add moved files

doc/howto/usage/cluster/fabric_cn.md

+# 使用fabric启动集群训练
+
+## 准备一个Linux集群
+可以在`paddle/scripts/cluster_train_v2/fabric/docker_cluster`目录下，执行`kubectl -f ssh_servers.yaml`启动一个测试集群，并使用`kubectl get po -o wide`获得这些节点的IP地址。
+
+## 启动集群作业
+
+`paddle.py` 提供了自动化脚本来启动不同节点中的所有 PaddlePaddle 集群进程。默认情况下，所有命令行选项可以设置为 `paddle.py` 命令选项并且 `paddle.py` 将透明、自动地将这些选项应用到 PaddlePaddle 底层进程。
+
+`paddle.py` 为方便作业启动提供了两个独特的命令选项。
+
+-  `job_dispatch_package`  设为本地 `workspace` 目录，它将被分发到 `conf.py` 中设置的所有节点。它有助于帮助频繁修改和访问工作区文件的用户减少负担，否则频繁的多节点工作空间部署可能会很麻烦。
+-  `job_workspace`  设为已部署的工作空间目录，`paddle.py` 将跳过分发阶段直接启动所有节点的集群作业。它可以帮助减少分发延迟。
+
+`cluster_train/run.sh` 提供了命令样例来运行 `doc/howto/usage/cluster/src/word2vec` 集群任务，只需用您定义的目录修改 `job_dispatch_package` 和 `job_workspace`，然后：
+```
+sh run.sh
+```
+
+集群作业将会在几秒后启动。
+
+## 终止集群作业
+`paddle.py`能获取`Ctrl + C` SIGINT 信号来自动终止它启动的所有进程。只需中断 `paddle.py` 任务来终止集群作业。如果程序崩溃你也可以手动终止。
+
+## 检查集群训练结果
+详细信息请检查 $workspace/log 里的日志，每一个节点都有相同的日志结构。
+
+`paddle_trainer.INFO`
+提供几乎所有训练的内部输出日志，与本地训练相同。这里检验运行时间模型的收敛。
+
+`paddle_pserver2.INFO`
+提供 pserver 运行日志，有助于诊断分布式错误。
+
+`server.log`
+提供 parameter server 进程的 stderr 和 stdout。训练失败时可以检查错误日志。
+
+`train.log`
+提供训练过程的 stderr 和 stdout。训练失败时可以检查错误日志。
+
+## 检查模型输出
+运行完成后，模型文件将被写入节点 0 的 `output` 目录中。
+工作空间中的 `nodefile` 表示当前集群作业的节点 ID。

doc/howto/usage/cluster/fabric_en.md

+# Cluster Training Using Fabric
+
+## Prepare a Linux cluster
+
+Run `kubectl -f ssh_servers.yaml` under the directory:  `paddle/scripts/cluster_train_v2/fabric/docker_cluster` will launch a demo cluster. Run `kubectl get po -o wide` to get IP addresses of these nodes.
+
+## Launching Cluster Job
+`paddle.py` provides automatical scripts to start all PaddlePaddle cluster processes in different nodes. By default, all command line options can be set as `paddle.py` command options and `paddle.py` will transparently and automatically set these options to PaddlePaddle lower level processes.
+
+`paddle.py`provides two distinguished command option for easy job launching.
+
+- `job_dispatch_package` set it with local `workspace` directory, it will be dispatched to all nodes which is set in `conf.py`. It could be helpful for frequently manipulating workspace files. otherwise, frequent multi-nodes workspace deployment is very annoying.
+- `job_workspace`  set it with already deployed workspace directory, `paddle.py` will skip dispatch stage to directly launch cluster job with all nodes. It could help to reduce heavy
+dispatch latency.
+
+`cluster_train/run.sh` provides command line sample to run `demo/recommendation` cluster job, just modify `job_dispatch_package` and `job_workspace` with your defined directory, then:
+```
+sh run.sh
+```
+
+The cluster Job will start in several seconds.
+
+## Kill Cluster Job
+`paddle.py` can capture `Ctrl + C` SIGINT signal to automatically kill all processes launched by it. So just stop `paddle.py` to kill cluster job. You should manually kill the job if the program crashed.
+
+## Check Cluster Training Result
+Check log in $workspace/log for details, each node owns same log structure.
+
+`paddle_trainer.INFO`
+It provides almost all internal output log for training,  same as local training. Check runtime model convergence here.
+
+`paddle_pserver2.INFO`
+It provides parameter server running log, which could help to diagnose distributed error.
+
+`server.log`
+It provides stderr and stdout of parameter server process. Check error log if training crashes.
+
+`train.log`
+It provides stderr and stdout of trainer process. Check error log if training crashes.
+
+## Check Model Output
+After one pass finished, model files will be written in `output` directory in node 0.
+`nodefile` in workspace indicates the node id of current cluster job.

