Update argo instructions for test/README.md

test/README.md

@@ -5,13 +5,13 @@ This folder contains the integration/e2e tests for ML pipeline. We use Argo work
 At a high level, a typical test workflow will
 
 - build docker images for all components
-- create a dedicate test namespace in the cluster 
+- create a dedicate test namespace in the cluster
 - deploy ml pipeline using the newly built components
 - run the test
 - delete the namespace
 - delete the images
 
-All these steps will be taking place in the same Kubernetes cluster. 
+All these steps will be taking place in the same Kubernetes cluster.
 You can use GKE to test against the code in a Github Branch. The image will be temporarily stored in the GCR repository in the same project.
 
 Tests are run automatically on each commit in a Kubernetes cluster using
@@ -28,14 +28,10 @@ Here are the one-time steps to prepare for your GKE testing cluster:
 - Follow the [main page](https://github.com/kubeflow/pipelines#setup-gke) to
 create a GKE cluster.
 - Install [Argo](https://github.com/argoproj/argo/blob/master/demo.md#argo-v20-getting-started)
-in the cluster. If you have Argo CLI installed locally, just run
-  ```
-  argo install
-  ```
+in the cluster.
 - Create cluster role binding.
   ```
-  kubectl create clusterrolebinding default-as-admin
-  --clusterrole=cluster-admin --serviceaccount default:default
+  kubectl create clusterrolebinding default-as-admin --clusterrole=cluster-admin --serviceaccount=default:default
   ```
 - Follow the
 [guideline](https://developer.github.com/v3/guides/managing-deploy-keys/) to
@@ -64,7 +60,7 @@ The workflow will create a temporary namespace with the same name as the Argo
 workflow. All the images will be stored in
 **gcr.io/project_id/workflow_name/branch_name/***. By default when the test is
 *finished, the namespace and images will be deleted.
-However you can keep them by providing additional parameter. 
+However you can keep them by providing additional parameter.
 ```
 argo submit integration_test_gke.yaml -p branch="my-branch" -p cleanup="false"
 ```
@@ -75,4 +71,4 @@ argo submit integration_test_gke.yaml -p branch="my-branch" -p cleanup="false"
 
 The cluster downloads a bunch of images during the first time the test runs. It will be faster the second time since the images are cached.
 The image building steps are running in parallel and usually takes 2~3 minutes in total. If you are experiencing high latency, it might due to the resource constrains
-on your GKE cluster. In that case you need to deploy a larger cluster. 
+on your GKE cluster. In that case you need to deploy a larger cluster.