diff --git a/manifests/gcp_marketplace/guide.md b/manifests/gcp_marketplace/guide.md
index 10d40960ac3..bd8fb372570 100644
--- a/manifests/gcp_marketplace/guide.md
+++ b/manifests/gcp_marketplace/guide.md
@@ -43,6 +43,7 @@ gcloud projects add-iam-policy-binding $PROJECT_ID \
   --member=serviceAccount:$SA_NAME@$PROJECT_ID.iam.gserviceaccount.com \
   --role=roles/storage.admin
 
+# Note that you can not bind multiple roles in one line.
 gcloud projects add-iam-policy-binding $PROJECT_ID \
   --member=serviceAccount:$SA_NAME@$PROJECT_ID.iam.gserviceaccount.com \
   --role=roles/ml.admin
@@ -98,3 +99,25 @@ as `Service Account User`. The Google Service Account is [Compute Engine default
 - Please also add your account as `Project Viewer` via [IAM](https://console.cloud.google.com/iam-admin/iam).
 
 For simplicity but not good for security, adding as `Project Editor` also can work.
+
+### Pipeline steps got insufficient permission
+If you see an error message stating that the pipeline got insufficient
+permissions, for example:
+
+```
+ Error executing an HTTP request: HTTP response code 403 with body '{
+ "error": {
+  "errors": [
+   {
+    "domain": "global",
+    "reason": "insufficientPermissions",
+    "message": "Insufficient Permission"
+   }
+  ],
+  "code": 403,
+  "message": "Insufficient Permission"
+ }
+}
+```
+please make sure following the procedure in [credential setup](#gcp-service-account-credentials). IAM configuration and/or
+ API enabling might take up to 5 mins to propagate. 
diff --git a/samples/contrib/parameterized_tfx_oss/README.md b/samples/contrib/parameterized_tfx_oss/README.md
index f53718382d6..a29d3e563ee 100644
--- a/samples/contrib/parameterized_tfx_oss/README.md
+++ b/samples/contrib/parameterized_tfx_oss/README.md
@@ -1,4 +1,4 @@
-# Parameterized TFX pipeline sample
+# Overview
 
 [Tensorflow Extended (TFX)](https://github.com/tensorflow/tfx) is a Google-production-scale machine
 learning platform based on TensorFlow. It provides a configuration framework to express ML pipelines
@@ -9,15 +9,19 @@ This sample demonstrates how to author a ML pipeline in TFX and run it on a KFP
 Please refer to inline comments for the purpose of each step.
 
 In order to successfully compile this sample, you'll need to have a TFX installation at HEAD.
-First, you can clone their repo and run `python setup.py install` from `tfx/tfx`. 
+First, you can clone their repo and run `python setup.py install` from `tfx/`. 
 The image used in the pipeline is specified as `tfx_image` in the 
 `KubeflowDagRunnerConfig`. Currently we're using our own patched version of TFX image containing visualization support.
 List of officially released nightly build image available can be found [here](https://hub.docker.com/r/tensorflow/tfx/tags)).
 
 After that, running 
 `python3 chicago_taxi_pipeline_simple.py` compiles the TFX pipeline into KFP pipeline package.
-This pipeline requires google storage permission to run. 
 
+# Permission
+
+This pipeline requires Google Cloud Storage permission to run. 
+If KFP was deployed through K8S marketplace, please follow instructions in [the guideline](https://github.com/kubeflow/pipelines/blob/master/manifests/gcp_marketplace/guide.md#gcp-service-account-credentials)
+to make sure the service account has `storage.admin` role.
 
 ## Caveats
 
@@ -28,8 +32,7 @@ objects `dsl.PipelineParam` and appending them to the `KubeflowDagRunner._params
 KubeflowDagRunner can correctly identify those pipeline parameters and interpret them as Argo
 placeholder correctly when compilation. However, this parameterization approach is a hack and 
 we do not have plan for long-term support. Instead we're working with TFX team to support 
-pipeline parameterization using their [RuntimeParameter](https://github.com/tensorflow/tfx/blob/46bb4f975c36ea1defde4b3c33553e088b3dc5b8/tfx/orchestration/data_types.py#L108).
-
+pipeline parameterization using their [RuntimeParameter](https://github.com/tensorflow/tfx/blob/46bb4f975c36ea1defde4b3c33553e088b3dc5b8/tfx/orchestration/data_types.py#L108). 
 ### Known issues
 * This approach only works for string-typed quantities. For example, you cannot parameterize 
 `num_steps` of `Trainer` in this way.
diff --git a/samples/core/xgboost_training_cm/README.md b/samples/core/xgboost_training_cm/README.md
index 3469fac2afc..7cb6e3dbf34 100644
--- a/samples/core/xgboost_training_cm/README.md
+++ b/samples/core/xgboost_training_cm/README.md
@@ -12,8 +12,11 @@ or not.
 
 ## Requirements
 
-Preprocessing uses Google Cloud DataProc. Therefore, you must enable the [DataProc API](https://cloud.google.com/endpoints/docs/openapi/enable-api) for the given GCP project.
-
+Preprocessing uses Google Cloud DataProc. Therefore, you must enable the 
+[Cloud Dataproc API](https://pantheon.corp.google.com/apis/library/dataproc.googleapis.com?q=dataproc) for the given GCP project. This is the 
+general [guideline](https://cloud.google.com/endpoints/docs/openapi/enable-api) for enabling GCP APIs.
+If KFP was deployed through K8S marketplace, please follow instructions in [the guideline](https://github.com/kubeflow/pipelines/blob/master/manifests/gcp_marketplace/guide.md#gcp-service-account-credentials)
+to make sure the service account used has the role `storage.admin` and `dataproc.admin`.
 
 ## Compile