Week 8 Microservices Tutorial
In this week’s lab, we will be learning about using Docker containers to develop microservices
Prerequisites
- Docker - for packaging microservices
- Maven
mvn
build system - an alternative to thegradle
build system forjava
Before you begin, Docker needs to be installed. For installation instructions, refer to the official Docker documentation You will build and run the microservices in Docker containers.
Introduction
This tutorial was inspired by Open Liberty’s guide
From development to production, and across your DevOps environments, you can deploy your microservices in a lightweight and portable manner by using containers. You can run a container from a container image. Each container image is a package of what you need to run your microservice or application, from the code to its dependencies and configuration. If you’re new to the development of applications in containers, you might want to start with the Using Docker containers to develop microservices guide before you work through this guide.
You’ll learn how to build container images and run containers using Docker for your microservices. You’ll learn about the Open Liberty container images and how to use them for your containerized applications. You’ll construct Dockerfile
files, create Docker images by using the docker build
command, and run the image as Docker containers by using docker run
command.
The two microservices that you’ll be working with are called system
and inventory
. The system
microservice returns the JVM system properties of the running container. The inventory
microservice adds the properties from the system
microservice to the inventory. This guide demonstrates how both microservices can run and communicate with each other in different Docker containers.
Getting started
Setting up the repository#
Clone the git repository for the example project
git clone https://gitlab.cecs.anu.edu.au/comp2120/2023/guide-containerize
cd guide-containerize
The start
directory contains the starting project that you will build upon.
The finish
directory contains the finished project that you will build.
Before you begin, make sure you have all the necessary prerequisites.
Packaging your microservices#
To begin, run the following command to navigate to the start directory:
cd start
You can find the starting Java project in the start
directory. This project is a multi-module Maven project that is made up of the system
and inventory
microservices. Each microservice is located in its own corresponding directory, system
and inventory
.
To try out the microservices by using Maven, run the following Maven goal to build the system
microservice and run it inside Open Liberty:
mvn -pl system liberty:run
Open another command-line session and run the following Maven goal to build the inventory
microservice and run it inside Open Liberty:
mvn -pl inventory liberty:run
After you see the following message in both command-line sessions, both of your services are ready:
To access the inventory
service, which displays the current contents of the inventory, see http://localhost:9081/inventory/systems
To access the system
service, which shows the system properties of the running JVM, see http://localhost:9081/system/properties
You can add the system properties of your localhost to the inventory
service at http://localhost:9081/inventory/systems/localhost.
After you are finished checking out the microservices, stop the Liberty instances by pressing CTRL+C
in the command-line sessions where you ran the system
and inventory
services. Alternatively, you can run the liberty:stop
goal in another command-line session:
mvn -pl system liberty:stop
mvn -pl inventory liberty:stop
To package your microservices, run the Maven package goal to build the application .war
files from the start directory so that the .war
files are in the system/target
and inventory/target
directories.
mvn package
To learn more about RESTful web services and how to build them, see Creating a RESTful web service for details about how to build the system
service. The inventory
service is built in a similar way.
Building your Docker images#
A Docker image is a binary file. It is made up of multiple layers and is used to run code in a Docker container. Images are built from instructions in Dockerfiles to create a containerized version of the application.
A Dockerfile
is a collection of instructions for building a Docker image that can then be run as a container. As each instruction is run in a Dockerfile
, a new Docker layer is created. These layers, which are known as intermediate images, are created when a change is made to your Docker image.
Every Dockerfile
begins with a parent or base image over which various commands are run. For example, you can start your image from scratch and run commands that download and install a Java runtime, or you can start from an image that already contains a Java installation.
Learn more about Docker on the official Docker page.
Creating your Dockerfiles#
You will be creating two Docker images to run the inventory
service and system
service. The first step is to create Dockerfiles for both services.
In this guide, you’re using an official image from the IBM Container Registry (ICR), icr.io/appcafe/open-liberty:full-java11-openj9-ubi
, as your parent image. This image is tagged with the word full
, meaning it includes all Liberty features. full
images are NOT recommended for development only because they significantly expand the image size with features that are not required by the application.
