# 安裝heapster插件 ## 準備鏡像 官方鏡像保存在 gcr.io 中需要翻墻才能下載，為了方便大家使用，我下載后放到了[時速云](http://www.tenxcloud.com)中，為公有鏡像供大家下載。 - index.tenxcloud.com/jimmy/heapster-amd64:v1.3.0-beta.1 - index.tenxcloud.com/jimmy/heapster-influxdb-amd64:v1.1.1 - index.tenxcloud.com/jimmy/heapster-grafana-amd64:v4.0.2 ## 準備YAML文件 到 [heapster release 頁面](https://github.com/kubernetes/heapster/releases) 下載最新版本的 heapster。 ``` bash wget https://github.com/kubernetes/heapster/archive/v1.3.0.zip unzip v1.3.0.zip mv v1.3.0.zip heapster-1.3.0 ``` 文件目錄： `heapster-1.3.0/deploy/kube-config/influxdb` ``` bash $ cd heapster-1.3.0/deploy/kube-config/influxdb $ ls *.yaml grafana-deployment.yaml grafana-service.yaml heapster-deployment.yaml heapster-service.yaml influxdb-deployment.yaml influxdb-service.yaml heapster-rbac.yaml ``` 我們自己創建了heapster的rbac配置`heapster-rbac.yaml`。 已經修改好的 yaml 文件見：[../manifests/heapster](https://github.com/rootsongjc/kubernetes-handbook/blob/master/manifests/heapster/) ## 配置 grafana-deployment ``` bash $ diff grafana-deployment.yaml.orig grafana-deployment.yaml 16c16 < image: gcr.io/google_containers/heapster-grafana-amd64:v4.0.2 --- > image: harbor-001.jimmysong.io/library/heapster-grafana-amd64:v4.0.2 40,41c40,41 < # value: /api/v1/proxy/namespaces/kube-system/services/monitoring-grafana/ < value: / --- > value: /api/v1/proxy/namespaces/kube-system/services/monitoring-grafana/ > #value: / ``` + 如果后續使用 kube-apiserver 或者 kubectl proxy 訪問 grafana dashboard，則必須將 `GF_SERVER_ROOT_URL` 設置為 `/api/v1/proxy/namespaces/kube-system/services/monitoring-grafana/`，否則后續訪問grafana時訪問時提示找不到`http://172.20.0.113:8086/api/v1/proxy/namespaces/kube-system/services/monitoring-grafana/api/dashboards/home` 頁面； ## 配置 heapster-deployment ``` bash $ diff heapster-deployment.yaml.orig heapster-deployment.yaml 16c16 < image: gcr.io/google_containers/heapster-amd64:v1.3.0-beta.1 --- > image: harbor-001.jimmysong.io/library/heapster-amd64:v1.3.0-beta.1 ``` ## 配置 influxdb-deployment influxdb 官方建議使用命令行或 HTTP API 接口來查詢數據庫，從 v1.1.0 版本開始默認關閉 admin UI，將在后續版本中移除 admin UI 插件。 開啟鏡像中 admin UI的辦法如下：先導出鏡像中的 influxdb 配置文件，開啟 admin 插件后，再將配置文件內容寫入 ConfigMap，最后掛載到鏡像中，達到覆蓋原始配置的目的： 注意：manifests 目錄已經提供了修改后的 ConfigMap 定義文件。 ``` bash $ # 導出鏡像中的 influxdb 配置文件 $ docker run --rm --entrypoint 'cat' -ti lvanneo/heapster-influxdb-amd64:v1.1.1 /etc/config.toml >config.toml.orig $ cp config.toml.orig config.toml $ # 修改：啟用 admin 接口 $ vim config.toml $ diff config.toml.orig config.toml 35c35 < enabled = false --- > enabled = true $ # 將修改后的配置寫入到 ConfigMap 對象中 $ kubectl create configmap influxdb-config --from-file=config.toml -n kube-system configmap "influxdb-config" created $ # 將 ConfigMap 中的配置文件掛載到 Pod 中，達到覆蓋原始配置的目的 $ diff influxdb-deployment.yaml.orig influxdb-deployment.yaml 16c16 < image: gcr.io/google_containers/heapster-influxdb-amd64:v1.1.1 --- > image: harbor-001.jimmysong.io/library/heapster-influxdb-amd64:v1.1.1 19a20,21 > - mountPath: /etc/ > name: influxdb-config 22a25,27 > - name: influxdb-config > configMap: > name: influxdb-config ``` ## 配置 monitoring-influxdb Service ``` $ diff influxdb-service.yaml.orig influxdb-service.yaml 12a13 > type: NodePort 15a17,20 > name: http > - port: 8083 > targetPort: 8083 > name: admin ``` - 定義端口類型為 NodePort，額外增加了 admin 端口映射，用于后續瀏覽器訪問 