doc/howto/usage/cluster/k8s_aws_en.md

+
+# Distributed PaddlePaddle Training on AWS with Kubernetes
+
+We will show you step by step on how to run distributed PaddlePaddle training on AWS cluster with Kubernetes. Let's start from core concepts.
+
+## Distributed PaddlePaddle Training Core Concepts
+
+### Distributed Training Job
+
+A distributed training job is represented by a [Kubernetes job](https://kubernetes.io/docs/user-guide/jobs/#what-is-a-job).
+
+Each Kuberentes job is described by a job config file, which specifies the information like the number of [pods](https://kubernetes.io/docs/user-guide/pods/#what-is-a-pod) in the job and environment variables.
+
+In a distributed training job, we would:
+
+1. prepare partitioned training data and configuration file on a distributed file system (in this tutorial we use Amazon Elastic File System), and
+1. create and submit the Kubernetes job config to the Kubernetes cluster to start the training job.
+
+### Parameter Servers and Trainers
+
+There are two roles in a PaddlePaddle cluster: *parameter server (pserver)* and *trainer*. Each parameter server process maintains a shard of the global model. Each trainer has its local copy of the model, and uses its local data to update the model. During the training process, trainers send model updates to parameter servers, parameter servers are responsible for aggregating these updates, so that trainers can synchronize their local copy with the global model.
+
+<center>![Model is partitioned into two shards. Managed by two parameter servers respectively.](src/pserver_and_trainer.png)</center>
+
+In order to communicate with pserver, trainer needs to know the ip address of each pserver. In kubernetes it's better to use a service discovery mechanism (e.g., DNS hostname) rather than static ip address, since any pserver's pod may be killed and a new pod could be schduled onto another node of different ip address. However, now we are using static ip. This will be improved.
+
+Parameter server and trainer are packaged into a same docker image. They will run once pod is scheduled by kubernetes job.
+
+### Trainer ID
+
+Each trainer process requires a trainer ID, a zero-based index value, passed in as a command-line parameter. The trainer process thus reads the data partition indexed by this ID.
+
+### Training
+
+The entry-point of a container is a shell script. It can see some environment variables pre-defined by Kubernetes. This includes one that gives the job's identity, which can be used in a remote call to the Kubernetes apiserver that lists all pods in the job.
+
+We rank each pod by sorting them by their ips. The rank of each pod could be the "pod ID". Because we run one trainer and one parameter server in each pod, we can use this "pod ID" as the trainer ID. A detailed workflow of the entry-point script is as follows:
+
+1. Query the api server to get pod information, and assign the `trainer_id` by sorting the ip.
+1. Copy the training data from EFS persistent volume into container.
+1. Parse the `paddle pserver` and `paddle trainer` startup parameters from environment variables, and then start up the processes.
+1. Trainer with `train_id` 0 will automatically write results onto EFS volume.
+
+
+## PaddlePaddle on AWS with Kubernetes
+
+### Choose AWS Service Region
+This tutorial requires several AWS services work in the same region. Before we create anything in AWS, please check the following link
+https://aws.amazon.com/about-aws/global-infrastructure/regional-product-services/
+Choose a region which has the following services available: EC2, EFS, VPS, CloudFormation, KMS, VPC, S3.
+In this tutorial, we use "Oregon(us-west-2)" as example.
+
+### Create AWS Account and IAM Account
+
+Under each AWS account, we can create multiple [IAM](http://docs.aws.amazon.com/IAM/latest/UserGuide/introduction.html) users. This allows us to grant some privileges to each IAM user and to create/operate AWS clusters as an IAM user.
+
+To sign up an AWS account, please
+follow
+[this guide](http://docs.aws.amazon.com/lambda/latest/dg/setting-up.html).
+To create IAM users and user groups under an AWS account, please
+follow
+[this guide](http://docs.aws.amazon.com/IAM/latest/UserGuide/id_users_create.html).
+
+Please be aware that this tutorial needs the following privileges for the user in IAM:
+
+- AmazonEC2FullAccess
+- AmazonS3FullAccess
+- AmazonRoute53FullAccess
+- AmazonRoute53DomainsFullAccess
+- AmazonElasticFileSystemFullAccess
+- AmazonVPCFullAccess
+- IAMUserSSHKeys
+- IAMFullAccess
+- NetworkAdministrator
+- AWSKeyManagementServicePowerUser
+
+
+### Download kube-aws and kubectl
+
+#### kube-aws
+
+[kube-aws](https://github.com/coreos/kube-aws) is a CLI tool to automate cluster deployment to AWS.
+##### Verify kube-aws integrity
+Note: if you are using a non-official release (e.g RC release) kube-aws, you can skip this setp.
+Import the CoreOS Application Signing Public Key:
+
+```
+gpg2 --keyserver pgp.mit.edu --recv-key FC8A365E
+```
+
+Validate the key fingerprint:
+
+```
+gpg2 --fingerprint FC8A365E
+```
+The correct key fingerprint is `18AD 5014 C99E F7E3 BA5F 6CE9 50BD D3E0 FC8A 365E`
+
+We can download `kube-aws` from its [release page](https://github.com/coreos/kube-aws/releases). In this tutorial, we use version 0.9.1
+
+Validate the tarball's GPG signature:
+
+```
+PLATFORM=linux-amd64
+ # Or
+PLATFORM=darwin-amd64
+
+gpg2 --verify kube-aws-${PLATFORM}.tar.gz.sig kube-aws-${PLATFORM}.tar.gz
+```
+##### Install kube-aws
+Extract the binary:
+
+```
+tar zxvf kube-aws-${PLATFORM}.tar.gz
+```
+
+Add kube-aws to your path:
+
+```
+mv ${PLATFORM}/kube-aws /usr/local/bin
+```
+
+
+#### kubectl
+
+[kubectl](https://kubernetes.io/docs/user-guide/kubectl-overview/) is a command line interface for running commands against Kubernetes clusters.
+
+Download `kubectl` from the Kubernetes release artifact site with the `curl` tool.
+
+```
+# OS X
+curl -O https://storage.googleapis.com/kubernetes-release/release/"$(curl -s https://storage.googleapis.com/kubernetes-release/release/stable.txt)"/bin/darwin/amd64/kubectl
+
+# Linux
+curl -O https://storage.googleapis.com/kubernetes-release/release/"$(curl -s https://storage.googleapis.com/kubernetes-release/release/stable.txt)"/bin/linux/amd64/kubectl
+```
+
+Make the kubectl binary executable and move it to your PATH (e.g. `/usr/local/bin`):
+
+```
+chmod +x ./kubectl
+sudo mv ./kubectl /usr/local/bin/kubectl
+```
+
+### Configure AWS Credentials
+
+First check out [this](http://docs.aws.amazon.com/cli/latest/userguide/installing.html) for installing the AWS command line interface.
+
+And then configure your AWS account information:
+
+```
+aws configure
+```
+
+
+Fill in the required fields:
+
+
+```
+AWS Access Key ID: YOUR_ACCESS_KEY_ID
+AWS Secrete Access Key: YOUR_SECRETE_ACCESS_KEY
+Default region name: us-west-2
+Default output format: json
+```
+
+`YOUR_ACCESS_KEY_ID`, and `YOUR_SECRETE_ACCESS_KEY` is the IAM key and secret from [Create AWS Account and IAM Account](#create-aws-account-and-iam-account)
+
+Verify that your credentials work by describing any instances you may already have running on your account:
+
+```
+aws ec2 describe-instances
+```
+
+### Define Cluster Parameters
+
+#### EC2 key pair
+
+The keypair that will authenticate SSH access to your EC2 instances. The public half of this key pair will be configured on each CoreOS node.
+
+Follow [EC2 Keypair User Guide](http://docs.aws.amazon.com/AWSEC2/latest/UserGuide/ec2-key-pairs.html) to create a EC2 key pair
+
+After creating a key pair, you will use the key pair name to configure the cluster.
+
+Key pairs are only available to EC2 instances in the same region. We are using us-west-2 in our tutorial, so make sure to creat key pairs in that region (Oregon).
+
+Your browser will download a `key-name.pem` file which is the key to access the EC2 instances. We will use it later.
+
+
+#### KMS key
+
+Amazon KMS keys are used to encrypt and decrypt cluster TLS assets. If you already have a KMS Key that you would like to use, you can skip creating a new key and provide the Arn string for your existing key.
+
+You can create a KMS key with the aws command line tool:
+
+```
+aws kms --region=us-west-2 create-key --description="kube-aws assets"
+{
+    "KeyMetadata": {
+        "CreationDate": 1458235139.724,
+        "KeyState": "Enabled",
+        "Arn": "arn:aws:kms:us-west-2:aaaaaaaaaaaaa:key/xxxxxxxxxxxxxxxxxxx",
