k8s應(yīng)用分享課件

kubernetes

生產(chǎn)級(jí)容器應(yīng)用集群平臺(tái)k8s應(yīng)用分享架構(gòu)k8s應(yīng)用分享組件、概念masterkube-apiserverkube-controller-managerkube-schedulernodekubeletkube-proxydeploymentdeploymentk8s應(yīng)用分享k8s重要流程說(shuō)明負(fù)載均衡服務(wù)發(fā)現(xiàn)更新回滾日志收集擴(kuò)容縮容監(jiān)控預(yù)警k8s應(yīng)用分享service概念將一組提供相同服務(wù)的pod定義為一個(gè)服務(wù)提供統(tǒng)一的訪問(wèn)地址入口服務(wù)端口暴露服務(wù)發(fā)現(xiàn)負(fù)載機(jī)制k8s應(yīng)用分享node1node2pod1label：app=opspod2label：app=dbpod3label：app=opspod4label：app=dbname=svc-opslabel：app=opsclusterip：192.168.6.6:5000svc2name=svc-dblabel：app=dbclusterip：192.168.6.7:5000pod1pod3pod2pod4svc1svc2k8s應(yīng)用分享node1ip：192.168.6.128port：5000node2ip：192.168.6.208port：5000pod1ip：172.16.1.1port：5000label：app=opspod2ip：172.16.1.2port：5000label：app=ops映射：192.168.6.128:5000172.16.1.1:500映射：192.168.6.208:5000172.16.1.2:500用戶serviceclusterip：192.168.6.111port：5000kube-proxy轉(zhuǎn)發(fā)映射映射請(qǐng)求請(qǐng)求proxypod1proxypod2service端口暴露k8s應(yīng)用分享服務(wù)注冊(cè)發(fā)現(xiàn)k8smastersvc信息：name：dbvalue：192.168.6.7:5000kube2skyetcd獲取寫入skydns查詢pod1label：name=ops數(shù)據(jù)庫(kù)地址：dbpod2label：name=db查詢db對(duì)應(yīng)的ip地址192.168.6.7:5000k8s應(yīng)用分享服務(wù)的負(fù)載均衡機(jī)制輪詢session會(huì)話保持k8s應(yīng)用分享應(yīng)用日志收集掛載到物理主機(jī)保持日志持久化定義容器組pod收集k8s應(yīng)用分享應(yīng)用日志收集

node1pod1label：name=ops日志路徑：/var/log/ops/graylogagent/var/log/ops/掛載采集pod1容器1label：name=ops/var/log/ops容器2label：name=graylog_agent/var/log/ops目錄共享k8s應(yīng)用分享pod更新回滾rollingupdate滾動(dòng)更新recreate重建k8s應(yīng)用分享rollingupdate滾動(dòng)更新serviceclusteripold版本pod1label：name=opsold版本pod2label：name=opsold版本pod3label：name=opsnew版本pod4label：name=opsnew版本pod4label：name=opsnew版本pod5label：name=opsnew版本pod5label：name=opsnew版本pod6label：name=opsnew版本pod6label：name=opsk8s應(yīng)用分享recreate重建serviceclusteripold版本pod1label：name=opsold版本pod2label：name=opsold版本pod3label：name=opsnew版本pod5label：name=opsnew版本pod6label：name=opsnew版本pod4label：name=opsk8s應(yīng)用分享pod擴(kuò)容縮容手動(dòng)擴(kuò)容縮容

命令：kubectlscaledeploymentxxxreplicas=1自動(dòng)擴(kuò)容縮容

根據(jù)cpu使用率k8s應(yīng)用分享自動(dòng)擴(kuò)容

HPApod1Heapster獲取cpuresources:requests:cpu:200mminReplicas:1maxReplicas:10targetCPUUtilizationPercentage:50scalerc/deploymentHPApod1Heapsterscalerc/deploymentpod2k8s應(yīng)用分享資源監(jiān)控cAdvisor（單節(jié)點(diǎn)收集）kubernetesdashboard（集群展示）Heapster+InfluxDB+Grafana（持久化展示）k8s應(yīng)用分享cAdvisor單節(jié)點(diǎn)監(jiān)控node1kubelet監(jiān)聽：4149cAdvisorpod1pod2pod3k8s應(yīng)用分享kubernetesdashboardnode1node2kubeletkubeletpod1label：name：heapsterpod2label：name：opspod3label：name：dashboardpod4label：name：opsmasterapi-server采集采集cAdvisorcAdvisornodeportservicek8s應(yīng)用分享Heapster+InfluxDB+Grafanamasterapi_serverheapster獲取節(jié)點(diǎn)信息node1kubeletcAdvisorpod1pod2node2kubeletcAdvisorpod3pod4influxdbGrafana采集采集寫入讀取k8s應(yīng)用分享持續(xù)部署（CD）開發(fā)者gitlabJenkinsdocker打包docker倉(cāng)庫(kù)k8s集群pushpostcallpushcall拉取鏡像部署k8s應(yīng)用分享