gocrane · qmhu · Nov 9, 2022 · Nov 9, 2022
diff --git a/deploy/metric-adapter/apiservice.yaml b/deploy/metric-adapter/apiservice.yaml
@@ -1,20 +1,5 @@
 apiVersion: apiregistration.k8s.io/v1
 kind: APIService
-metadata:
-  name: v1beta1.custom.metrics.k8s.io
-spec:
-  service:
-    name: metric-adapter
-    namespace: crane-system
-  group: custom.metrics.k8s.io
-  version: v1beta1
-  insecureSkipTLSVerify: true
-  groupPriorityMinimum: 100
-  versionPriority: 100
-
----
-apiVersion: apiregistration.k8s.io/v1
-kind: APIService
 metadata:
   name: v1beta1.external.metrics.k8s.io
 spec:

diff --git a/site/content/en/docs/Best Practices/effective-hpa-with-prometheus-adapter.md b/site/content/en/docs/Best Practices/effective-hpa-with-prometheus-adapter.md
@@ -36,7 +36,7 @@ Since Crane is installed, the result is as follows.
 ```sh
 NAME                                   SERVICE                           AVAILABLE   AGE
 v1beta1.batch                          Local                             True        35d
-v1beta1.custom.metrics.k8s.io          crane-system/metric-adapter       True        18d
+v1beta1.custom.metrics.k8s.io          Local                             True        18d
 v1beta1.discovery.k8s.io               Local                             True        35d
 v1beta1.events.k8s.io                  Local                             True        35d
 v1beta1.external.metrics.k8s.io        crane-system/metric-adapter       True        18d
@@ -47,7 +47,6 @@ v1beta1.metrics.k8s.io                 kube-system/metrics-service       True
 Remove the installed ApiService by crane
 
 ```bash
-kubectl delete apiservice v1beta1.custom.metrics.k8s.io
 kubectl delete apiservice v1beta1.external.metrics.k8s.io
 ```
 
@@ -98,7 +97,7 @@ spec:
 
 #### RemoteAdapter Capabilities
 
-![](/images/remote-adapter.png)
+![remote adapter](/images/remote-adapter.png)
 
 Kubernetes restricts an ApiService to configure only one backend service, so in order to use the Metric provided by Crane and the Metric provided by PrometheusAdapter within a cluster, Crane supports a RemoteAdapter to solve this problem
 
@@ -241,8 +240,8 @@ Annotation in the Effective HPA adds the configuration according to the followin
 
 ```yaml
 annotations:
-  # metric-query.autoscaling.crane.io 是固定的前缀，后面是 Metric 名字，需跟 spec.metrics 中的 Metric.name 相同，支持 Pods 类型和 External 类型
-  metric-query.autoscaling.crane.io/http_requests: "sum(rate(http_requests_total[5m])) by (pod)"
+  # metric-query.autoscaling.crane.io is a static prefix，after that should be the Metric Type and Metric Name，It should be the same as your spec.metrics's Metric.name，Supported Metric type are Pods(pods) and External (external)
+  metric-query.autoscaling.crane.io/pods.http_requests: "sum(rate(http_requests_total[5m])) by (pod)"
 ```
 
 <summary>sample-app-hpa.yaml</summary>
@@ -253,8 +252,8 @@ kind: EffectiveHorizontalPodAutoscaler
 metadata:
   name: php-apache
   annotations:
-    # metric-query.autoscaling.crane.io 是固定的前缀，后面是 Metric 名字，需跟 spec.metrics 中的 Metric.name 相同，支持 Pods 类型和 External 类型
-    metric-query.autoscaling.crane.io/http_requests: "sum(rate(http_requests_total[5m])) by (pod)"
+    # metric-query.autoscaling.crane.io 是固定的前缀，后面是 Metric 的 type 和 名字，需跟 spec.metrics 中的 Metric.name 相同，支持 Pods 类型(pods)和 External 类型(external)
+    metric-query.autoscaling.crane.io/pods.http_requests: "sum(rate(http_requests_total[5m])) by (pod)"
 spec:
   # ScaleTargetRef is the reference to the workload that should be scaled.
   scaleTargetRef:
@@ -328,7 +327,7 @@ spec:
           sampleInterval: 60s
       expressionQuery:
         expression: sum(rate(http_requests_total[5m])) by (pod)
-      resourceIdentifier: crane_custom.pods_http_requests
+      resourceIdentifier: pods.http_requests
       type: ExpressionQuery
   predictionWindowSeconds: 3600
   targetRef:
@@ -357,7 +356,7 @@ status:
               value: "0.01683"
             ......
       ready: true
-      resourceIdentifier: crane_custom.pods_http_requests  
+      resourceIdentifier: pods.http_requests  
 ```
 