+        "AWSAccountId": "xxxxxxxxxxxxx",
+        "Enabled": true,
+        "KeyUsage": "ENCRYPT_DECRYPT",
+        "KeyId": "xxxxxxxxx",
+        "Description": "kube-aws assets"
+    }
+}
+```
+
+We will need to use the value of `Arn` later.
+
+And then let's add several inline policies in your IAM user permission.
+
+Go to [IAM Console](https://console.aws.amazon.com/iam/home?region=us-west-2#/home). Click on button `Users`, click user that we just created, and then click on `Add inline policy` button, and select `Custom Policy`.
+
+Paste into following inline policies:
+
+```
+ (Caution: node_0, node_1, node_2 directories represents PaddlePaddle node and train_id, not the Kubernetes node){
+    "Version": "2012-10-17",
+    "Statement": [
+        {
+            "Sid": "Stmt1482205552000",
+            "Effect": "Allow",
+            "Action": [
+                "kms:Decrypt",
+                "kms:Encrypt"
+            ],
+            "Resource": [
+                "arn:aws:kms:*:AWS_ACCOUNT_ID:key/*"
+            ]
+        },
+		{
+            "Sid": "Stmt1482205746000",
+            "Effect": "Allow",
+            "Action": [
+                "cloudformation:CreateStack",
+                "cloudformation:UpdateStack",
+                "cloudformation:DeleteStack",
+                "cloudformation:DescribeStacks",
+                "cloudformation:DescribeStackResource",
+                "cloudformation:GetTemplate",
+                "cloudformation:DescribeStackEvents"
+            ],
+            "Resource": [
+                "arn:aws:cloudformation:us-west-2:AWS_ACCOUNT_ID:stack/MY_CLUSTER_NAME/*"
+            ]
+        }
+    ]
+}
+```
+`Version` : Its value has to be exactly "2012-10-17".
+`AWS_ACCOUNT_ID`: You can get it from following command line:
+
+```
+aws sts get-caller-identity --output text --query Account
+```
+
+`MY_CLUSTER_NAME`: Pick a MY_CLUSTER_NAME that you like, you will use it later as well. 
+Please note, stack name must satisfy regular expression pattern: [a-zA-Z][-a-zA-Z0-9*]*, which means no "_" or "-" in stack name, or kube-aws will throw error in later steps.
+
+#### External DNS name
+
+When the cluster is created, the controller will expose the TLS-secured API on a DNS name.
+
+DNS name should have a CNAME points to cluster DNS name or an A record points to the cluster IP address.
+
+We will need to use DNS name later in tutorial. If you don't already own one, you can choose any DNS name (e.g., `paddle`) and modify `/etc/hosts` to associate cluster IP with that DNS name for your local machine. And add name service (route53) in aws to associate the IP to paddle for cluster. We will find the cluster IP in later steps.
+
+#### S3 bucket
+
+You need to create an S3 bucket before startup the Kubernetes cluster.
+
+There are some bugs in aws cli in creating S3 bucket, so let's use the [S3 Console](https://console.aws.amazon.com/s3/home?region=us-west-2).
+
+Click on `Create Bucket`, fill in a unique BUCKET_NAME, and make sure region is us-west-2 (Oregon).
+
+
+#### Initialize Assets
+
+Create a directory on your local machine to hold the generated assets:
+
+```
+$ mkdir my-cluster
+$ cd my-cluster
+```
+
+Initialize the cluster CloudFormation stack with the KMS Arn, key pair name, and DNS name from the previous step:
+
+```
+kube-aws init \
+--cluster-name=MY_CLUSTER_NAME \
+--external-dns-name=MY_EXTERNAL_DNS_NAME \
+--region=us-west-2 \
+--availability-zone=us-west-2a \
+--key-name=KEY_PAIR_NAME \
+--kms-key-arn="arn:aws:kms:us-west-2:xxxxxxxxxx:key/xxxxxxxxxxxxxxxxxxx"
+```
+
+`MY_CLUSTER_NAME`: the one you picked in [KMS key](#kms-key)
+
+`MY_EXTERNAL_DNS_NAME`: see [External DNS name](#external-dns-name)
+
+`KEY_PAIR_NAME`: see [EC2 key pair](#ec2-key-pair)
+
+`--kms-key-arn`: the "Arn" in [KMS key](#kms-key)
+
+Here `us-west-2a` is used for parameter `--availability-zone`, but supported availability zone varies among AWS accounts.
+
+Please check if `us-west-2a` is supported by `aws ec2 --region us-west-2 describe-availability-zones`, if not switch to other supported availability zone. (e.g., `us-west-2a`, or `us-west-2b`)
+
+
+There will now be a cluster.yaml file in the asset directory. This is the main configuration file for your cluster.
+
+By default `kube-aws` will only create one worker node. Let's edit `cluster.yaml` and change `workerCount` from 1 to 3.
+
+
+#### Render contents of the asset directory
+
+In the simplest case, you can have kube-aws generate both your TLS identities and certificate authority for you.
+
+```
+kube-aws render credentials --generate-ca
+```
+
+The next command generates the default set of cluster assets in your asset directory.
+
+```
+kube-aws render stack
+```
+Assets (templates and credentials) that are used to create, update and interact with your Kubernetes cluster will be created under your current folder.
+
+
+### Kubernetes Cluster Start Up
+
+#### Create the instances defined in the CloudFormation template
+
+Now let's create your cluster (choose any `PREFIX` for the command below):
+
+```
+kube-aws up --s3-uri s3://BUCKET_NAME/PREFIX
+```
+
+`BUCKET_NAME`: the bucket name that you used in [S3 bucket](#s3-bucket)
+
+
+#### Configure DNS
+
+You can invoke `kube-aws status` to get the cluster API endpoint after cluster creation.
+
+```
+$ kube-aws status
+Cluster Name:		paddle-cluster
+Controller DNS Name:	paddle-cl-ElbAPISe-EEOI3EZPR86C-531251350.us-west-2.elb.amazonaws.com
+```
+
+If you own a DNS name, set the A record to any of the above ip. __Or__ you can set up CNAME point to `Controller DNS Name` (`paddle-cl-ElbAPISe-EEOI3EZPR86C-531251350.us-west-2.elb.amazonaws.com`)
+
+##### Find IP address
+
+Use command `dig` to check the load balancer hostname to get the ip address.
+
+```
+$ dig paddle-cl-ElbAPISe-EEOI3EZPR86C-531251350.us-west-2.elb.amazonaws.com
+
+;; QUESTION SECTION:
+;paddle-cl-ElbAPISe-EEOI3EZPR86C-531251350.us-west-2.elb.amazonaws.com. IN A
+
+;; ANSWER SECTION:
+paddle-cl-ElbAPISe-EEOI3EZPR86C-531251350.us-west-2.elb.amazonaws.com. 59 IN A 54.241.164.52
+paddle-cl-ElbAPISe-EEOI3EZPR86C-531251350.us-west-2.elb.amazonaws.com. 59 IN A 54.67.102.112
+```
+
+In the above output, both ip `54.241.164.52`, `54.67.102.112` will work.
+
+*If you own a DNS name*, set the A record to any of the above ip. Then you can skip to the step "Access the cluster".
+
+*If you do not own a DNS name*:
+##### Update local DNS association
+Edit `/etc/hosts` to associate above ip with the DNS name.
+##### Add Route53 private name service in VPC
+ - Open [Route53 Console](https://console.aws.amazon.com/route53/home)
+ - Create hosted zone with following config
+   - Domain name: "paddle"
+   - Type: "Private hosted zone for amazon VPC"
+   - VPC ID: `<Your VPC ID>`
+
+   ![route53 zone setting](src/route53_create_zone.png)
+ - Add A record
+    - Click on the zone "paddle" just created
+    - Click the button "Create record set"
+        - Name : leave blank
+        - type: "A"
+        - Value: `<kube-controller ec2 private ip>`
+
+        ![route53 create recordset](src/route53_create_recordset.png)
+ - Verify name service
+    - Connect to any instance created by kube-aws via ssh
+    - Run command "host paddle", see if the ip returned is the private ip of kube-controller
+
+#### Access the cluster
+
+Once the API server is running, you should see:
+
+```
+$ kubectl --kubeconfig=kubeconfig get nodes 
+NAME                                       STATUS    AGE
+ip-10-0-0-134.us-west-2.compute.internal   Ready     6m
+ip-10-0-0-238.us-west-2.compute.internal   Ready     6m
+ip-10-0-0-50.us-west-2.compute.internal    Ready     6m
+ip-10-0-0-55.us-west-2.compute.internal    Ready     6m
+```
+
+
+### Setup Elastic File System for Cluster
+
+Training data is usually served on a distributed filesystem, we use Elastic File System (EFS) on AWS.
+
+1. Create security group for EFS in [security group console](https://us-west-2.console.aws.amazon.com/ec2/v2/home?region=us-west-2#SecurityGroups:sort=groupId)
+  1. Look up security group id for `paddle-cluster-sg-worker` (`sg-055ee37d` in the image below)
+  <center>![](src/worker_security_group.png)</center>
+  2. Add security group `paddle-efs` with `ALL TCP` inbound rule and custom source as group id of `paddle-cluster-sg-worker`. And VPC of `paddle-cluster-vpc`. Make sure availability zone is same as the one you used in [Initialize Assets](#initialize-assets).
+  <center>![](src/add_security_group.png)</center>
+