To minimize your image footprint in production, you can use one of the kernel-slim
images, such as icr.io/appcafe/open-liberty:kernel-slim-java11-openj9-ubi
. This image installs the basic Liberty runtime. You can then add all the necessary features for your application with the usage pattern that is detailed in the Open Liberty https://openliberty.io/docs/latest/container-images.html#build[container image documentation]. To use the default image that comes with the Open Liberty runtime, define the FROM
instruction as FROM icr.io/appcafe/open-liberty
. You can find all official images on the Open Liberty container image repository.
Create the Dockerfile
for the inventory service.
The FROM
instruction initializes a new build stage, which indicates the parent image of the built image. If you don’t need a parent image, then you can use FROM scratch
, which makes your image a base image.
It is also recommended to label your Docker images with the LABEL
command, as the label information can help you manage your images. For more information, see Best practices for writing Dockerfiles.
The COPY
instructions are structured as COPY
[--chown=<user>:<group>]
<source>
<destination>
. They copy local files into the specified destination within your Docker image. In this case, the inventory
Liberty configuration files that are located at src/main/liberty/config
are copied to the /config/
destination directory. The inventory
application WAR file inventory.war
, which was created from running mvn package
, is copied to the /config/apps
destination directory.
The COPY
instructions use the 1001
user ID and 0
group because the icr.io/appcafe/open-liberty:full-java11-openj9-ubi
image runs by default with the USER 1001
(non-root) user for security purposes. Otherwise, the files and directories that are copied over are owned by the root user.
Place the RUN configure.sh
command at the end to get a pre-warmed Docker image. It improves the startup time of running your Docker container.
The Dockerfile
for the system
service follows the same instructions as the inventory
service, except that some labels
are updated, and the system.war
archive is copied into /config/apps
.
Create the Dockerfile
for the system service. system/Dockerfile
Building your Docker image#
Now that your microservices are packaged and you have written your Dockerfiles, you will build your Docker images by using the docker build
command.
Run the following commands to build container images for your application:
docker build -t system:1.0-SNAPSHOT system/.
docker build -t inventory:1.0-SNAPSHOT inventory/.
The -t
flag in the docker build
command allows the Docker image to be labeled (tagged) in the name[:tag]
format. The tag for an image describes the specific image version. If the optional [:tag]
tag is not specified, the latest
tag is created by default.
To verify that the images are built, run the docker images
command to list all local Docker images:
docker images
Or, run the docker images
command with --filter
option to list your images:
docker images -f "label=org.opencontainers.image.authors=Your Name"
Your inventory
and system
images appear in the list of all Docker images:
REPOSITORY TAG IMAGE ID CREATED SIZE
inventory 1.0-SNAPSHOT 08fef024e986 4 minutes ago 1GB
system 1.0-SNAPSHOT 1dff6d0b4f31 5 minutes ago 977MB
Running your microservices in Docker containers#
Now that your two images are built, you will run your microservices in Docker containers:
docker run -d --name system -p 9080:9080 system:1.0-SNAPSHOT
docker run -d --name inventory -p 9081:9081 inventory:1.0-SNAPSHOT
The following table describes the flags in these commands:
Flag | Description |
-d | Runs the container in the background. |
–name | Specifies a name for the container. |
-p | Maps the host ports to the container ports. For example: -p <HOST_PORT>:<CONTAINER_PORT> |
Next, run the docker ps
command to verify that your containers are started:
docker ps
Make sure that your containers are running and show Up
as their status:
CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES
2b584282e0f5 inventory:1.0-SNAPSHOT "/opt/ol/helpers/run…" 2 seconds ago Up 1 second 9080/tcp, 9443/tcp, 0.0.0.0:9081->9081/tcp inventory
99a98313705f system:1.0-SNAPSHOT "/opt/ol/helpers/run…" 3 seconds ago Up 2 seconds 0.0.0.0:9080->9080/tcp, 9443/tcp system
If a problem occurs and your containers exit prematurely, the containers don’t appear in the container list that the docker ps
command displays. Instead, your containers appear with an Exited
status when they run the docker ps -a
command. Run the docker logs system
and docker logs inventory
commands to view the container logs for any potential problems. Run the docker stats system
and docker stats inventory
commands to display a live stream of usage statistics for your containers. You can also double-check that your Dockerfiles are correct. When you find the cause of the issues, remove the faulty containers with the docker rm system
and docker rm inventory
commands. Rebuild your images, and start the containers again.