 Looking at the HPA object created by Effective HPA, you can observe that a Metric has been created based on custom metrics predictions: ``crane_custom.pods_http_requests``.
@@ -386,7 +385,7 @@ spec:
     type: Pods
   - pods:
       metric:
-        name: crane_custom.pods_http_requests
+        name: pods.http_requests
         selector:
           matchLabels:
             autoscaling.crane.io/effective-hpa-uid: 1322c5ac-a1c6-4c71-98d6-e85d07b22da0

diff --git a/site/content/en/docs/Getting started/installation.md b/site/content/en/docs/Getting started/installation.md
@@ -85,6 +85,55 @@ helm install fadvisor -n crane-system --create-namespace crane/fadvisor
 {{< /tab >}}
 {{% /tabpane %}}
 
+### Using Existing Prometheus（Optional）
+
+Normally in a production environment offen already have a specified Prometheus. You can modify the Crane Chart Release configuration by using the following command or modify the Craned Deployment container Args directly.
+
+```bash
+helm upgrade crane -n crane-system --set craned.containerArgs.prometheus-address=http://{prometheus-ip}:{port} --create-namespace crane/crane
+```
+
+At the same time, the cost analytics of the Crane Dashboard need to deploy [kube-state-metrics](https://github.com/kubernetes/kube-state-metrics) (Prometheus Chart Will install it by default), also you need to config additional extraScrapeConfigs in your Prometheus, You can refer to [here](https://github.com/gocrane/helm-charts/blob/main/integration/prometheus/override_values.yaml#L56).
+
+Finally, fadvisor needs to configure recording rules to aggregate cost data, You can refer to [here](https://github.com/gocrane/helm-charts/blob/main/integration/prometheus/override_values.yaml#L6) to configuration in your Prometheus.
+
+### Using Existing Grafana（Optional）
+
+The Crane Dashboard supports the Grafana report embedded with the Iframe to show the cost distribution. If you want to use an external Grafana to embed into the Crane Dashboard, you need to modify the nginx configuration in the configmap at first.
+
+```bash
+kubectl edit configmap -n monitor grafana
+```
+
+Change `grafana.{{ .Release.Namespace }}.svc.cluster.local` to be existing Grafana server address，Change `http://$upstream_grafana:8082` to be the existing Grafana server port。
+
+```yaml
+ location /grafana {
+    set $upstream_grafana grafana.{{ .Release.Namespace }}.svc.cluster.local;
+    proxy_connect_timeout       180;
+    proxy_send_timeout          180;
+    proxy_read_timeout          180;
+    proxy_pass http://$upstream_grafana:8082;
+    proxy_redirect off;
+    rewrite /grafana/(.*) /$1 break;
+    proxy_http_version 1.1;
+    proxy_set_header  Host $http_host;
+    proxy_set_header  Upgrade $http_upgrade;
+    proxy_set_header  Connection $connection_upgrade;
+    proxy_set_header  X-Real-IP  $remote_addr;
+    proxy_set_header  X-Forwarded-For $proxy_add_x_forwarded_for;
+}
+
+```
+
+Next up you need to config your grafana based on [here](https://github.com/gocrane/helm-charts/blob/main/integration/grafana/override_values.yaml), The idea was that Grafana supported embedding panels, but required that the corresponding permission configuration be turned on.
+
+- Make sure Service configuration is the same as nginx 
+- Config datasources
+- Config dashboardProviders
+- Config dashboards
+- Config grafana.ini
+
 ### Deploying Crane-scheduler(optional)
 ```bash
 helm install scheduler -n crane-system --create-namespace crane/scheduler
@@ -176,7 +225,6 @@ echo "Dashboard link: http://${NODE_IP}:${PORT}"
 
 ### LoadBalancer
 
-
 #### Quick Start
 