+2. Create the Elastic File System in [EFS console](https://us-west-2.console.aws.amazon.com/efs/home?region=us-west-2#/wizard/1) with `paddle-cluster-vpc` VPC. Make sure subnet is `paddle-cluster-Subnet0` andd security group is `paddle-efs`.
+<center>![](src/create_efs.png)</center>
+
+
+### Start PaddlePaddle Training Demo on AWS
+
+#### Configure Kubernetes Volume that Points to EFS
+
+First we need to create a [PersistentVolume](https://kubernetes.io/docs/user-guide/persistent-volumes/) to provision EFS volumn.
+
+Save following snippet as `pv.yaml`
+```
+apiVersion: v1
+kind: PersistentVolume
+metadata:
+  name: efsvol
+spec:
+  capacity:
+    storage: 100Gi
+  accessModes:
+    - ReadWriteMany
+  nfs:
+    server: EFS_DNS_NAME
+    path: "/"
+```
+
+`EFS_DNS_NAME`: DNS name as shown in description of `paddle-efs` that we created. Looks similar to `fs-2cbf7385.efs.us-west-2.amazonaws.com`
+
+Run following command to create a persistent volumn:
+```
+kubectl --kubeconfig=kubeconfig create -f pv.yaml
+```
+
+Next let's create a [PersistentVolumeClaim](https://kubernetes.io/docs/user-guide/persistent-volumes/) to claim the persistent volume.
+
+Save following snippet as `pvc.yaml`.
+```
+kind: PersistentVolumeClaim
+apiVersion: v1
+metadata:
+  name: efsvol
+spec:
+  accessModes:
+    - ReadWriteMany
+  resources:
+    requests:
+      storage: 50Gi
+```
+
+Run following command to create a persistent volumn claim:
+```
+kubectl --kubeconfig=kubeconfig create -f pvc.yaml
+```
+
+#### Prepare Training Data
+
+We will now launch a kubernetes job that downloads, saves and evenly splits training data into 3 shards on the persistent volumn that we just created.
+
+save following snippet as `paddle-data-job.yaml`
+```
+apiVersion: batch/v1
+kind: Job
+metadata:
+  name: paddle-data
+spec:
+  template:
+    metadata:
+      name: pi
+    spec:
+      containers:
+      - name: paddle-data
+        image: paddledev/paddle-tutorial:k8s_data
+        imagePullPolicy: Always
+        volumeMounts:
+        - mountPath: "/efs"
+          name: efs
+        env:
+        - name: OUT_DIR
+          value: /efs/paddle-cluster-job
+        - name: SPLIT_COUNT
+          value: "3"
+      volumes:
+        - name: efs
+          persistentVolumeClaim:
+            claimName: efsvol
+      restartPolicy: Never
+```
+
+Run following command to launch the job:
+```
+kubectl --kubeconfig=kubeconfig create -f paddle-data-job.yaml
+```
+
+Job may take 7 min to finish, use following command to check job status. Do not proceed until `SUCCESSFUL` for `paddle-data` job is `1`
+```
+$ kubectl --kubeconfig=kubeconfig get jobs
+NAME          DESIRED   SUCCESSFUL   AGE
+paddle-data   1         1            6m
+```
+
+Data preparation is done by docker image `paddledev/paddle-tutorial:k8s_data`, see [here](src/k8s_data/README.md) for how to build this docker image and source code.
+
+#### Start Training
+
+Now we are ready to start paddle training job. Save following snippet as `paddle-cluster-job.yaml`
+```
+apiVersion: batch/v1
+kind: Job
+metadata:
+  name: paddle-cluster-job
+spec:
+  parallelism: 3
+  completions: 3
+  template:
+    metadata:
+      name: paddle-cluster-job
+    spec:
+      volumes:
+      - name: efs
+        persistentVolumeClaim:
+          claimName: efsvol
+      containers:
+      - name: trainer
+        image: paddledev/paddle-tutorial:k8s_train
+        command: ["bin/bash",  "-c", "/root/start.sh"]
+        env:
+        - name: JOB_NAME
+          value: paddle-cluster-job
+        - name: JOB_PATH
+          value: /home/jobpath
+        - name: JOB_NAMESPACE
+          value: default
+        - name: TRAIN_CONFIG_DIR
+          value: quick_start
+        - name: CONF_PADDLE_NIC
+          value: eth0
+        - name: CONF_PADDLE_PORT
+          value: "7164"
+        - name: CONF_PADDLE_PORTS_NUM
+          value: "2"
+        - name: CONF_PADDLE_PORTS_NUM_SPARSE
+          value: "2"
+        - name: CONF_PADDLE_GRADIENT_NUM
+          value: "3"
+        - name: TRAINER_COUNT
+          value: "3"
+        volumeMounts:
+        - mountPath: "/home/jobpath"
+          name: efs
+        ports:
+        - name: jobport0
+          hostPort: 7164
+          containerPort: 7164
+        - name: jobport1
+          hostPort: 7165
+          containerPort: 7165
+        - name: jobport2
+          hostPort: 7166
+          containerPort: 7166
+        - name: jobport3
+          hostPort: 7167
+          containerPort: 7167
+      restartPolicy: Never
+```
+
+`parallelism: 3, completions: 3` means this job will simultaneously start 3 PaddlePaddle pods, and this job will be finished when there are 3 finished pods.
+
+`env` field represents container's environment variables, we specify PaddlePaddle parameters by environment variables.
+
+`ports` indicates that TCP port 7164 - 7167 are exposed for communication between `pserver` ans trainer. port starts continously from `CONF_PADDLE_PORT` (7164) to `CONF_PADDLE_PORT + CONF_PADDLE_PORTS_NUM + CONF_PADDLE_PORTS_NUM_SPARSE - 1` (7167). We use multiple ports for dense and sparse paramter updates to improve latency.
+
+Run following command to launch the job.
+```
+kubectl --kubeconfig=kubeconfig create -f paddle-claster-job.yaml
+```
+
+Inspect individual pods
+
+```
+$ kubectl --kubeconfig=kubeconfig get pods
+NAME                       READY     STATUS    RESTARTS   AGE
+paddle-cluster-job-cm469   1/1       Running   0          9m
+paddle-cluster-job-fnt03   1/1       Running   0          9m
+paddle-cluster-job-jx4xr   1/1       Running   0          9m
+```
+
+Inspect individual console output
+```
+kubectl --kubeconfig=kubeconfig log -f POD_NAME
+```
+
+`POD_NAME`: name of any pod (e.g., `paddle-cluster-job-cm469`).
+
+Run `kubectl --kubeconfig=kubeconfig describe job paddle-cluster-job` to check training job status. It will complete in around 20 minutes.
+
+The details for start `pserver` and `trainer` are hidden inside docker image `paddledev/paddle-tutorial:k8s_train`, see [here](src/k8s_train/README.md) for how to build the docker image and source code.
+
+#### Inspect Training Output
+
+Training output (model snapshot and logs) will be saved in EFS. We can ssh into worker EC2 instance, mount EFS and check training output.
+
+1. ssh Into Worker EC2 instance
+```
+chmod 400 key-name.pem
+ssh -i key-name.pem core@INSTANCE_IP
+```
+
+`INSTANCE_IP`: public IP address of EC2 kubernetes worker node. Go to [EC2 console](https://us-west-2.console.aws.amazon.com/ec2/v2/home?region=us-west-2#Instances:sort=instanceId) and check `public IP` of any `paddle-cluster-kube-aws-worker` instance.
+
+2. Mount EFS
+```
+mkdir efs
+sudo mount -t nfs4 -o nfsvers=4.1,rsize=1048576,wsize=1048576,hard,timeo=600,retrans=2 EFS_DNS_NAME:/ efs
+```
+
+`EFS_DNS_NAME`: DNS name as shown in description of `paddle-efs` that we created. Look similar to `fs-2cbf7385.efs.us-west-2.amazonaws.com`.
+
+Now folder `efs` will have structure similar to:
+```
+-- paddle-cluster-job
+    |-- ...
+    |-- output
+    |   |-- node_0
+    |   |   |-- server.log
+    |   |   `-- train.log
+    |   |-- node_1
+    |   |   |-- server.log
+    |   |   `-- train.log
+    |   |-- node_2
+    |   |   |-- server.log
+    |   |   `-- train.log
+    |   |-- pass-00000
+    |   |   |-- ___fc_layer_0__.w0
+    |   |   |-- ___fc_layer_0__.wbias
+    |   |   |-- done
+    |   |   |-- path.txt
+    |   |   `-- trainer_config.lr.py
+	|   |-- pass-00001...
+```
+`server.log` contains log for `pserver`. `train.log` contains log for `trainer`. Model description and snapshot is stored in `pass-0000*`.
+
+### Kubernetes Cluster Tear Down
+
+#### Delete EFS
+
+Go to [EFS Console](https://us-west-2.console.aws.amazon.com/efs/home?region=us-west-2) and delete the EFS volumn that we created.
+
+#### Delete security group
+
+Go to [Security Group Console](https://us-west-2.console.aws.amazon.com/ec2/v2/home?region=us-west-2#SecurityGroups:sort=groupId) and delete security group `paddle-efs`.
+
+
+#### Delete S3 Bucket
+
+Go to [S3 Console](https://console.aws.amazon.com/s3/home?region=us-west-2#) and delete the S3 bucket that we created.
+
+#### Destroy Cluster
+
+```
+kube-aws destroy
+```
+
+The command will return immediately, but it might take 5 min to tear down the whole cluster.
+
+You can go to [CludFormation Console](https://us-west-2.console.aws.amazon.com/cloudformation/home?region=us-west-2#/stacks?filter=active) to check destroy process.

