5. Java RMI

JAVA RMI

This uses the calculator example for RMI that can be found on moodle here.

The code can be found here in original directory. A solution to this task can be found in the solution directory.

Below is an example architecture of 2 rmi clients in separate docker containers and a server container that also runs an rmi registry:

   ---------------------------------------------------------------- rmicw network
         |             |                 |
         |             |                 |port 1099
         |             |                 |
       +----------+  +----------+      +---------------+
       |rmi       |  |rmi       |      |rmi server     |
       |client 1  |  |client 2  |      |and registry   |
       +----------+  +----------+      +---------------+
    +-----------------------------------------------------+
    |                                                     |
    |                       docker                        |
    +-----------------------------------------------------+

The rmicw is a network that lets all containers communicate. It is good practice to have separate networks for services to communicate over. To create the network, run:

docker network create rmicw

Lets Begin

Firstly, separate the code into 3 catagories:

• Code only the client needs
  • into a folder called client-code
• Code only the server needs
  • into a folder called server-code
• Code both the client and server need (like interfaces)
  • into a folder called shared-code

Dockerfile for the Client

In the client-code directory, make a new file called Dockerfile. For the client, we can start off with alpine linux as it is a nice small base image. We will also need to install a JDK such as openjdk8. This looks like so:

FROM alpine:latest
RUN apk add --no-cache openjdk8
ENV JAVA_HOME=/usr/lib/jvm/java-1.8-openjdk
ENV PATH="$JAVA_HOME/bin:${PATH}"

Here the Java is also "installed" (added to the PATH).

Next, the files for both the shared code and client code need to be copied into the same folder in the container.

WORKDIR /app
COPY client-code/ .
COPY shared-code/ .

Now both sets of code are in the /app directory in the container image.

Finally, we need to compile the code and tell the container to run the code when instantiated!

RUN javac *.java
ENTRYPOINT [ "java", "calculatorclient" ]

Now the docker container for a client can be created. This should be run from the projects root directory (where you can see the separated folders created in the lets begin step):

docker build --rm -f client-code/Dockerfile -t 311tutorial:rmi-client .

Using -f here tells docker to use the dockerfile from the client code directory as it typically defaults to ./Dockerfile.

Dockerfile for the Server

In the server-code directory, make a new Dockerfile. Much like for the client, it can start off with alpine with a Java install.

FROM alpine:latest
RUN apk add --no-cache openjdk8
ENV JAVA_HOME=/usr/lib/jvm/java-1.8-openjdk
ENV PATH="$JAVA_HOME/bin:${PATH}"

Next, the files for both the shared code and server code need to be copied into the same folder in the container.

WORKDIR /app
COPY server-code/ .
COPY shared-code/ .

Now both sets of code are in the /app directory in the container image.

To be able to run both the registry and the code when the container starts, a simple bash script can be created like so:

#!/bin/bash

rmiregistry &
java calculatorserver

Call this file entrypoint.sh and put it in the server-code folder, and make it executable in the container:

RUN chmod +x entrypoint.sh

Finally, we need to compile the code and tell the container to run the entrypoint script at the start:

RUN javac *.java
ENTRYPOINT [ "sh", "entrypoint.sh" ]

Now the docker container for a server can be created. This should be run from the projects root directory (where you can see the separated folders created in the lets begin step):

docker build --rm -f server-code/Dockerfile -t 311tutorial:rmi-server .

Running your server and registry

• Using the --network flag, we can attach a container to a network
• using the --name flag, a hostname for the container can be set

docker run --rm --network rmicw --name rmiserver 311tutorial:rmi-server

Running your client

From another terminal window...
Using the --network rmicw flag, we can attach a container to a network:

docker run --rm --network rmicw 311tutorial:rmi-client

Oh No

The client is performing a registry lookup on localhost, but the registry is no longer at localhost! As the --name flag gave the server container the hostname rmiserver, the naming lookup must be performed there, so replacing localhost with rmiserver should fix that (in calculatorclient.java line 22).

You could have the host name set via a environment variable though, right...?

This code might help with that:

import java.util.stream.*;
import java.nio.file.*;
import java.io.*;

public class dockerutility {

  public static String getRegistryHost() {

    boolean isDocker = false;

    try (Stream < String > stream =
      Files.lines(Paths.get("/proc/1/cgroup"))) {
      isDocker = stream.anyMatch(line -> line.contains("/docker"));
    } catch (IOException e) {
      System.out.println("Not running in Docker 😢");
    }
    if (isDocker) {
      System.out.println("Running in Docker! 🐳");
      if (System.getenv("REGISTRY_HOST") != null)
        return System.getenv("REGISTRY_HOST");
      System.out.println("🐳⚠️  Please set the REGISTRY_HOST environment variable!");
    }
    return "localhost";
  }

}

The just run the client like so:

docker run --rm --network rmicw -e REGISTRY_HOST=|registry-container-hostname| 311tutorial:rmi-client