Merge pull request #4259 from spidernet-io/feature/rdma

Add RDMA ref in AI scene doc.
spidernet-io · Nov 4, 2024 · 364fb54 · 364fb54
2 parents 11535ca + 5096553
commit 364fb54
Show file tree

Hide file tree

Showing 4 changed files with 8 additions and 3 deletions.
diff --git a/docs/usage/install/ai/get-started-macvlan-zh_CN.md b/docs/usage/install/ai/get-started-macvlan-zh_CN.md
@@ -150,7 +150,9 @@
     helm install spiderpool spiderpool/spiderpool -n spiderpool --set rdma.rdmaSharedDevicePlugin.install=true
     ```
 
-    > 如果您是中国用户，可以指定参数 `--set global.imageRegistryOverride=ghcr.m.daocloud.io` 来使用国内的镜像源。
+    > - 如果您是中国用户，可以指定参数 `--set global.imageRegistryOverride=ghcr.m.daocloud.io` 来使用国内的镜像源。
+    > - 设置 `--set spiderpoolAgent.prometheus.enabled --set spiderpoolAgent.prometheus.enabledRdmaMetric=true` 和 `--set grafanaDashboard.install=true` 命令行参数可以开启 RDMA metrics exporter 和 Grafana dashboard，更多可以查看 [RDMA metrics](../../rdma-metrics.md).
+
 
    完成后，安装的组件如下
 

diff --git a/docs/usage/install/ai/get-started-macvlan.md b/docs/usage/install/ai/get-started-macvlan.md
@@ -151,7 +151,8 @@ The network planning for the cluster is as follows:
     helm install spiderpool spiderpool/spiderpool -n spiderpool --set rdma.rdmaSharedDevicePlugin.install=true
     ```
 
-    > If you are a user in China, you can specify the helm option `--set global.imageRegistryOverride=ghcr.m.daocloud.io` to use a domestic image source.
+    > - If you are a user in China, you can specify the helm option `--set global.imageRegistryOverride=ghcr.m.daocloud.io` to use a domestic image source.
+    > - Set `--set spiderpoolAgent.prometheus.enabled --set spiderpoolAgent.prometheus.enabledRdmaMetric=true` and `--set grafanaDashboard.install=true` flag to enable the RDMA metrics exporter and Grafana dashboard. Refer to [RDMA metrics](../../rdma-metrics.md).
 
     After completion, the installed components are as follows:
 

diff --git a/docs/usage/install/ai/get-started-sriov-zh_CN.md b/docs/usage/install/ai/get-started-sriov-zh_CN.md
@@ -171,7 +171,8 @@ Spiderpool 使用了 [sriov-network-operator](https://github.com/k8snetworkplumb
     helm install spiderpool spiderpool/spiderpool -n spiderpool --set sriov.install=true
     ```
 
-    > 如果您是中国用户，可以指定参数 `--set global.imageRegistryOverride=ghcr.m.daocloud.io` 来使用国内的镜像源。
+    > - 如果您是中国用户，可以指定参数 `--set global.imageRegistryOverride=ghcr.m.daocloud.io` 来使用国内的镜像源。
+    > - 设置 `--set spiderpoolAgent.prometheus.enabled --set spiderpoolAgent.prometheus.enabledRdmaMetric=true` 和 `--set grafanaDashboard.install=true` 命令行参数可以开启 RDMA metrics exporter 和 Grafana dashboard，更多可以查看 [RDMA metrics](../../rdma-metrics.md).
 
     完成后，安装的组件如下
 

diff --git a/docs/usage/install/ai/get-started-sriov.md b/docs/usage/install/ai/get-started-sriov.md
@@ -619,6 +619,7 @@ To simplify the complexity of configuring multiple network cards for AI applicat
     > - By default, the webhook automatic injection of network resources feature is disabled and needs to be manually enabled by the user.
     > - You can specify namespaces to exclude from injection using `spiderpoolController.podResourceInject.namespacesExclude`, and specify namespaces to include for injection using `spiderpoolController.podResourceInject.namespacesInclude`.
     > - After installing Spiderpool, you can update the configuration by modifying the podResourceInject field in the spiderpool-config configMap.
+    > - Set `--set spiderpoolAgent.prometheus.enabled --set spiderpoolAgent.prometheus.enabledRdmaMetric=true` and `--set grafanaDashboard.install=true` flag to enable the RDMA metrics exporter and Grafana dashboard. Refer to [RDMA metrics](../../rdma-metrics.md).
 
 2. Create SpiderMultusConfig and specify labels, and configure RDMA-related settings: