含的標籤選擇器（Label selector）來管理全部帶相應label的pod。下面是一個RC的定義模板文件：docker

"kind": "ReplicationController",
    "apiVersion": "v1",
    "metadata": {
        "name": "haohua2aa22-13275-c059ac0f",
        "namespace": "90ab293349704831aa3977c450235fe1",
        "selfLink": "/api/v1/namespaces/90ab293349704831aa3977c450235fe1/replicationcontrollers/haohua2aa22-13275-c059ac0f",
        "uid": "87ee2f1e-4f16-11e6-96bb-fa163e70fcd5",
        "resourceVersion": "9565340",
        "generation": 1,
        "creationTimestamp": "2016-07-21T07:41:26Z",
        "labels": {
            "name": "haohua2aa22-13275-c059ac0f"
        }
    },
    "spec": {
        "replicas": 1,
        "selector": {
            "name": "haohua2aa22-13275-c059ac0f"
        },
        "podCreatePolicy": "New",
        "template": {
            "metadata": {
                "name": "haohua2aa22-13275-c059ac0f",
                "namespace": "90ab293349704831aa3977c450235fe1",
                "creationTimestamp": null,
                "labels": {
                    "name": "haohua2aa22-13275-c059ac0f"
                }
            },
            "spec": {              
                "containers": [
                    {
                        ...(此處省略)
                        ],
                    }
                ],               
            }
    }

能夠看到這個RC裏面包含了一個容器的pod：haohua2aa22-13275-c059ac0f，該pod有1個副本（ "replicas": 1,）。 pod的模板文件中有一個labels標識pod是屬於哪一個rcjson

"labels": {
            "name": "haohua2aa22-13275-c059ac0f"
        },

RC內pod變動大概有幾種場景：api

1.經過設置副本個數能夠動態的調整pod資源，若是將replicas設置爲2，則控制器會自動去建立一個pod而且標籤也設置爲haohua2aa22-13275-c059ac0f。app

2.若是想刪除資源，則須要將replicas設置爲0，這時候控制器就會自動刪除pod副本資源。須要注意若是直接刪除RC，pod資源是不會被清理的。frontend

3.若是想更新service內rc中的pod，則須要用rolling Updates滾動更新模式，即逐個替換pod的方式來支持service的滾動更新，新建一個只有1個pod的rc，若新的rc副本數量加1，則舊的rc副本數減1，直到舊的rc副本數量爲0，則刪除舊的rc。ide

deployments&RS概念

deployments是用來爲pods和Replica Sets提供更新用的。Replica Sets被官網稱之爲下一代的RC，可見取而代之的決心，Replica Set 和Replication Controller僅有的區別是slector的支持，RS支持新的基於集合（ set-based）的選擇器，而RC只支持基於相等（ equality-based ）的選擇器。ui

這邊的兩個概念基於集合（ set-based）；基於相等（ equality-based ）怎麼來理解呢。spa

基於相等的就是說只能經過兩種操做來設置：code

environment = production
tier != frontend

基於集合的不僅是包含等於和不等於兩種規則，還能夠包含 in 和notin的方式來篩選。

environment in (production, qa)
tier notin (frontend, backend)
partition
!partition

下面來看看kubernetes 1.3的deployments是如何用模板文件定義的這是一個官網的3的deployments是如何用模板文件定義的例子，我在1.3的環境下試着用命令建立了1個deployment

apiVersion: extensions/v1beta1
kind: Deployment
metadata:
  name: nginx-deployment
spec:
  replicas: 3
  template:
    metadata:
      labels:
        app: nginx
    spec:
      containers:
      - name: nginx
        image: nginx:1.7.9
        ports:
        - containerPort: 80

kubectl create -f /root/cxq/jsonxml/nginx-deployment.yaml
deployment "nginx-deployment" created
root@qa-control-master2-cluster-21:~/cxq/jsonxml# kubectl get deployment
NAME               DESIRED   CURRENT   UP-TO-DATE   AVAILABLE   AGE
nginx-deployment   3         3         3            0           12s

這個時候查詢當前的rs，發現多了一個名字跟deployment有點關聯性的rs

root@qa-control-master2-cluster-21:~/cxq/jsonxml# kubectl get rs
NAME                          DESIRED   CURRENT   AGE
nginx-deployment-1159050644   3         3         8m

