update doc for join command (#295)

* update doc for join command Signed-off-by: ycyaoxdu <[email protected]> * Update README.md Co-authored-by: Jian Qiu <[email protected]> Signed-off-by: ycyaoxdu <[email protected]> --------- Signed-off-by: ycyaoxdu <[email protected]> Co-authored-by: Jian Qiu <[email protected]>
open-cluster-management-io · Feb 28, 2023 · e4bfc3f · e4bfc3f
1 parent 2e33a5a
commit e4bfc3f
Show file tree

Hide file tree

Showing 3 changed files with 13 additions and 6 deletions.
diff --git a/README.md b/README.md
@@ -85,21 +85,25 @@ Initialize the hub by deploying the hub side resources to manage clusters.
 `clusteradm init [--use-bootstrap-token]`
 
 it returns the command line to launch on the spoke to join the hub.
+> NOTE: Do not run init command against a [multicluster-controlplane](https://github.com/open-cluster-management-io/multicluster-controlplane) instance. It is already an initialized hub on start. Instead, use `clusteradm get token --use-bootstrap-token` to get the join command.
 
 ### get token
 
 Get the latest token to import a new managed cluster.
 
 `clusteradm get token --context ${CTX_HUB_CLUSTER}`
-
 ### join
 
 Install the agent on the spoke.
 
-`clusteradm join --hub-token <token> --hub-apiserver <hub_apiserver_url> --cluster-name c1`
+`clusteradm join --hub-token <token> --hub-apiserver <hub_apiserver_url> --cluster-name c1 [--ca-file <path-to-ca-file>] [--force-internal-endpoint-lookup]`
 
 it returns the command line to launch on the hub the accept the spoke onboarding.
 
+> NOTE: The `--ca-file` flag is used to provide a valid CA for hub. The ca data is fetched from cluster-info configmap in kube-public namespace of the hub cluster, then from kube-root-ca.crt configmap in kube-public namespace if the cluster-info configmap does not exist.
+
+> NOTE: If you're trying to join a hub cluster which is initialized from a kind cluster, please set the `--force-internal-endpoint-lookup` flag.
+
 ### accept
 
 Accept the CSRs on the hub to approve the spoke clusters to join the hub.
@@ -111,6 +115,7 @@ Accept the CSRs on the hub to approve the spoke clusters to join the hub.
 Uninstall the agent on the spoke
 
 `clusteradm unjoin --cluster-name c1`
+> Note: the applied resources on managed cluster would be checked and prompt a warning if still exist any.
 
 ### clean
 

diff --git a/pkg/cmd/join/cmd.go b/pkg/cmd/join/cmd.go
@@ -11,10 +11,12 @@ import (
 )
 
 var example = `
-# Join a cluster to the hub
+# Join a cluster to the hub 
 %[1]s join --hub-token <tokenID.tokenSecret> --hub-apiserver <hub_apiserver_url> --cluster-name <cluster_name>
-# Join a cluster to the hub with hosted mode
-%[1]s join --hub-token <tokenID.tokenSecret> --hub-apiserver <hub_apiserver_url> --cluster-name <cluster_name> --mode hosted --managed-cluster-kubeconfig <path_to_kubeconfig>
+# join a cluster to the hub with hosted mode
+%[1]s join --hub-token <tokenID.tokenSecret> --hub-apiserver <hub_apiserver_url> --cluster-name <cluster_name> --mode hosted --managed-cluster-kubeconfig <managed-cluster-kubeconfig-file>
+# join a cluster to the hub while the hub provided no valid CA data in kube-public namespace
+%[1]s join --hub-token <tokenID.tokenSecret> --hub-apiserver <hub_apiserver_url> --cluster-name <cluster_name> --ca-file <ca-file>
 `
 
 // NewCmd ...

diff --git a/pkg/cmd/join/exec.go b/pkg/cmd/join/exec.go
@@ -128,7 +128,7 @@ func (o *Options) complete(cmd *cobra.Command, args []string) (err error) {
 
 	// code logic of building hub client in join process:
 	// 1. use the token and insecure to fetch the ca data from cm in kube-public ns
-	// 2. if not found, assume using a authorized ca.
+	// 2. if not found, assume using an authorized ca.
 	// 3. use the ca and token to build a secured client and call hub
 
 	//Create an unsecure bootstrap