Add custom image sample documentation (kubeflow#678)

* Add documentation for custom sample

* Add TARGET env var

* Add requirements.txt and update README

* kfservingsdk example

* Add expected output

* Update model name

docs/samples/custom/hello-world/Dockerfile

@@ -0,0 +1,17 @@
+# Use the official lightweight Python image.
+# https://hub.docker.com/_/python
+FROM python:3.7-slim
+
+# Copy local code to the container image.
+ENV APP_HOME /app
+WORKDIR $APP_HOME
+COPY app.py requirements.txt ./
+
+# Install production dependencies.
+RUN pip install --no-cache-dir -r ./requirements.txt
+
+# Run the web service on container startup. Here we use the gunicorn
+# webserver, with one worker process and 8 threads.
+# For environments with multiple CPU cores, increase the number of workers
+# to be equal to the cores available.
+CMD exec gunicorn --bind :$PORT --workers 1 --threads 8 app:app

docs/samples/custom/hello-world/README.md

@@ -0,0 +1,69 @@
+# Predict on a InferenceService using a Custom Image
+
+## Setup
+
+1. Your ~/.kube/config should point to a cluster with [KFServing installed](https://github.com/kubeflow/kfserving/blob/master/docs/DEVELOPER_GUIDE.md#deploy-kfserving).
+2. Your cluster's Istio Ingress gateway must be network accessible.
+
+## Build and push the sample Docker Image
+
+The goal of custom image support is to allow users to bring their own wrapped model inside a container and serve it with KFServing. Please note that you will need to ensure that your container is also running a web server e.g. Flask to expose your model endpoints.
+
+In this example we use Docker to build the sample python server into a container. To build and push with Docker Hub, run these commands replacing {username} with your Docker Hub username:
+
+```
+# Build the container on your local machine
+docker build -t {username}/custom-sample .
+
+# Push the container to docker registry
+docker push {username}/custom-sample
+```
+
+## Create the InferenceService
+
+In the `custom.yaml` file edit the container image and replace {username} with your Docker Hub username.
+
+Apply the CRD
+
+```
+kubectl apply -f custom.yaml
+```
+
+Expected Output
+
+```
+$ inferenceservice.serving.kubeflow.org/custom-sample created
+```
+
+## Run a prediction
+
+```
+MODEL_NAME=custom-sample
+CLUSTER_IP=$(kubectl -n istio-system get service istio-ingressgateway -o jsonpath='{.status.loadBalancer.ingress[0].ip}')
+SERVICE_HOSTNAME=$(kubectl get inferenceservice ${MODEL_NAME} -o jsonpath='{.status.url}' | cut -d "/" -f 3)
+
+curl -v -H "Host: ${SERVICE_HOSTNAME}" http://$CLUSTER_IP/v1/models/${MODEL_NAME}:predict
+```
+
+Expected Output
+
+```
+*   Trying 184.172.247.174...
+* TCP_NODELAY set
+* Connected to 184.172.247.174 (184.172.247.174) port 31380 (#0)
+> GET /v1/models/custom-sample:predict HTTP/1.1
+> Host: custom-sample.default.example.com
+> User-Agent: curl/7.64.1
+> Accept: */*
+>
+< HTTP/1.1 200 OK
+< content-length: 31
+< content-type: text/html; charset=utf-8
+< date: Thu, 13 Feb 2020 21:34:54 GMT
+< server: istio-envoy
+< x-envoy-upstream-service-time: 15
+<
+Hello Python KFServing Sample!
+* Connection #0 to host 184.172.247.174 left intact
+* Closing connection 0
+```

docs/samples/custom/hello-world/app.py

@@ -0,0 +1,15 @@
+import os
+
+from flask import Flask
+
+app = Flask(__name__)
+
+
+@app.route('/v1/models/custom-sample:predict')
+def hello_world():
+    greeting_target = os.environ.get('GREETING_TARGET', 'World')
+    return 'Hello {}!\n'.format(greeting_target)
+
+
+if __name__ == "__main__":
+    app.run(debug=True, host='0.0.0.0', port=int(os.environ.get('PORT', 8080)))

docs/samples/custom/custom.yaml → docs/samples/custom/hello-world/custom.yaml

@@ -3,13 +3,13 @@ kind: InferenceService
 metadata:
   labels:
     controller-tools.k8s.io: "1.0"
-  name: service-custom-sample
+  name: custom-sample
 spec:
   default:
     predictor:
       custom:
         container:
-          image: seldonio/mock_classifier:1.0
+          image: {username}/custom-sample
           env:
-            - name: PREDICTIVE_UNIT_SERVICE_PORT
-              value: "8080"
+            - name: GREETING_TARGET
+              value: "Python KFServing Sample"