doc/howto/usage/cluster/k8s_cn.md

+# Kubernetes单机训练
+
+在这篇文档里，我们介绍如何在 Kubernetes 集群上启动一个单机使用CPU的Paddle训练作业。在下一篇中，我们将介绍如何启动分布式训练作业。
+
+## 制作Docker镜像
+
+在一个功能齐全的Kubernetes机群里，通常我们会安装Ceph等分布式文件系统来存储训练数据。这样的话，一个分布式Paddle训练任务中的每个进程都可以从Ceph读取数据。在这个例子里，我们只演示一个单机作业，所以可以简化对环境的要求，把训练数据直接放在
+Paddle的Docker image里。为此，我们需要制作一个包含训练数据的Paddle镜像。
+
+Paddle 的 [Quick Start Tutorial](http://www.paddlepaddle.org/doc/demo/quick_start/index_en.html) 
+里介绍了用Paddle源码中的脚本下载训练数据的过程。
+而 `paddledev/paddle:cpu-demo-latest` 镜像里有 Paddle 源码与demo，（ 请注意，默认的
+Paddle镜像 `paddledev/paddle:cpu-latest` 是不包括源码的, Paddle的各版本镜像可以参考 [Docker installation guide](http://www.paddlepaddle.org/doc/build/docker_install.html) ），所以我们使用这个镜像来下载训练数据到Docker container中，然后把这个包含了训练数据的container保存为一个新的镜像。
+  
+### 运行容器
+
+```
+$ docker run --name quick_start_data -it paddledev/paddle:cpu-demo-latest
+```
+
+### 下载数据
+
+进入容器`/root/paddle/demo/quick_start/data`目录，使用`get_data.sh`下载数据
+
+```
+$ root@fbd1f2bb71f4:~/paddle/demo/quick_start/data# ./get_data.sh
+
+Downloading Amazon Electronics reviews data...
+--2016-10-31 01:33:43--  http://snap.stanford.edu/data/amazon/productGraph/categoryFiles/reviews_Electronics_5.json.gz
+Resolving snap.stanford.edu (snap.stanford.edu)... 171.64.75.80
+Connecting to snap.stanford.edu (snap.stanford.edu)|171.64.75.80|:80... connected.
+HTTP request sent, awaiting response... 200 OK
+Length: 495854086 (473M) [application/x-gzip]
+Saving to: 'reviews_Electronics_5.json.gz'
+
+ 10% [=======>                                         ] 874,279     64.7KB/s  eta 2h 13m
+
+```
+
+### 修改启动脚本
+
+下载完数据后，修改`/root/paddle/demo/quick_start/train.sh`文件，内容如下（增加了一条cd命令）
+```
+set -e
+cd /root/paddle/demo/quick_start
+cfg=trainer_config.lr.py
+#cfg=trainer_config.emb.py
+#cfg=trainer_config.cnn.py
+#cfg=trainer_config.lstm.py
+#cfg=trainer_config.bidi-lstm.py
+#cfg=trainer_config.db-lstm.py
+paddle train \
+  --config=$cfg \
+  --save_dir=./output \
+  --trainer_count=4 \
+  --log_period=20 \
+  --num_passes=15 \
+  --use_gpu=false \
+  --show_parameter_stats_period=100 \
+  --test_all_data_in_one_period=1 \
+  2>&1 | tee 'train.log'
+```
+
+### 提交镜像
+
+修改启动脚本后，退出容器，使用`docker commit`命令创建新镜像。
+
+```
+$ docker commit quick_start_data mypaddle/paddle:quickstart
+```
+
+## 使用 Kubernetes 进行训练
+
+>针对任务运行完成后容器自动退出的场景，Kubernetes有Job类型的资源来支持。下文就是用Job类型的资源来进行训练。
+
+### 编写yaml文件
+
+在训练时，输出结果可能会随着容器的消耗而被删除，需要在创建容器前挂载卷以便我们保存训练结果。使用我们之前构造的镜像，可以创建一个 [Kubernetes Job](http://kubernetes.io/docs/user-guide/jobs/#what-is-a-job)，简单的yaml文件如下：
+
+```
+apiVersion: batch/v1
+kind: Job
+metadata:
+  name: quickstart
+spec:
+  parallelism: 1
+  completions: 1
+  template:
+    metadata:
+      name: quickstart
+    spec:
+      volumes:
+      - name: output
+        hostPath: 
+          path: /home/work/paddle_output     
+      containers:
+      - name: pi
+        image: mypaddle/paddle:quickstart
+        command: ["bin/bash",  "-c", "/root/paddle/demo/quick_start/train.sh"]
+        volumeMounts:
+        - name: output
+          mountPath: /root/paddle/demo/quick_start/output
+      restartPolicy: Never
+```
+
+### 创建Paddle Job
+
+使用上文创建的yaml文件创建Kubernetes Job，命令为：
+
+```
+$ kubectl  create -f paddle.yaml
+```
+
+查看job的详细情况：
+
+```
+$ kubectl  get job
+NAME         DESIRED   SUCCESSFUL   AGE
+quickstart   1         0            58s
+
+$ kubectl  describe job quickstart
+Name:		quickstart
+Namespace:	default
+Image(s):	registry.baidu.com/public/paddle:cpu-demo-latest
+Selector:	controller-uid=f120da72-9f18-11e6-b363-448a5b355b84
+Parallelism:	1
+Completions:	1
+Start Time:	Mon, 31 Oct 2016 11:20:16 +0800
+Labels:		controller-uid=f120da72-9f18-11e6-b363-448a5b355b84,job-name=quickstart
+Pods Statuses:	0 Running / 1 Succeeded / 0 Failed
+Volumes:
+  output:
+    Type:	HostPath (bare host directory volume)
+    Path:	/home/work/paddle_output
+Events:
+  FirstSeen	LastSeen	Count	From			SubobjectPath	Type		Reason			Message
+  ---------	--------	-----	----			-------------	--------	------			-------
+  1m		1m		1	{job-controller }			Normal		SuccessfulCreate	Created pod: quickstart-fa0wx
+```
+
+### 查看训练结果
+
+根据Job对应的Pod信息，可以查看此Pod运行的宿主机。
+
+```
+kubectl  describe pod quickstart-fa0wx
+Name:		quickstart-fa0wx
+Namespace:	default
+Node:		paddle-demo-let02/10.206.202.44
+Start Time:	Mon, 31 Oct 2016 11:20:17 +0800
+Labels:		controller-uid=f120da72-9f18-11e6-b363-448a5b355b84,job-name=quickstart
+Status:		Succeeded
+IP:		10.0.0.9
+Controllers:	Job/quickstart
+Containers:
+  quickstart:
+    Container ID:	docker://b8561f5c79193550d64fa47418a9e67ebdd71546186e840f88de5026b8097465
+    Image:		registry.baidu.com/public/paddle:cpu-demo-latest
+    Image ID:		docker://18e457ce3d362ff5f3febf8e7f85ffec852f70f3b629add10aed84f930a68750
+    Port:
+    Command:
+      bin/bash
+      -c
+      /root/paddle/demo/quick_start/train.sh
+    QoS Tier:
+      cpu:		BestEffort
+      memory:		BestEffort
+    State:		Terminated
+      Reason:		Completed
+      Exit Code:	0
+      Started:		Mon, 31 Oct 2016 11:20:20 +0800
+      Finished:		Mon, 31 Oct 2016 11:21:46 +0800
+    Ready:		False
+    Restart Count:	0
+    Environment Variables:
+Conditions:
+  Type		Status
+  Ready 	False
+Volumes:
+  output:
+    Type:	HostPath (bare host directory volume)
+    Path:	/home/work/paddle_output
+```
+
+我们还可以登录到宿主机上查看训练结果。
+
+```
+[root@paddle-demo-let02 paddle_output]# ll
+total 60
+drwxr-xr-x 2 root root 4096 Oct 31 11:20 pass-00000
+drwxr-xr-x 2 root root 4096 Oct 31 11:20 pass-00001
+drwxr-xr-x 2 root root 4096 Oct 31 11:21 pass-00002
+drwxr-xr-x 2 root root 4096 Oct 31 11:21 pass-00003
+drwxr-xr-x 2 root root 4096 Oct 31 11:21 pass-00004
+drwxr-xr-x 2 root root 4096 Oct 31 11:21 pass-00005
+drwxr-xr-x 2 root root 4096 Oct 31 11:21 pass-00006
+drwxr-xr-x 2 root root 4096 Oct 31 11:21 pass-00007
+drwxr-xr-x 2 root root 4096 Oct 31 11:21 pass-00008
+drwxr-xr-x 2 root root 4096 Oct 31 11:21 pass-00009
+drwxr-xr-x 2 root root 4096 Oct 31 11:21 pass-00010
+drwxr-xr-x 2 root root 4096 Oct 31 11:21 pass-00011
+drwxr-xr-x 2 root root 4096 Oct 31 11:21 pass-00012
+drwxr-xr-x 2 root root 4096 Oct 31 11:21 pass-00013
+drwxr-xr-x 2 root root 4096 Oct 31 11:21 pass-00014
+```