influxdb 的 admin UI 界面； ## 執行所有定義文件 ``` bash $ pwd /root/heapster-1.3.0/deploy/kube-config/influxdb $ ls *.yaml grafana-service.yaml heapster-rbac.yaml influxdb-cm.yaml influxdb-service.yaml grafana-deployment.yaml heapster-deployment.yaml heapster-service.yaml influxdb-deployment.yaml $ kubectl create -f . deployment "monitoring-grafana" created service "monitoring-grafana" created deployment "heapster" created serviceaccount "heapster" created clusterrolebinding "heapster" created service "heapster" created configmap "influxdb-config" created deployment "monitoring-influxdb" created service "monitoring-influxdb" created ``` ## 檢查執行結果 檢查 Deployment ``` bash $ kubectl get deployments -n kube-system | grep -E 'heapster|monitoring' heapster 1 1 1 1 2m monitoring-grafana 1 1 1 1 2m monitoring-influxdb 1 1 1 1 2m ``` 檢查 Pods ``` bash $ kubectl get pods -n kube-system | grep -E 'heapster|monitoring' heapster-110704576-gpg8v 1/1 Running 0 2m monitoring-grafana-2861879979-9z89f 1/1 Running 0 2m monitoring-influxdb-1411048194-lzrpc 1/1 Running 0 2m ``` 檢查 kubernets dashboard 界面，看是顯示各 Nodes、Pods 的 CPU、內存、負載等利用率曲線圖；  ## 訪問 grafana 1. 通過 kube-apiserver 訪問： 獲取 monitoring-grafana 服務 URL ``` bash $ kubectl cluster-info Kubernetes master is running at https://172.20.0.113:6443 Heapster is running at https://172.20.0.113:6443/api/v1/proxy/namespaces/kube-system/services/heapster KubeDNS is running at https://172.20.0.113:6443/api/v1/proxy/namespaces/kube-system/services/kube-dns kubernetes-dashboard is running at https://172.20.0.113:6443/api/v1/proxy/namespaces/kube-system/services/kubernetes-dashboard monitoring-grafana is running at https://172.20.0.113:6443/api/v1/proxy/namespaces/kube-system/services/monitoring-grafana monitoring-influxdb is running at https://172.20.0.113:6443/api/v1/proxy/namespaces/kube-system/services/monitoring-influxdb To further debug and diagnose cluster problems, use 'kubectl cluster-info dump'. ``` 瀏覽器訪問 URL： `http://172.20.0.113:8080/api/v1/proxy/namespaces/kube-system/services/monitoring-grafana` 2. 通過 kubectl proxy 訪問： 創建代理 ``` bash $ kubectl proxy --address='172.20.0.113' --port=8086 --accept-hosts='^*$' Starting to serve on 172.20.0.113:8086 ``` 瀏覽器訪問 URL：`http://172.20.0.113:8086/api/v1/proxy/namespaces/kube-system/services/monitoring-grafana`  ## 訪問 influxdb admin UI 獲取 influxdb http 8086 映射的 NodePort ``` $ kubectl get svc -n kube-system|grep influxdb monitoring-influxdb 10.254.22.46 <nodes> 8086:32299/TCP,8083:30269/TCP 9m ``` 通過 kube-apiserver 的**非安全端口**訪問 influxdb 的 admin UI 界面： `http://172.20.0.113:8080/api/v1/proxy/namespaces/kube-system/services/monitoring-influxdb:8083/` 在頁面的 “Connection Settings” 的 Host 中輸入 node IP， Port 中輸入 8086 映射的 nodePort 如上面的 32299，點擊 “Save” 即可（我的集群中的地址是172.20.0.113:32299）：  ## 注意 在安裝好 Grafana 之后我們使用的是默認的 template 配置，頁面上的 namespace 選擇里只有 `default` 和 `kube-system`，并不是說其他的 namespace 里的指標沒有得到監控，只是我們沒有在 Grafana 中開啟他它們的顯示而已。見 [Cannot see other namespaces except, kube-system and default #1279](https://github.com/kubernetes/heapster/issues/1279)。  將 Templating 中的 namespace 的 Data source 設置為 influxdb-datasource，Refresh 設置為 on Dashboard Load 保存設置，刷新瀏覽器，即可看到其他 namespace 選項。 ## 參考 [使用Heapster獲取集群對象的metric數據](../practice/using-heapster-to-get-object-metrics.md)