make example working well

doc/install.md

@@ -1,12 +1,10 @@
 # Install EDL dependencies
 
-## Kubernetes
-
 Before everything, make sure you have a running Kubernetes v1.7.* cluster and a working `kubectl`.
 You may prepare a Kubernetes on your laptop using [minikube](#minikube), create a Kubernetes Cluster on [AWS](#AWS)
 or any other way as you expediently.
 
-### kubectl
+## kubectl
 
 `kubectl` is the CLI tool to manager your Kubernetes cluster, and kubectl `v1.9.0` works well for Kubernets v1.7.*,
 you can install it by following commands:
@@ -27,6 +25,16 @@ chmod +x ./kubectl
 sudo mv ./kubectl /usr/local/bin/kubectl
 ```
 
+- Windows
+
+```bash
+wget -O kubectl.exe https://storage.googleapis.com/kubernetes-release/release/v1.9.0/bin/windows/amd64/kubectl.exe
+```
+
+And then add the binary file to your PATH.
+
+## Kubernetes Cluster
+
 ### minikube
 
 If you just trying to play EDL in your laptop, go with [minikube](https://github.com/kubernetes/minikube) with the following command is good enough to get you ready.
@@ -44,11 +52,59 @@ kubectl version
 ### AWS
 
 If you are trying to deploy EDL on a real Kubernetes cluster, [kops](https://github.com/kubernetes/kops) is
-an easy way to get a production grad Kubernetes.
+an easy way to get a production grad Kubernetes on AWS.
 
-You can checkout the [tutorial](https://github.com/kubernetes/kops/blob/master/docs/aws.md#configure-dns) for the details of lunching a Kubernetes cluster on AWS.
+You can checkout the [tutorial](https://github.com/kubernetes/kops/blob/master/docs/aws.md) for the details of launching a Kubernetes cluster on AWS.
 
-You can use the following command to create the cluster instead of the simple one in the tutorial:
+- Prepare kops environment
+
+- Configurate DNS
+
+Because of I bought a domain `yancey.co.uk` via AWS, and had a hosted zone in Route53, so I don't need to do more work about DNS configuration,
+for the other condition, please following [Configure DNS](https://github.com/kubernetes/kops/blob/master/docs/aws.md#configure-dns).
+
+Please testing your DNS setup before continue:
+
+```bash
+dig ns yancey.co.uk
+```
+
+Should return something similar to:
+
+```bash
+yancey.co.uk.		172778	IN	NS	ns-1939.awsdns-50.co.uk.
+yancey.co.uk.		172778	IN	NS	ns-1101.awsdns-09.org.
+yancey.co.uk.		172778	IN	NS	ns-320.awsdns-40.com.
+yancey.co.uk.		172778	IN	NS	ns-563.awsdns-06.net.
+```
+
+**NOTE**: DNS is a critical component for `kops`, please validated your DNS configuration before moving on.
+
+- Cluster State Storage
+
+```bash
+aws s3api create-bucket \
+    --bucket prefix-example-com-state-store \
+    --region us-east-1
+```
+
+- Prepare local environment
+
+```bash
+export NAME=myfirstcluster.example.com
+export KOPS_STATE_STORE=s3://prefix-example-com-state-store
+
+```
+
+- Create your cluster
+
+We will need to note which availability zones are available to us. In this example we will be deploying our cluster to the us-west-2 region.
+
+```bash
+aws ec2 describe-availability-zones --region us-west-2
+```
+
+Then you can use the following command to create the cluster:
 
 ```bash
 kops create cluster \
@@ -64,14 +120,54 @@ kops create cluster \
     ${NAME} --yes
 ```
 
+It would take serival minutes to create the cluster ...
+
+- Validating the cluster
+
+You can use the following command to validate the cluster:
+
+```bash
+kops validate cluster
+```
+
+When you see the output like:
+
+```text
+
+Validating cluster kops.example.yancey.co.uk
+
+INSTANCE GROUPS
+NAME			ROLE	MACHINETYPE	MIN	MAX	SUBNETS
+master-us-west-2a	Master	t2.medium	1	1	us-west-2a
+nodes			Node	t2.xlarge	3	3	us-west-2a
+
+NODE STATUS
+NAME						ROLE	READY
+ip-172-20-35-244.us-west-2.compute.internal	node	True
+ip-172-20-54-167.us-west-2.compute.internal	node	True
+ip-172-20-61-99.us-west-2.compute.internal	master	True
+ip-172-20-62-223.us-west-2.compute.internal	node	True
+
+Your cluster kops.example.yancey.co.uk is ready
+```
+
+That means your cluster is ready
+
 - Configure RBAC for your cluster
 
+The default authauthentication policy of the Kubernetes cluster created by `kops` is `RBAC`,
+to allow EDL controller changing the number of trainer process, we need to assign the controller `admin`
+permission.
+
 ```bash
 kubectl create ../k8s/admin.yaml
 ```
 
 - Delete you cluster
 
+You can execute the following command the delete the cluster configuration and resource that
+you created above.
+
 ```bash
 kops delete cluster ${NAME} --yes
 ```

example/examplejob.yaml

@@ -3,13 +3,13 @@ kind: TrainingJob
 metadata:
   name: example
 spec:
-  image: "paddlepaddle/paddlecloud-job"
+  image: "yancey1989/edl-tutorial"
   port: 7164
   ports_num: 1
   ports_num_for_sparse: 1
   fault_tolerant: true
   trainer:
-    entrypoint: "python /workspace/vgg16_v2.py"
+    entrypoint: "python /workspace/train_ft.py"
     workspace: "/workspace"
     passes: 50
     min-instance: 2

k8s/edl_controller.yaml

@@ -12,7 +12,7 @@ spec:
     spec:
       containers:
       - name: training-job-controller
-        image: paddlepaddle/edl-controller
+        image: yancey1989/edl-controller
         env:
         - name: https_proxy
           value: ""

k8s/thirdpartyresource.yaml

@@ -1,7 +1,7 @@
 apiVersion: extensions/v1beta1
 kind: ThirdPartyResource
 metadata:
-  name: trainingjobs.paddlepaddle.org
+  name: training-job.paddlepaddle.org
 description: "PaddlePaddle TrainingJob operator"
 versions:
 - name: v1