doc/howto/usage/cluster/k8s_distributed_cn.md

+# Kubernetes分布式训练
+
+前一篇文章介绍了如何在Kubernetes集群上启动一个单机PaddlePaddle训练作业 (Job)。在这篇文章里，我们介绍如何在Kubernetes集群上进行分布式PaddlePaddle训练作业。关于PaddlePaddle的分布式训练，文章 [Cluster Training](https://github.com/baidu/Paddle/blob/develop/doc/cluster/opensource/cluster_train.md)介绍了一种通过SSH远程分发任务，进行分布式训练的方法，与此不同的是，本文将介绍在Kubernetes容器管理平台上快速构建PaddlePaddle容器集群，进行分布式训练的方案。
+
+有关Kubernetes相关概念以及如何搭建和配置Kubernetes集群，可以参考[k8s_basis](./k8s_basis_cn.md)。
+
+## 整体方案
+
+在训练之前，用户将配置与训练数据切分好放在分布式文件系统预先分配好的目录中(不同的分布式文件系统，需要使用其制定的方式挂载后并导入数据)，训练时，程序从此目录拷贝文件到容器内进行训练，将结果保存到此目录里。整体的结构图如下：
+
+![paddle on kubernetes结构图](src/k8s-paddle-arch.png)
+
+上图描述了一个3节点的分布式训练场景，在每个Pod上都通过volume方式挂载分布式文件系统的一个目录用于保存训练数据和输出结果。Kubernetes为这次训练创建了3个pod并且调度到了3个node上运行，每个pod包含一个PaddlePaddle容器。在容器创建后，会启动pserver与trainer进程，读取volume中的数据进行这次分布式训练。
+
+根据前文的描述，要在已有的Kubernetes集群上进行PaddlePaddle的分布式训练，按照下面步骤即可：
+
+1. [制作PaddlePaddle镜像](#制作镜像)
+1. [将训练文件与切分好的数据上传到共享存储](#上传训练文件)
+1. [编写本次训练的YAML文件，创建一个Kubernetes job](#创建Job)
+1. [训练结束后查看输出结果](#查看输出)
+
+下面就根据这几个步骤分别介绍。
+
+### 制作镜像
+
+PaddlePaddle镜像需要提供`paddle pserver`与`paddle train`进程的运行环境，用这个镜像创建的容器需要有以下两个功能：
+
+- 拷贝训练文件到容器内
+- 生成`paddle pserver`与`paddle train`进程的启动参数，并且启动训练
+
+因为官方镜像 `paddledev/paddle:cpu-latest` 内已经包含PaddlePaddle的执行程序但是还没上述功能，所以我们可以在这个基础上，添加启动脚本，制作新镜像来完成以上的工作。参考镜像的[*Dockerfile*](https://github.com/PaddlePaddle/Paddle/blob/develop/doc/howto/usage/cluster/k8s/src/k8s_train/Dockerfile)。
+
+```bash
+$ cd doc/howto/usage/k8s/src/k8s_train
+$ docker build -t [YOUR_REPO]/paddle:mypaddle .
+```
+
+然后将构建成功的镜像上传到镜像仓库。
+
+```bash
+docker push  [YOUR_REPO]/paddle:mypaddle
+```
+
+注意上述命令中`[YOUR_REPO]`表示读者所使用的Docker镜像仓库地址，读者需要替换成自己使用的仓库地址。下文使用`[YOUR_REPO]/paddle:mypaddle`这个地址来表示此步骤所构建出的镜像。
+
+### 准备训练数据
+
+这里我们通过在Kubernetes集群上启动一个Job来下载并切割数据，也可以通过修改[k8s_train](./src/k8s_train/README.md)的内容来定制image.
+
+在启动Job之前，需要根据不同的分布式存储来绑定一个[persistentVolumeClaim](https://kubernetes.io/docs/user-guide/persistent-volumes/),生成的数据将会存储在这个volume下.
+
+```yaml
+apiVersion: batch/v1
+kind: Job
+metadata:
+  name: paddle-data
+spec:
+  template:
+    metadata:
+      name: pi
+    spec:
+      hostNetwork: true
+      containers:
+      - name: paddle-data
+        image: paddledev/paddle-tutorial:k8s_data
+        imagePullPolicy: Always
+        volumeMounts:
+        - mountPath: "/mnt"
+          name: nfs
+        env:
+        - name: OUT_DIR
+          value: /home/work/mfs/paddle-cluster-job
+        - name: SPLIT_COUNT
+          value: "3"
+      volumes:
+        - name: nfs
+          persistentVolumeClaim:
+            claimName: mfs
+      restartPolicy: Never
+```
+
+完成后volume中的文件内容大致如下：
+```base
+[root@paddle-kubernetes-node0 nfsdir]$ tree -d
+.
+`-- paddle-cluster-job
+    |-- 0
+    |   `-- data
+    |-- 1
+    |   `-- data
+    |-- 2
+    |   `-- data
+    |-- output
+    |-- quick_start
+```
+
+目录中paddle-cluster-job是本次训练对应的job name，本次训练要求有3个PaddlePaddle节点，在paddle-cluster-job/data目录中存放切分好的数据，文件夹0，1，2分别代表3个节点的trainer_id。recommendation文件夹内存放训练文件，output文件夹存放训练结果与日志。
+
+### 创建Job
+
+Kubernetes可以通过YAML文件来创建相关对象，然后可以使用命令行工具创建job。
+
+Job YAML文件描述了这次训练使用的Docker镜像，需要启动的节点个数以及 `paddle pserver`与 `paddle train`进程启动的必要参数，也描述了容器需要使用的存储卷挂载的情况。YAML文件中各个字段的具体含义，可以查看[Kubernetes Job API](http://kubernetes.io/docs/api-reference/batch/v1/definitions/#_v1_job)。例如，本次训练的YAML文件可以写成：
+
+```yaml
+apiVersion: batch/v1
+kind: Job
+metadata:
+  name: paddle-cluster-job
+spec:
+  parallelism: 3
+  completions: 3
+  template:
+    metadata:
+      name: paddle-cluster-job
+    spec:
+      volumes:
+      - name: jobpath
+        hostPath:
+          path: /home/work/mfs
+      containers:
+      - name: trainer
+        image: [YOUR_REPO]/paddle:mypaddle
+        command: ["bin/bash",  "-c", "/root/start.sh"]
+        env:
+        - name: JOB_NAME
+          value: paddle-cluster-job
+        - name: JOB_PATH
+          value: /home/jobpath
+        - name: JOB_NAMESPACE
+          value: default
+        - name: TRAIN_CONFIG_DIR
+          value: recommendation
+        - name: CONF_PADDLE_NIC
+          value: eth0
+        - name: CONF_PADDLE_PORT
+          value: "7164"
+        - name: CONF_PADDLE_PORTS_NUM
+          value: "2"
+        - name: CONF_PADDLE_PORTS_NUM_SPARSE
+          value: "2"
+        - name: CONF_PADDLE_GRADIENT_NUM
+          value: "3"
+        volumeMounts:
+        - name: jobpath
+          mountPath: /home/jobpath
+      restartPolicy: Never
+```
+
+文件中，`metadata`下的`name`表示这个job的名字。`parallelism，completions`字段表示这个job会同时开启3个PaddlePaddle节点，成功训练且退出的pod数目为3时，这个job才算成功结束。然后申明一个存储卷`jobpath`，代表宿主机目录`/home/work/mfs`，在对容器的描述`containers`字段中，将此目录挂载为容器的`/home/jobpath`目录，这样容器的`/home/jobpath`目录就成为了共享存储，放在这个目录里的文件其实是保存到了MFS上。
+
+`env`字段表示容器的环境变量，我们将`paddle`运行的一些参数通过这种方式传递到容器内。
+
+环境变量 | 说明
+--- | ---
+JOB_PATH | 共享存储挂在的路径
+JOB_NAME | Job的名字
+TRAIN_CONFIG_DIR | 本次训练文件所在目录，与JOB_PATH,JOB_NAME组合可以找到本次训练需要的文件路径
+CONF_PADDLE_NIC | `paddle pserver`进程需要的`--nics`参数，即网卡名
+CONF_PADDLE_PORT | `paddle paserver`的`--port`参数
+CONF_PADDLE_PORTS_NUM | 稠密更新的端口数量，即`--ports_num`参数
+CONF_PADDLE_PORTS_NUM_SPARSE | 稀疏更新的端口数量，即`--ports_num_for_sparse`参数
+CONF_PADDLE_GRADIENT_NUM | 训练节点数量，即`--num_gradient_servers参数`
+
+这些参数的具体描述，读者可以查看[这里](http://www.paddlepaddle.org/doc/ui/cmd_argument/detail_introduction.html#parameter-server-and-distributed-communication)。
+
+编写完YAML文件后，可以使用Kubernetes的命令行工具创建job。
+
+```bash
+kubectl create -f job.yaml
+```
+
+创建成功后，Kubernetes就会创建3个pod作为PaddlePaddle节点然后拉取镜像，启动容器开始训练。
+
+
+### 查看输出
+
+在训练过程中，可以在共享存储上查看输出的日志和模型，例如output目录下就存放了输出结果。注意node_0，node_1，node_2这几个目录表示PaddlePaddle节点与trainer_id，并不是Kubernetes中的node概念。
+
+```bash
+[root@paddle-kubernetes-node0 output]# tree -d
+.
+├── node_0
+│   ├── server.log
+│   └── train.log
+├── node_1
+│   ├── server.log
+│   └── train.log
+├── node_2
+......
+├── pass-00002
+│   ├── done
+│   ├── ___embedding_0__.w0
+│   ├── ___embedding_1__.w0
+......
+```
+
+我们可以通过日志查看容器训练的情况，例如：
+
+```bash
+[root@paddle-kubernetes-node0 node_0]# cat train.log
+I1116 09:10:17.123121    50 Util.cpp:155] commandline:
+ /usr/local/bin/../opt/paddle/bin/paddle_trainer
+    --nics=eth0 --port=7164
+    --ports_num=2 --comment=paddle_process_by_paddle
+    --pservers=192.168.129.66,192.168.223.143,192.168.129.71
+    --ports_num_for_sparse=2 --config=./trainer_config.py
+    --trainer_count=4 --num_passes=10 --use_gpu=0
+    --log_period=50 --dot_period=10 --saving_period=1
+    --local=0 --trainer_id=0
+    --save_dir=/home/jobpath/paddle-cluster-job/output
+I1116 09:10:17.123440    50 Util.cpp:130] Calling runInitFunctions
+I1116 09:10:17.123764    50 Util.cpp:143] Call runInitFunctions done.
+[WARNING 2016-11-16 09:10:17,227 default_decorators.py:40] please use keyword arguments in paddle config.
+[INFO 2016-11-16 09:10:17,239 networks.py:1282] The input order is [movie_id, title, genres, user_id, gender, age, occupation, rating]
+[INFO 2016-11-16 09:10:17,239 networks.py:1289] The output order is [__square_error_cost_0__]
+I1116 09:10:17.392917    50 Trainer.cpp:170] trainer mode: Normal
+I1116 09:10:17.613910    50 PyDataProvider2.cpp:257] loading dataprovider dataprovider::process
+I1116 09:10:17.680917    50 PyDataProvider2.cpp:257] loading dataprovider dataprovider::process
+I1116 09:10:17.681543    50 GradientMachine.cpp:134] Initing parameters..
+I1116 09:10:18.012390    50 GradientMachine.cpp:141] Init parameters done.
+I1116 09:10:18.018641    50 ParameterClient2.cpp:122] pserver 0 192.168.129.66:7164
+I1116 09:10:18.018950    50 ParameterClient2.cpp:122] pserver 1 192.168.129.66:7165
+I1116 09:10:18.019069    50 ParameterClient2.cpp:122] pserver 2 192.168.223.143:7164
+I1116 09:10:18.019492    50 ParameterClient2.cpp:122] pserver 3 192.168.223.143:7165
+I1116 09:10:18.019716    50 ParameterClient2.cpp:122] pserver 4 192.168.129.71:7164
+I1116 09:10:18.019836    50 ParameterClient2.cpp:122] pserver 5 192.168.129.71:7165
+```
+
+
+## 一些细节的补充
+
+### 使用环境变量
+
+使用容器方式运行训练任务的Kubernetes Job，通常会使用环境变量配置Job的配置信息`start_paddle.py`提供了一个启动脚本，将环境变量转换成paddle的命令行参数：
+```
+API = "/api/v1/namespaces/"
+JOBSELECTOR = "labelSelector=job-name="
+JOB_PATH = os.getenv("JOB_PATH") + "/" + os.getenv("JOB_NAME")
+JOB_PATH_OUTPUT = JOB_PATH + "/output"
+JOBNAME = os.getenv("JOB_NAME")
+NAMESPACE = os.getenv("JOB_NAMESPACE")
+PADDLE_NIC = os.getenv("CONF_PADDLE_NIC")
+PADDLE_PORT = os.getenv("CONF_PADDLE_PORT")
+PADDLE_PORTS_NUM = os.getenv("CONF_PADDLE_PORTS_NUM")
+PADDLE_PORTS_NUM_SPARSE = os.getenv("CONF_PADDLE_PORTS_NUM_SPARSE")
+PADDLE_SERVER_NUM = os.getenv("CONF_PADDLE_GRADIENT_NUM")
+```
+
+### Pod间通信
+`start_paddle.py`脚本开始时，会先进行参数的初始化与解析。
+
+```python
+parser = argparse.ArgumentParser(prog="start_paddle.py",
+                                     description='simple tool for k8s')
+    args, train_args_list = parser.parse_known_args()
+    train_args = refine_unknown_args(train_args_list)
+    train_args_dict = dict(zip(train_args[:-1:2], train_args[1::2]))
+    podlist = getPodList()
+```
+
+然后通过函数`getPodList()`访问Kubernetes的接口来查询此job对应的所有pod信息。当所有pod都处于running状态（容器运行都运行）时，再通过函数`getIdMap(podlist)`获取trainer_id。
+
+```python
+    podlist = getPodList()
+    # need to wait until all pods are running
+    while not isPodAllRunning(podlist):
+        time.sleep(10)
+        podlist = getPodList()
+    idMap = getIdMap(podlist)
+```
+* *注意*: `getPodList()`会获取当前namespace下的所有pod，如果已经有pod运行，可能会导致出错。这种集群节点管理方式会在将来使用[statfulsets](https://kubernetes.io/docs/concepts/abstractions/controllers/statefulsets/)代替。
+
+在函数`getIdMap(podlist)`内部，我们通过读取`podlist`中每个pod的IP地址，将IP排序生成的序号作为trainer_id。
+
+```python
+def getIdMap(podlist):
+    '''
+    generate tainer_id by ip
+    '''
+    ips = []
+    for pod in podlist["items"]:
+        ips.append(pod["status"]["podIP"])
+    ips.sort()
+    idMap = {}
+    for i in range(len(ips)):
+        idMap[ips[i]] = i
+    return idMap
+```
+
+在得到`idMap`后，通过函数`startPaddle(idMap, train_args_dict)`构造`paddle pserver`与`paddle train`的启动参数并执行进程。
+
+### 启动任务
+
+在函数`startPaddle`中，最主要的工作就是解析出`paddle pserver`与`paddle train`的启动参数。例如`paddle train`参数的解析，解析环境变量得到`PADDLE_NIC`，`PADDLE_PORT`，`PADDLE_PORTS_NUM`等参数，然后通过自身的IP地址在`idMap`中获取`trainerId`。
+
+```python
+    program = 'paddle train'
+    args = " --nics=" + PADDLE_NIC
+    args += " --port=" + str(PADDLE_PORT)
+    args += " --ports_num=" + str(PADDLE_PORTS_NUM)
+    args += " --comment=" + "paddle_process_by_paddle"
+    ip_string = ""
+    for ip in idMap.keys():
+        ip_string += (ip + ",")
+    ip_string = ip_string.rstrip(",")
+    args += " --pservers=" + ip_string
+    args_ext = ""
+    for key, value in train_args_dict.items():
+        args_ext += (' --' + key + '=' + value)
+    localIP = socket.gethostbyname(socket.gethostname())
+    trainerId = idMap[localIP]
+    args += " " + args_ext + " --trainer_id=" + \
+        str(trainerId) + " --save_dir=" + JOB_PATH_OUTPUT
+```

