Use the "pio instance" command.

docs/manual/engines/index.md

@@ -5,25 +5,59 @@ title: Engines
 
 # Engines
 
-Each engine represents a type of prediction. 
+Each engine represents a type of prediction.
 
 ## Getting Started with Engine Instance
 
-Let say you want to launch an instance of Item Recommendation Engine for product recommendation.
+Let say you want to launch an instance of Item Recommendation Engine for product
+recommendation. First of all, you will need to create an engine instance
+project. The new project will contain configuration for your engine instance.
+
+```
+$ $PIO_HOME/bin/pio instance io.prediction.engines.itemrec
+$ cd io.prediction.engines.itemrec
+```
+where `$PIO_HOME` is your installation path of PredictionIO.
 
-### Specific which App the instance is targeted to
+
+### Specific the Target App
+
+Inside the engine instance project, edit `params/datasource.json` and change the
+value of `appId` to fit your app.
 
 ```
-(TODO)
+{
+ "appId": 1,
+ "actions": [
+ "view",
+ "like",
+ "conversion",
+ ...
 ```
 
 ### Launching an Engine Instance
+
 ```
-$ cd $PIO_HOME/engines/item-rec
-$ ../../bin/pio train
-$ ../../bin/pio <TODO>
+$ $PIO_HOME/bin/pio train
+...
+2014-09-11 16:25:44,591 INFO spark.SparkContext - Job finished: collect at Workflow.scala:674, took 0.078664 s
+2014-09-11 16:25:44,737 INFO workflow.CoreWorkflow$ - Saved engine instance with ID: KxOsC2FRSdGGe1lv0oaHiw
 ```
-You may launch more than one engine instance in PredictionIO.
+where `$PIO_HOME` is your installation path of PredictionIO.
+
+If your training was successful, you should see the lines shown above. Now you are ready to deploy the instance:
+
+```
+$ $PIO_HOME/bin/pio deploy
+...
+[INFO] [09/11/2014 16:26:16.525] [pio-server-akka.actor.default-dispatcher-2] [akka:https://pio-server/user/IO-HTTP/listener-0] Bound to localhost/127.0.0.1:8000
+[INFO] [09/11/2014 16:26:16.526] [pio-server-akka.actor.default-dispatcher-5] [akka:https://pio-server/user/master] Bind successful. Ready to serve.
+```
+
+> The `deploy` command runs the engine instance in the foreground. To run more
+than one engine instance, either launch a new console, or put the process into
+the background, then repeat the same command on a different port (by adding a
+`--port` argument).
 
 If it is your first time using PredictionIO, these [tutorials and samples](/tutorials/engines/) should be helpful.
 
@@ -35,13 +69,17 @@ You may set up a crontab in Linux to update the predictive model with new data r
 ```
 $ crontab -e
 
-0 */6 * * * cd <PIO_HOME>/engines/item-rec; ../../bin/pio train; ../../bin/pio deploy 
+0 */6 * * * cd <engine instance project directory>; $PIO_HOME/bin/pio train; $PIO_HOME/bin/pio deploy
 ```
-where `<PIO_HOME>` is your installation path of PredictionIO.
+where `$PIO_HOME` is your installation path of PredictionIO.
+
+> The `deploy` command will automatically undeploy any running engine instance
+on the same IP address and port. It is not necessary to use the `undeploy`
+command.
 
 ## Built-in Engines
 
-PredictionIO comes with the following engines. 
+PredictionIO comes with the following engines.
 
 * [Item Ranking](/engines/itemrank)
 * [Item Recommendation](/engines/itemrec)
@@ -51,4 +89,4 @@ You may start with these [tutorials and samples](tutorials.html).
 
 ## Building your own Engine
 
-Please read the [Engine Builders' Guide](/enginebuilders/) for details.
+Please read the [Engine Builders' Guide](/enginebuilders/) for details.

docs/manual/tutorials/engines/quickstart.md

@@ -164,10 +164,10 @@ can run multiple recommendation *engine instances* at the same time with
 different settings, or even for different projects.
 
 To launch an engine instance for this simple app, first create an engine
-project:
+instance project:
 
 ```
-$ $PIO_HOME/bin/pio new io.prediction.engines.itemrec
+$ $PIO_HOME/bin/pio instance io.prediction.engines.itemrec
 $ cd io.prediction.engines.itemrec
 ```