To access the application, go to the http://localhost:9081/inventory/systems URL. An empty list is expected because no system properties are stored in the inventory yet.
Next, retrieve the system
container’s IP address by running the following:
docker inspect -f "{{.NetworkSettings.IPAddress }}" system
The command returns the system container IP address:
172.17.0.2
In this case, the IP address for the system
service is 172.17.0.2
. Take note of this IP address to construct the URL to view the system properties.
Go to the \http://localhost:9081/inventory/systems/pass:c[[system-ip-address]]
URL by replacing [system-ip-address]
with the IP address that you obtained earlier. You see a result in JSON format with the system properties of your local JVM. When you go to this URL, these system properties are automatically stored in the inventory. Go back to the http://localhost:9081/inventory/systems URL and you see a new entry for [system-ip-address]
.
Externalizing Liberty’s configuration#
As mentioned at the beginning of this guide, one of the advantages of using containers is that they are portable and can be moved and deployed efficiently across all of your DevOps environments. Configuration often changes across different environments, and by externalizing your Liberty’s configuration, you can simplify the development process.
Imagine a scenario where you are developing an Open Liberty application on port 9081
but to deploy it to production, it must be available on port 9091
. To manage this scenario, you can keep two different versions of the server.xml
file; one for production and one for development. However, trying to maintain two different versions of a file might lead to mistakes. A better solution would be to externalize the configuration of the port number and use the value of an environment variable that is stored in each environment.
In this example, you will use an environment variable to externally configure the HTTP port number of the inventory
service.
In the inventory/server.xml
file, the default.http.port
variable is declared and is used in the httpEndpoint
element to define the service endpoint. The default value of the default.http.port
variable is 9081
. However, this value is only used if no other value is specified. You can replace this value in the container by using the -e flag for the podman run command.
Run the following commands to stop and remove the inventory
container and rerun it with the default.http.port
environment variable set:
docker stop inventory
docker rm inventory
docker run -d --name inventory -e default.http.port=9091 -p 9091:9091 inventory:1.0-SNAPSHOT
The -e
flag can be used to create and set the values of environment variables in a Docker container. In this case, you are setting the default.http.port
environment variable to 9091
for the inventory
container.
Now, when the service is starting up, Open Liberty finds the default.http.port
environment variable and uses it to set the value of the default.http.port
variable to be used in the HTTP endpoint.
The inventory
service is now available on the new port number that you specified. You can see the contents of the inventory at the http://localhost:9091/inventory/systems URL. You can add your local system properties at \http://localhost:9091/inventory/systems/pass:c[[system-ip-address]]
by replacing [system-ip-address]
with the IP address that you obtained in the previous section. The system
service remains unchanged and is available at the http://localhost:9080/system/properties URL.
You can externalize the configuration of more than just the port numbers. To learn more about Open Liberty configuration, check out the Server Configuration Overview docs.
Optimizing the image size#
As mentioned previously, the parent image that is used in each Dockerfile
contains the full
tag, which includes all of the Liberty features. This parent image with the full
tag is recommended for development, but while deploying to production it is recommended to use a parent image with the kernel-slim
tag. The kernel-slim
tag provides a bare minimum Liberty runtime with the ability to add the features required by the application.
Replace the Dockerfile
for the inventory service. inventory/Dockerfile
Replace the parent image with icr.io/appcafe/open-liberty:kernel-slim-java11-openj9-ubi
at the top of your Dockerfile
. This image contains the kernel-slim
tag that is recommended when deploying to production.
Place RUN features.sh
command after the COPY
command that copies the local /config/
directory into the Docker
image. The features.sh
script adds the Liberty features that your application is required to operate.
Ensure that you repeat these instructions for the system
service.