doc/howto/usage/cluster/k8s_en.md

+# Paddle On Kubernetes
+
+>In this article, we will introduce how to run Paddle training job on single CPU machine using Kubernetes. In next article, we will introduce how to run Paddle training job on distributed cluster.
+
+## Build Docker Image
+
+In distributed Kubernetes cluster, we will use Ceph or other shared storage system for storing training related data so that all processes in Paddle training can retrieve data from Ceph. In this example, we will only demo training job on single machine. In order to simplify the requirement of the environment, we will directly put training data into Paddle's Docker Image, so we need to create a Paddle Docker image that already includes the training data.
+
+Paddle's [Quick Start Tutorial](http://www.paddlepaddle.org/doc/demo/quick_start/index_en.html) introduces how to download and train data by using script from Paddle's source code.
+And `paddledev/paddle:cpu-demo-latest` image has the Paddle source code and demo. (Caution: Default Paddle image `paddledev/paddle:cpu-latest` doesn't include the source code, Paddle's different versions of image can be referred here: [Docker installation guide](http://www.paddlepaddle.org/doc/build/docker_install.html)), so we run this container and download the training data, and then commit the whole container to be a new Docker image.
+  
+### Run Docker Container
+
+```
+$ docker run --name quick_start_data -it paddledev/paddle:cpu-demo-latest
+```
+
+### Download Training Data
+
+Getting into `/root/paddle/demo/quick_start/data` Directory，using `get_data.sh` to download training data.
+Then getting into `/root/paddle/demo/quick_start` Directory, using `preprocess.sh` to pre-process training data.
+
+```
+$ root@fbd1f2bb71f4:~/paddle/demo/quick_start/data# ./get_data.sh
+
+Downloading Amazon Electronics reviews data...
+--2016-10-31 01:33:43--  http://snap.stanford.edu/data/amazon/productGraph/categoryFiles/reviews_Electronics_5.json.gz
+Resolving snap.stanford.edu (snap.stanford.edu)... 171.64.75.80
+Connecting to snap.stanford.edu (snap.stanford.edu)|171.64.75.80|:80... connected.
+HTTP request sent, awaiting response... 200 OK
+Length: 495854086 (473M) [application/x-gzip]
+Saving to: 'reviews_Electronics_5.json.gz'
+
+ 10% [=======>                                         ] 874,279     64.7KB/s  eta 2h 13m
+
+```
+
+### Modify Startup Script
+
+After downloading the data，modify `/root/paddle/demo/quick_start/train.sh` file contents are as follows (one more cd cmd):
+```
+set -e
+cd /root/paddle/demo/quick_start
+cfg=trainer_config.lr.py
+#cfg=trainer_config.emb.py
+#cfg=trainer_config.cnn.py
+#cfg=trainer_config.lstm.py
+#cfg=trainer_config.bidi-lstm.py
+#cfg=trainer_config.db-lstm.py
+paddle train \
+  --config=$cfg \
+  --save_dir=./output \
+  --trainer_count=4 \
+  --log_period=20 \
+  --num_passes=15 \
+  --use_gpu=false \
+  --show_parameter_stats_period=100 \
+  --test_all_data_in_one_period=1 \
+  2>&1 | tee 'train.log'
+```
+
+### Commit Docker Image
+
+```
+$ docker commit quick_start_data mypaddle/paddle:quickstart
+```
+
+## Use Kubernetes For Training
+
+>We will use Kubernetes job for training process, following steps shows how to do the training with Kubernetes.
+
+### Create Yaml Files
+
+The output result in container will be demolished when job finished (container stopped running), so we need to mount the volume out to the local disk when creating the container to store the training result. Using our previously created image, we can create a [Kubernetes Job](http://kubernetes.io/docs/user-guide/jobs/#what-is-a-job), the yaml contents are as follows:
+
+```
+apiVersion: batch/v1
+kind: Job
+metadata:
+  name: quickstart
+spec:
+  parallelism: 1
+  completions: 1
+  template:
+    metadata:
+      name: quickstart
+    spec:
+      volumes:
+      - name: output
+        hostPath: 
+          path: /home/work/paddle_output     
+      containers:
+      - name: pi
+        image: mypaddle/paddle:quickstart
+        command: ["bin/bash",  "-c", "/root/paddle/demo/quick_start/train.sh"]
+        volumeMounts:
+        - name: output
+          mountPath: /root/paddle/demo/quick_start/output
+      restartPolicy: Never
+```
+
+### Start Paddle Job
+
+Using the above yaml file to start the Kubernetes job.
+
+```
+$ kubectl  create -f paddle.yaml
+```
+
+Get the detailed status of the job:
+
+```
+$ kubectl  get job
+NAME         DESIRED   SUCCESSFUL   AGE
+quickstart   1         0            58s
+
+$ kubectl  describe job quickstart
+Name:		quickstart
+Namespace:	default
+Image(s):	registry.baidu.com/public/paddle:cpu-demo-latest
+Selector:	controller-uid=f120da72-9f18-11e6-b363-448a5b355b84
+Parallelism:	1
+Completions:	1
+Start Time:	Mon, 31 Oct 2016 11:20:16 +0800
+Labels:		controller-uid=f120da72-9f18-11e6-b363-448a5b355b84,job-name=quickstart
+Pods Statuses:	0 Running / 1 Succeeded / 0 Failed
+Volumes:
+  output:
+    Type:	HostPath (bare host directory volume)
+    Path:	/home/work/paddle_output
+Events:
+  FirstSeen	LastSeen	Count	From			SubobjectPath	Type		Reason			Message
+  ---------	--------	-----	----			-------------	--------	------			-------
+  1m		1m		1	{job-controller }			Normal		SuccessfulCreate	Created pod: quickstart-fa0wx
+```
+
+### Get Training Result
+
+We can use kubectl command to take a look at the status of related pod.
+
+```
+$ kubectl  describe pod quickstart-fa0wx
+Name:		quickstart-fa0wx
+Namespace:	default
+Node:		paddle-demo-let02/10.206.202.44
+Start Time:	Mon, 31 Oct 2016 11:20:17 +0800
+Labels:		controller-uid=f120da72-9f18-11e6-b363-448a5b355b84,job-name=quickstart
+Status:		Succeeded
+IP:		10.0.0.9
+Controllers:	Job/quickstart
+Containers:
+  quickstart:
+    Container ID:	docker://b8561f5c79193550d64fa47418a9e67ebdd71546186e840f88de5026b8097465
+    Image:		registry.baidu.com/public/paddle:cpu-demo-latest
+    Image ID:		docker://18e457ce3d362ff5f3febf8e7f85ffec852f70f3b629add10aed84f930a68750
+    Port:
+    Command:
+      bin/bash
+      -c
+      /root/paddle/demo/quick_start/train.sh
+    QoS Tier:
+      cpu:		BestEffort
+      memory:		BestEffort
+    State:		Terminated
+      Reason:		Completed
+      Exit Code:	0
+      Started:		Mon, 31 Oct 2016 11:20:20 +0800
+      Finished:		Mon, 31 Oct 2016 11:21:46 +0800
+    Ready:		False
+    Restart Count:	0
+    Environment Variables:
+Conditions:
+  Type		Status
+  Ready 	False
+Volumes:
+  output:
+    Type:	HostPath (bare host directory volume)
+    Path:	/home/work/paddle_output
+```
+
+We can also ssh to Kubernetes node to take a look at the training result.
+
+```
+[root@paddle-demo-let02 paddle_output]# ll
+total 60
+drwxr-xr-x 2 root root 4096 Oct 31 11:20 pass-00000
+drwxr-xr-x 2 root root 4096 Oct 31 11:20 pass-00001
+drwxr-xr-x 2 root root 4096 Oct 31 11:21 pass-00002
+drwxr-xr-x 2 root root 4096 Oct 31 11:21 pass-00003
+drwxr-xr-x 2 root root 4096 Oct 31 11:21 pass-00004
+drwxr-xr-x 2 root root 4096 Oct 31 11:21 pass-00005
+drwxr-xr-x 2 root root 4096 Oct 31 11:21 pass-00006
+drwxr-xr-x 2 root root 4096 Oct 31 11:21 pass-00007
+drwxr-xr-x 2 root root 4096 Oct 31 11:21 pass-00008
+drwxr-xr-x 2 root root 4096 Oct 31 11:21 pass-00009
+drwxr-xr-x 2 root root 4096 Oct 31 11:21 pass-00010
+drwxr-xr-x 2 root root 4096 Oct 31 11:21 pass-00011
+drwxr-xr-x 2 root root 4096 Oct 31 11:21 pass-00012
+drwxr-xr-x 2 root root 4096 Oct 31 11:21 pass-00013
+drwxr-xr-x 2 root root 4096 Oct 31 11:21 pass-00014
+```

doc/howto/usage/cluster/openmpi_cn.md

+# 在OpenMPI集群中提交训练作业
+
+## 准备OpenMPI集群
+
+执行下面的命令以启动3个节点的OpenMPI集群和一个"head"节点：
+
+```bash
+paddle/scripts/cluster_train_v2/openmpi/docker_cluster
+kubectl create -f head.yaml
+kubectl create -f mpi-nodes.yaml
+```
+
+然后可以从head节点ssh无密码登录到OpenMPI的每个节点上。
+
+## 启动集群作业
+
+您可以按照下面的步骤在OpenMPI集群中提交paddle训练任务：
+
+```bash
+# 获得head和node节点的IP地址
+kubectl get po -o wide
+# 将node节点的IP地址保存到machines文件中
+kubectl get po -o wide | grep nodes | awk '{print $6}' > machines
+# 拷贝必要的文件到head节点
+scp -i ssh/id_rsa.mpi.pub machines prepare.py train.py start_mpi_train.sh tutorial@[headIP]:~
+# ssh 登录到head节点
+ssh -i ssh/id_rsa.mpi.pub tutorial@[headIP]
+# --------------- 以下操作均在head节点中执行 ---------------
+# 准备训练数据
+python prepare.py
+# 拷贝训练程序和字典文件到每台MPI节点
+cat machines | xargs -i scp word_dict.pickle train.py start_mpi_train.sh machines {}:/home/tutorial
+# 创建日志目录
+mpirun -hostfile machines -n 3 mkdir /home/tutorial/logs
+# 拷贝训练数据到各自的节点
+scp train.txt-00000 test.txt-00000 [node1IP]:/home/tutorial
+scp train.txt-00001 test.txt-00001 [node2IP]:/home/tutorial
+scp train.txt-00002 test.txt-00002 [node3IP]:/home/tutorial
+# 启动训练任务
+mpirun -hostfile machines -n 3  /home/tutorial/start_mpi_train.sh
+```

doc/howto/usage/cluster/openmpi_en.md

+# Cluster Training Using OpenMPI
+
+## Prepare an OpenMPI cluster
+
+Run the following command to start a 3-node MPI cluster and one "head" node.
+
+```bash
+cd paddle/scripts/cluster_train_v2/openmpi/docker_cluster
+kubectl create -f head.yaml
+kubectl create -f mpi-nodes.yaml
+```
+
+Then you can log in to every OpenMPI node using ssh without input any passwords.
+
+## Launching Cluster Job
+
+Follow the steps to launch a PaddlePaddle training job in OpenMPI cluster:\
+
+```bash
+# find out node IP addresses
+kubectl get po -o wide
+# generate a "machines" file containing node IP addresses
+kubectl get po -o wide | grep nodes | awk '{print $6}' > machines
+# copy necessary files onto "head" node
+scp -i ssh/id_rsa.mpi.pub machines prepare.py train.py start_mpi_train.sh tutorial@[headIP]:~
+# login to head node using ssh
+ssh -i ssh/id_rsa.mpi.pub tutorial@[headIP]
+# --------------- in head node ---------------
+# prepare training data
+python prepare.py
+# copy training data and dict file to MPI nodes
+cat machines | xargs -i scp word_dict.pickle train.py start_mpi_train.sh machines {}:/home/tutorial
+# creat a directory for storing log files
+mpirun -hostfile machines -n 3 mkdir /home/tutorial/logs
+# copy training data to every node
+scp train.txt-00000 test.txt-00000 [node1IP]:/home/tutorial
+scp train.txt-00001 test.txt-00001 [node2IP]:/home/tutorial
+scp train.txt-00002 test.txt-00002 [node3IP]:/home/tutorial
+# start the job
+mpirun -hostfile machines -n 3  /home/tutorial/start_mpi_train.sh
+```