docs/samples/custom/hello-world/requirements.txt

@@ -0,0 +1,2 @@
+Flask==1.1.1
+gunicorn==20.0.4

docs/samples/custom/kfserving-sdk/Dockerfile

@@ -0,0 +1,15 @@
+# Use the official lightweight Python image.
+# https://hub.docker.com/_/python
+FROM python:3.7-slim
+
+ENV APP_HOME /app
+WORKDIR $APP_HOME
+
+# Install production dependencies.
+COPY requirements.txt ./
+RUN pip install --no-cache-dir -r ./requirements.txt
+
+# Copy local code to container image
+COPY model.py  imagenet_classes.txt ./
+
+CMD ["python", "model.py"]

docs/samples/custom/kfserving-sdk/README.md

@@ -0,0 +1,73 @@
+# Predict on a InferenceService using a KFServing Model Server
+
+## Setup
+
+1. Your ~/.kube/config should point to a cluster with [KFServing installed](https://github.com/kubeflow/kfserving/blob/master/docs/DEVELOPER_GUIDE.md#deploy-kfserving).
+2. Your cluster's Istio Ingress gateway must be network accessible.
+
+## Build and push the sample Docker Image
+
+The goal of custom image support is to allow users to bring their own wrapped model inside a container and serve it with KFServing. Please note that you will need to ensure that your container is also running a web server e.g. Flask to expose your model endpoints.
+
+In this example we use Docker to build the sample python server into a container. To build and push with Docker Hub, run these commands replacing {username} with your Docker Hub username:
+
+```
+# Build the container on your local machine
+docker build -t {username}/kfserving-custom-model .
+
+# Push the container to docker registry
+docker push {username}/kfserving-custom-model
+```
+
+## Create the InferenceService
+
+In the `custom.yaml` file edit the container image and replace {username} with your Docker Hub username.
+
+Apply the CRD
+
+```
+kubectl apply -f custom.yaml
+```
+
+Expected Output
+
+```
+$ inferenceservice.serving.kubeflow.org/kfserving-custom-model created
+```
+
+## Run a prediction
+
+```
+MODEL_NAME=kfserving-custom-model
+INPUT_PATH=@./input.json
+CLUSTER_IP=$(kubectl -n istio-system get service istio-ingressgateway -o jsonpath='{.status.loadBalancer.ingress[0].ip}')
+SERVICE_HOSTNAME=$(kubectl get inferenceservice ${MODEL_NAME} -o jsonpath='{.status.url}' | cut -d "/" -f 3)
+
+curl -v -H "Host: ${SERVICE_HOSTNAME}" http://${CLUSTER_IP}/v1/models/${MODEL_NAME}:predict -d $INPUT_PATH
+```
+
+Expected Output:
+```
+*   Trying 169.47.250.204...
+* TCP_NODELAY set
+* Connected to 169.47.250.204 (169.47.250.204) port 80 (#0)
+> POST /v1/models/kfserving-custom-model:predict HTTP/1.1
+> Host: kfserving-custom-model.default.example.com
+> User-Agent: curl/7.64.1
+> Accept: */*
+> Content-Length: 105318
+> Content-Type: application/x-www-form-urlencoded
+> Expect: 100-continue
+>
+< HTTP/1.1 100 Continue
+* We are completely uploaded and fine
+< HTTP/1.1 200 OK
+< content-length: 232
+< content-type: text/html; charset=UTF-8
+< date: Fri, 21 Feb 2020 20:19:37 GMT
+< server: istio-envoy
+< x-envoy-upstream-service-time: 258
+<
+* Connection #0 to host 169.47.250.204 left intact
+{"predictions": {"Labrador retriever": 0.4158518612384796, "golden retriever": 0.1659165322780609, "Saluki, gazelle hound": 0.16286855936050415, "whippet": 0.028539149090647697, "Ibizan hound, Ibizan Podenco": 0.023924754932522774}}* Closing connection 0
+```

docs/samples/custom/kfserving-sdk/custom.yml

@@ -0,0 +1,12 @@
+apiVersion: serving.kubeflow.org/v1alpha2
+kind: InferenceService
+metadata:
+  labels:
+    controller-tools.k8s.io: "1.0"
+  name: kfserving-custom-model
+spec:
+  default:
+    predictor:
+      custom:
+        container:
+          image: {username}/kfserving-custom-model