Replace the Dockerfile
for the system service.
system/Dockerfile
Continue by running the following commands to stop and remove your current Docker
containers that are using the full
parent image:
docker stop inventory system
docker rm inventory system
Next, build your new Docker
images with the kernel-slim
parent image:
docker build -t system:1.0-SNAPSHOT system/.
docker build -t inventory:1.0-SNAPSHOT inventory/.
Verify that the images have been built by executing the following command to list all the local Docker
images:
docker images
Notice that the images for the inventory
and system
services now have a reduced image size.
REPOSITORY TAG IMAGE ID CREATED SIZE
inventory 1.0-SNAPSHOT d5a3d1b2c20e 4 minutes ago 682MB
system 1.0-SNAPSHOT 6346cf87eae0 5 minutes ago 694MB
After confirming that the images have been built, run the following commands to start the Docker
containers:
docker run -d --name system -p 9080:9080 system:1.0-SNAPSHOT
docker run -d --name inventory -p 9081:9081 inventory:1.0-SNAPSHOT
Once your Docker
containers are running, run the following command to see the list of required features installed by features.sh
:
docker exec -it inventory /opt/ol/wlp/bin/productInfo featureInfo
Your list of Liberty features should be similar to the following:
jndi-1.0
servlet-5.0
cdi-3.0
concurrent-2.0
jsonb-2.0
jsonp-2.0
mpConfig-3.0
restfulWS-3.0
restfulWSClient-3.0
To ensure that your containers are working properly, try accessing the system
service to show the system properties of the running JVM.
See http://localhost:9080/system/properties[http://localhost:9080/system/properties]
Next, replace [system-ip-address]
with the IP address that you obtained earlier and add your localhost system properties to the inventory
service by visiting:
\http://localhost:9081/inventory/systems/pass:c[[system-ip-address]]
Then, verify the addition of your localhost system properties to your inventory
service.
See http://localhost:9081/inventory/systems
Testing the microservices#
You can test your microservices manually by hitting the endpoints or with automated tests that check your running Docker containers.
Create the SystemEndpointIT
class. system/src/test/java/it/io/openliberty/guides/system/SystemEndpointIT.java
The testGetProperties()
method checks for a 200
response code from the system
service endpoint.
Create the InventoryEndpointIT
class.
inventory/src/test/java/it/io/openliberty/guides/inventory/InventoryEndpointIT.java
- The
testEmptyInventory()
method checks that theinventory
service has a total of 0 systems before anything is added to it. - The
testHostRegistration()
method checks that thesystem
service was added toinventory
properly. - The
testSystemPropertiesMatch()
checks that thesystem
properties match what was added into theinventory
service. - The
testUnknownHost()
method checks that an error is raised if an unknown host name is being added into theinventory
service. - The
systemServiceIp
variable has the same value as the IP address that you retrieved in the previous section when you manually added thesystem
service into theinventory
service. This value of the IP address is passed in when you run the tests.
Activity 3: Running the tests#
Run the Maven package
goal to compile the test classes. Run the Maven failsafe
goal to test the services that are running in the Docker containers by replacing the [system-ip-address]
with the IP address that you determined previously.
mvn package
mvn failsafe:integration-test -Dsystem.ip=[system-ip-address] -Dinventory.http.port=9081 -Dsystem.http.port=9080
If the tests pass, you see output similar to the following example:
-------------------------------------------------------
T E S T S
-------------------------------------------------------
Running it.io.openliberty.guides.system.SystemEndpointIT
Tests run: 1, Failures: 0, Errors: 0, Skipped: 0, Time elapsed: 0.653 s - in it.io.openliberty.guides.system.SystemEndpointIT
Results:
Tests run: 1, Failures: 0, Errors: 0, Skipped: 0
-------------------------------------------------------
T E S T S
-------------------------------------------------------
Running it.io.openliberty.guides.inventory.InventoryEndpointIT
Tests run: 4, Failures: 0, Errors: 0, Skipped: 0, Time elapsed: 0.935 s - in it.io.openliberty.guides.inventory.InventoryEndpointIT
Results:
Tests run: 4, Failures: 0, Errors: 0, Skipped: 0
When you are finished with the services, run the following commands to stop and remove your containers:
docker stop inventory system
docker rm inventory system
Great work! You have just built Docker images and run two microservices on Open Liberty in containers.