doc/howto/usage/cluster/src/Dockerfile

+FROM paddledev/paddle:cpu-latest
+
+MAINTAINER zjsxzong89@gmail.com
+
+COPY start.sh /root/
+COPY start_paddle.py /root/
+CMD ["bash"," -c","/root/start.sh"]
\ No newline at end of file

doc/howto/usage/cluster/src/k8s_data/Dockerfile

+FROM alpine
+
+RUN apk update && apk upgrade && apk add coreutils
+ADD quick_start /quick_start
+ADD get_data.sh /bin/
+RUN chmod +x /bin/get_data.sh
+ENTRYPOINT ["/bin/get_data.sh"]

doc/howto/usage/cluster/src/k8s_data/README.md

+To build PaddlePaddle data preparation image in tutorial [Distributed PaddlePaddle Training on AWS with Kubernetes](../../k8s_aws_en.md), run following commands:
+
+```
+cp -r ../../../../../../demo/quick_start .
+docker build . -t prepare-data-image-name
+```

doc/howto/usage/cluster/src/k8s_data/get_data.sh

+#!/bin/sh
+
+out_dir=$OUT_DIR
+split_count=$SPLIT_COUNT
+
+set -e
+
+mkdir -p $out_dir
+cp -r /quick_start $out_dir/
+
+mkdir -p $out_dir/0/data
+cd $out_dir/0/data
+wget http://paddlepaddle.bj.bcebos.com/demo/quick_start_preprocessed_data/preprocessed_data.tar.gz
+tar zxvf preprocessed_data.tar.gz
+rm preprocessed_data.tar.gz
+
+split -d --number=l/$split_count -a 5 train.txt train.
+mv train.00000 train.txt
+
+cd $out_dir
+end=$(expr $split_count - 1)
+for i in $(seq 1 $end); do
+    mkdir -p $i/data
+    cp -r 0/data/* $i/data
+    mv $i/data/train.`printf %05d $i` $i/data/train.txt
+done;

doc/howto/usage/cluster/src/k8s_train/Dockerfile

+FROM paddledev/paddle:cpu-latest
+
+COPY start.sh /root/
+COPY start_paddle.py /root/
+RUN chmod +x /root/start.sh
+CMD ["bash"," -c","/root/start.sh"]

doc/howto/usage/cluster/src/k8s_train/README.md

+To build PaddlePaddle training image in tutorial [Distributed PaddlePaddle Training on AWS with Kubernetes](../../k8s_aws_en.md), run following command:
+
+```
+docker build . -t train-image-name
+```

doc/howto/usage/cluster/src/k8s_train/start.sh

+#!/bin/sh
+
+set -eu
+
+jobconfig=${JOB_PATH}"/"${JOB_NAME}"/"${TRAIN_CONFIG_DIR}
+cd /root
+cp -rf $jobconfig/* .
+
+python /root/start_paddle.py \
+  --dot_period=10 \
+  --ports_num=$CONF_PADDLE_PORTS_NUM \
+  --ports_num_for_sparse=$CONF_PADDLE_PORTS_NUM_SPARSE \
+  --log_period=50 \
+  --num_passes=10 \
+  --trainer_count=$TRAINER_COUNT \
+  --saving_period=1 \
+  --local=0 \
+  --config=trainer_config.lr.py \
+  --use_gpu=0

doc/howto/usage/cluster/src/k8s_train/start_paddle.py

+#!/usr/bin/python
+# Copyright (c) 2016 PaddlePaddle Authors. All Rights Reserved
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import requests
+import time
+import socket
+import os
+import argparse
+
+# configuration for cluster
+API = "/api/v1/namespaces/"
+JOBSELECTOR = "labelSelector=job-name="
+JOB_PATH = os.getenv("JOB_PATH") + "/" + os.getenv("JOB_NAME")
+JOB_PATH_OUTPUT = JOB_PATH + "/output"
+JOBNAME = os.getenv("JOB_NAME")
+NAMESPACE = os.getenv("JOB_NAMESPACE")
+PADDLE_NIC = os.getenv("CONF_PADDLE_NIC")
+PADDLE_PORT = os.getenv("CONF_PADDLE_PORT")
+PADDLE_PORTS_NUM = os.getenv("CONF_PADDLE_PORTS_NUM")
+PADDLE_PORTS_NUM_SPARSE = os.getenv("CONF_PADDLE_PORTS_NUM_SPARSE")
+PADDLE_SERVER_NUM = os.getenv("CONF_PADDLE_GRADIENT_NUM")
+
+tokenpath = '/var/run/secrets/kubernetes.io/serviceaccount/token'
+
+
+def refine_unknown_args(cmd_args):
+    '''
+    refine unknown parameters to handle some special parameters
+    '''
+    new_args = []
+    for arg in cmd_args:
+        if arg.startswith("--") and arg.find("=") != -1:
+            equal_pos = arg.find("=")  # find first = pos
+            arglist = list(arg)
+            arglist[equal_pos] = " "
+            arg = "".join(arglist)
+            arg = arg.lstrip("-")
+            new_args += arg.split(" ")
+        elif arg.startswith("--") and arg.find("=") == -1:
+            arg = arg.lstrip("-")
+            new_args.append(arg)
+        else:
+            new_args.append(arg)
+    return new_args
+
+
+def isPodAllRunning(podlist):
+    '''
+    check all pod is running
+    '''
+    require = len(podlist["items"])
+    running = 0
+    for pod in podlist["items"]:
+        if pod["status"]["phase"] == "Running":
+            running += 1
+    print "waiting for pods running, require:", require, "running:", running
+    if require == running:
+        return True
+    return False
+
+
+def getPodList():
+    '''
+    get all container status of the job
+    '''
+    apiserver = "https://" + \
+        os.getenv("KUBERNETES_SERVICE_HOST") + ":" + \
+        os.getenv("KUBERNETES_SERVICE_PORT_HTTPS")
+
+    pod = API + NAMESPACE + "/pods?"
+    job = JOBNAME
+    if os.path.isfile(tokenpath):
+        tokenfile = open(tokenpath, mode='r')
+        token = tokenfile.read()
+        Bearer = "Bearer " + token
+        headers = {"Authorization": Bearer}
+        return requests.get(apiserver + pod + JOBSELECTOR + job,
+                            headers=headers,
+                            verify=False).json()
+    else:
+        return requests.get(apiserver + pod + JOBSELECTOR + job,
+                            verify=False).json()
+
+
+def getIdMap(podlist):
+    '''
+    generate tainer_id by ip
+    '''
+    ips = []
+    for pod in podlist["items"]:
+        ips.append(pod["status"]["podIP"])
+    ips.sort()
+    idMap = {}
+    for i in range(len(ips)):
+        idMap[ips[i]] = i
+    return idMap
+
+
+def startPaddle(idMap={}, train_args_dict=None):
+    '''
+    start paddle pserver and trainer
+    '''
+    program = 'paddle train'
+    args = " --nics=" + PADDLE_NIC
+    args += " --port=" + str(PADDLE_PORT)
+    args += " --ports_num=" + str(PADDLE_PORTS_NUM)
+    args += " --comment=" + "paddle_process_by_paddle"
+    ip_string = ""
+    for ip in idMap.keys():
+        ip_string += (ip + ",")
+    ip_string = ip_string.rstrip(",")
+    args += " --pservers=" + ip_string
+    args_ext = ""
+    for key, value in train_args_dict.items():
+        args_ext += (' --' + key + '=' + value)
+    localIP = socket.gethostbyname(socket.gethostname())
+    trainerId = idMap[localIP]
+    args += " " + args_ext + " --trainer_id=" + \
+        str(trainerId) + " --save_dir=" + JOB_PATH_OUTPUT
+    logDir = JOB_PATH_OUTPUT + "/node_" + str(trainerId)
+    if not os.path.exists(JOB_PATH_OUTPUT):
+        os.makedirs(JOB_PATH_OUTPUT)
+    if not os.path.exists(logDir):
+        os.mkdir(logDir)
+    copyCommand = 'cp -rf ' + JOB_PATH + \
+        "/" + str(trainerId) + "/data/*" + " ./data/"
+    os.system(copyCommand)
+    startPserver = 'nohup paddle pserver' + \
+        " --port=" + str(PADDLE_PORT) + \
+        " --ports_num=" + str(PADDLE_PORTS_NUM) + \
+        " --ports_num_for_sparse=" + str(PADDLE_PORTS_NUM_SPARSE) + \
+        " --nics=" + PADDLE_NIC + \
+        " --comment=" + "paddle_process_by_paddle" + \
+        " --num_gradient_servers=" + str(PADDLE_SERVER_NUM) +\
+        " > " + logDir + "/server.log 2>&1 &"
+    print startPserver
+    os.system(startPserver)
+    # wait until pservers completely start
+    time.sleep(20)
+    startTrainer = program + args + " 2>&1 | tee " + \
+        logDir + "/train.log"
+    print startTrainer
+    os.system(startTrainer)
+
+
+if __name__ == '__main__':
+    parser = argparse.ArgumentParser(
+        prog="start_paddle.py", description='simple tool for k8s')
+    args, train_args_list = parser.parse_known_args()
+    train_args = refine_unknown_args(train_args_list)
+    train_args_dict = dict(zip(train_args[:-1:2], train_args[1::2]))
+    podlist = getPodList()
+    # need to wait until all pods are running
+    while not isPodAllRunning(podlist):
+        time.sleep(20)
+        podlist = getPodList()
+    idMap = getIdMap(podlist)
+    startPaddle(idMap, train_args_dict)