 ```bash

diff --git a/site/content/zh/docs/Best Practices/effective-hpa-with-prometheus-adapter.md b/site/content/zh/docs/Best Practices/effective-hpa-with-prometheus-adapter.md
@@ -36,7 +36,7 @@ kubectl get apiservice
 ```bash
 NAME                                   SERVICE                           AVAILABLE   AGE
 v1beta1.batch                          Local                             True        35d
-v1beta1.custom.metrics.k8s.io          crane-system/metric-adapter       True        18d
+v1beta1.custom.metrics.k8s.io          Local                             True        18d
 v1beta1.discovery.k8s.io               Local                             True        35d
 v1beta1.events.k8s.io                  Local                             True        35d
 v1beta1.external.metrics.k8s.io        crane-system/metric-adapter       True        18d
@@ -47,7 +47,6 @@ v1beta1.metrics.k8s.io                 kube-system/metrics-service       True
 删除 crane 安装的 ApiService
 
 ```bash
-kubectl delete apiservice v1beta1.custom.metrics.k8s.io
 kubectl delete apiservice v1beta1.external.metrics.k8s.io
 ```
 
@@ -98,7 +97,7 @@ spec:
 
 #### RemoteAdapter 能力
 
-![](../images/remote-adapter.png)
+![remote adapter](/images/remote-adapter.png)
 
 Kubernetes 限制一个 ApiService 只能配置一个后端服务，因此，为了在一个集群内使用 Crane 提供的 Metric 和 PrometheusAdapter 提供的 Metric，Crane 支持了 RemoteAdapter 解决此问题
 
@@ -241,8 +240,8 @@ kubectl get --raw /apis/custom.metrics.k8s.io/v1beta1 | jq .
 
 ```yaml
 annotations:
-  # metric-query.autoscaling.crane.io 是固定的前缀，后面是 Metric 名字，需跟 spec.metrics 中的 Metric.name 相同，支持 Pods 类型和 External 类型
-  metric-query.autoscaling.crane.io/http_requests: "sum(rate(http_requests_total[5m])) by (pod)"
+  # metric-query.autoscaling.crane.io 是固定的前缀，后面是 Metric 的 type 和 名字，需跟 spec.metrics 中的 Metric.name 相同，支持 Pods 类型(pods)和 External 类型(external)
+  metric-query.autoscaling.crane.io/pods.http_requests: "sum(rate(http_requests_total[5m])) by (pod)"
 ```
 
 <summary>sample-app-hpa.yaml</summary>
@@ -253,8 +252,8 @@ kind: EffectiveHorizontalPodAutoscaler
 metadata:
   name: php-apache
   annotations:
-    # metric-query.autoscaling.crane.io 是固定的前缀，后面是 Metric 名字，需跟 spec.metrics 中的 Metric.name 相同，支持 Pods 类型和 External 类型
-    metric-query.autoscaling.crane.io/http_requests: "sum(rate(http_requests_total[5m])) by (pod)"
+    # metric-query.autoscaling.crane.io 是固定的前缀，后面是 Metric 的 type 和 名字，需跟 spec.metrics 中的 Metric.name 相同，支持 Pods 类型(pods)和 External 类型(external)
+    metric-query.autoscaling.crane.io/pods.http_requests: "sum(rate(http_requests_total[5m])) by (pod)"
 spec:
   # ScaleTargetRef is the reference to the workload that should be scaled.
   scaleTargetRef:
@@ -328,7 +327,7 @@ spec:
           sampleInterval: 60s
       expressionQuery:
         expression: sum(rate(http_requests_total[5m])) by (pod)
-      resourceIdentifier: crane_custom.pods_http_requests
+      resourceIdentifier: pods.http_requests
       type: ExpressionQuery
   predictionWindowSeconds: 3600
   targetRef:
@@ -357,7 +356,7 @@ status:
               value: "0.01683"
             ......
       ready: true
-      resourceIdentifier: crane_custom.pods_http_requests  
+      resourceIdentifier: pods.http_requests  
 ```
 