在後面隨機加了一串數字。檢查rs的json說明文件，看下跟以前的rc有什麼區別【重點在此，請注意】：

kubectl get rs -o json
{
    "kind": "List",
    "apiVersion": "v1",
    "metadata": {},
    "items": [
        {
            "kind": "ReplicaSet",
            "apiVersion": "extensions/v1beta1",
            "metadata": {
                "name": "nginx-deployment-1159050644",
                "namespace": "default",
                "selfLink": "/apis/extensions/v1beta1/namespaces/default/replicasets/nginx-deployment-1159050644",
                "uid": "57fd01cc-4f29-11e6-9cce-fa163ee959c4",
                "resourceVersion": "179452",
                "generation": 1,
                "creationTimestamp": "2016-07-21T09:56:06Z",
                "labels": {
                    "app": "nginx",
                    "pod-template-hash": "1159050644"
                },
                "annotations": {
                    "deployment.kubernetes.io/revision": "1"
                }
            },
            "spec": {
                "replicas": 3,
                "selector": {
                    "matchLabels": {
                        "app": "nginx",
                        "pod-template-hash": "1159050644"
                    }
                },
                "template": {
                    "metadata": {
                        "creationTimestamp": null,
                        "labels": {
                            "app": "nginx",
                            "pod-template-hash": "1159050644"
                        }
                    },
                    "spec": {
                        "containers": [
                            {
                            ...(此處省略)
                    }
                }
            },
        }
    ]
}

能夠看到labels的值不是隻有一條name：value的格式，而是有兩條。

"labels": {
                            "app": "nginx",
                            "pod-template-hash": "1159050644"
                        }

說明app=nginx或者包含nginx的均可以被篩選出來做爲rs的pod 查看pod，而且模擬selector的篩選方式實驗一下

當前系統有4個pod，其中前3個屬於新建的rs
 kubectl get pod -o wide
NAME                                READY     STATUS    RESTARTS   AGE       IP             NODE
nginx-deployment-1159050644-kg654   1/1       Running   0          4m        192.168.5.10   10.180.217.225
nginx-deployment-1159050644-x6wk8   1/1       Running   0          4m        192.168.5.8    10.180.217.225
nginx-deployment-1159050644-z2kqi   1/1       Running   0          3h        192.168.5.6    10.180.217.225
test-5-5tcql                        1/1       Running   0          1d        192.168.5.2    10.180.217.225

使用過濾器equality-based方法篩選，能夠準確的篩選出來3條app=nginx的pod

root@qa-control-master2-cluster-21:~/cxq/jsonxml#  kubectl get pod -o wide -l  app=nginx
NAME                                READY     STATUS    RESTARTS   AGE       IP             NODE
nginx-deployment-1159050644-kg654   1/1       Running   0          5m        192.168.5.10   10.180.217.225
nginx-deployment-1159050644-x6wk8   1/1       Running   0          5m        192.168.5.8    10.180.217.225
nginx-deployment-1159050644-z2kqi   1/1       Running   0          3h        192.168.5.6    10.180.217.225

使用過濾器set-based方式篩選，看下可否篩選出來

使用in的規則篩選，則篩選出app中包含nginx的pod
root@qa-control-master2-cluster-21:~/cxq/jsonxml#  kubectl get pod -o wide -l  'app in (nginx)'
NAME                                READY     STATUS    RESTARTS   AGE       IP             NODE
nginx-deployment-1159050644-kg654   1/1       Running   0          6m        192.168.5.10   10.180.217.225
nginx-deployment-1159050644-x6wk8   1/1       Running   0          6m        192.168.5.8    10.180.217.225
nginx-deployment-1159050644-z2kqi   1/1       Running   0          3h        192.168.5.6    10.180.217.225
使用notin規則，篩選出來app不包含nginx的pod（剩餘的1條）
root@qa-control-master2-cluster-21:~/cxq/jsonxml#  kubectl get pod -o wide -l  'app notin (nginx)'
NAME           READY     STATUS    RESTARTS   AGE       IP             NODE
test-5-5tcql   1/1       Running   0          1d        192.168.5.2    10.180.217.225