 查看 Effective HPA 创建的 HPA 对象，可以观测到已经创建出基于自定义指标预测的 Metric: `crane_custom.pods_http_requests`
@@ -386,7 +385,7 @@ spec:
     type: Pods
   - pods:
       metric:
-        name: crane_custom.pods_http_requests
+        name: pods.http_requests
         selector:
           matchLabels:
             autoscaling.crane.io/effective-hpa-uid: 1322c5ac-a1c6-4c71-98d6-e85d07b22da0

diff --git a/site/content/zh/docs/Getting started/installation.md b/site/content/zh/docs/Getting started/installation.md
@@ -95,6 +95,55 @@ helm install fadvisor -n crane-system --create-namespace crane/fadvisor
 {{< /tab >}}
 {{% /tabpane %}}
 
+### 使用外部的 Prometheus（可选）
+
+通常在生产环境，安装时需要配置外部的 Prometheus，你可以通过以下命令修改 Crane 的 Chart Release 配置或者直接修改 Craned Deployment 的容器 Args。
+
+```bash
+helm upgrade crane -n crane-system --set craned.containerArgs.prometheus-address=http://{prometheus-ip}:{port} --create-namespace crane/crane
+```
+
+同时，Crane Dashboard 的成本展示需要部署[kube-state-metrics](https://github.com/kubernetes/kube-state-metrics)（Prometheus Chart 中默认会安装），并且需要在你的 Prometheus 中配置额外的 extraScrapeConfigs，可以参考[这里](https://github.com/gocrane/helm-charts/blob/main/integration/prometheus/override_values.yaml#L56)。
+
+最后，Fadvisor 需要配置 recording rules 来实现成本数据的聚合，可以参考[这里](https://github.com/gocrane/helm-charts/blob/main/integration/prometheus/override_values.yaml#L6)配置到你的 Prometheus 中。
+
+### 使用外部的 Grafana（可选）
+
+Crane Dashboard 支持通过 Iframe 内嵌 Grafana 报表展示成本分布。如果希望使用外部的 Grafana 内嵌到 Crane Dashboard，首先需要修改 configmap 中的 nginx 配置。
+
+```bash
+kubectl edit configmap -n monitor grafana
+```
+
+配置 `grafana.{{ .Release.Namespace }}.svc.cluster.local` 成外部的 Grafana 服务地址，配置 `http://$upstream_grafana:8082` 成外部的 Grafana 服务端口。
+
+```yaml
+ location /grafana {
+    set $upstream_grafana grafana.{{ .Release.Namespace }}.svc.cluster.local;
+    proxy_connect_timeout       180;
+    proxy_send_timeout          180;
+    proxy_read_timeout          180;
+    proxy_pass http://$upstream_grafana:8082;
+    proxy_redirect off;
+    rewrite /grafana/(.*) /$1 break;
+    proxy_http_version 1.1;
+    proxy_set_header  Host $http_host;
+    proxy_set_header  Upgrade $http_upgrade;
+    proxy_set_header  Connection $connection_upgrade;
+    proxy_set_header  X-Real-IP  $remote_addr;
+    proxy_set_header  X-Forwarded-For $proxy_add_x_forwarded_for;
+}
+
+```
+
+接下来需要参考[这里](https://github.com/gocrane/helm-charts/blob/main/integration/grafana/override_values.yaml)进行配置，原理是 Grafana 支持前端图表的内嵌，但是需要把对应的权限配置打开。
+
+- 确定 Service 和 nginx 配置一致
+- 配置 datasources
+- 配置 dashboardProviders
+- 配置 dashboards
+- 配置 grafana.ini
+
 ### 安装 Crane-scheduler（可选）
 ```console
 helm install scheduler -n crane-system --create-namespace crane/scheduler