Build a Scala application image in 5 minutes, no Dockerfile required.
At the end of this tutorial, you'll have a working OCI image of a Scala application that can run locally. You will learn about the Cloud Native Buildpack (CNB) ecosystem, and how to utilize the pack CLI to build images without the need to write or maintain a Dockerfile.
You can now also use the
heroku/scala
CNB on Heroku Fir generation via the pilot program. See the Getting started on Heroku Fir Dev Center tutorial.
We assume you have docker installed and a working copy of git. Next, you will need to install the CLI tool for building CNBs, pack CLI. If you're on a Mac you can install it via Homebrew:
$ brew install buildpacks/tap/pack
Ensure that pack
is installed correctly:
$ pack --version
0.35.1+git-3a22a7f.build-6099
Once pack
is installed, the only configuration you'll need for this tutorial is to set a default builder:
$ pack config default-builder heroku/builder:24
Builder 'heroku/builder:24' is now the default builder
You can view your default builder at any time:
$ pack config default-builder
The current default builder is 'heroku/builder:24'
Note: The
heroku/builder:24
supports both amd64 (also known as x86) and arm64 (such as aarch64 used with newer Mac machines) architectures. If needed, you can configure the architecture fordocker
andpack
CLIs using the--platform
argument if needed. For example--platform linux/amd64
.
Note
Skip ahead if you want to build the application first and get into the details later. You won't need to know about builders for the rest of this tutorial.
In short, a builder is a delivery mechanism for buildpacks. A builder contains references to base images and individual buildpacks. A base image contains the operating system and system dependencies. Buildpacks are the components that will configure an image to run your application, that’s where the bulk of the logic lives and why the project is called “Cloud Native Buildpacks” and not “Cloud Native Builders.”
You can view the contents of a builder via the command pack builder inspect
. For example:
$ pack builder inspect heroku/builder:24 | grep Buildpacks: -m1 -A10
Buildpacks:
ID NAME VERSION HOMEPAGE
heroku/deb-packages Heroku .deb Packages 0.0.3 https://github.com/heroku/buildpacks-deb-packages
heroku/dotnet Heroku .NET 0.1.9 https://github.com/heroku/buildpacks-dotnet
heroku/go Heroku Go 0.5.0 https://github.com/heroku/buildpacks-go
heroku/gradle Heroku Gradle 6.0.4 https://github.com/heroku/buildpacks-jvm
heroku/java Heroku Java 6.0.4 https://github.com/heroku/buildpacks-jvm
heroku/jvm Heroku OpenJDK 6.0.4 https://github.com/heroku/buildpacks-jvm
heroku/maven Heroku Maven 6.0.4 https://github.com/heroku/buildpacks-jvm
heroku/nodejs Heroku Node.js 3.4.0 https://github.com/heroku/buildpacks-nodejs
heroku/nodejs-corepack Heroku Node.js Corepack 3.4.0 https://github.com/heroku/buildpacks-nodejs
Note
Your output version numbers may differ.
This output shows the various buildpacks that represent the different languages that are supported by this builder such as heroku/go
and heroku/nodejs-engine
.
How do you configure a CNB? Give them an application. While Dockerfile is procedural, buildpacks, are declarative. A buildpack will determine what your application needs to function by inspecting the code on disk.
For this example, we're using a pre-built Scala application. Download it now:
$ git clone https://github.com/heroku/scala-getting-started
$ cd scala-getting-started
Verify you're in the correct directory:
$ ls -A
.env
.g8
.git
.github
.gitignore
LICENSE
Procfile
Procfile.windows
README.md
app
app.json
build.sbt
build.sc
conf
project
public
system.properties
This tutorial was built using the following commit SHA:
$ git log --oneline | head -n1
b16773f Update readme and index with fir info (#14)
Now build an image named my-image-name
by executing the heroku builder against the application by running the
pack build
command:
$ pack build my-image-name --path .
===> ANALYZING
Image with name "my-image-name" not found
===> DETECTING
3 of 4 buildpacks participating
heroku/jvm 6.0.4
heroku/sbt 6.0.4
heroku/procfile 3.2.0
===> RESTORING
Skipping buildpack layer analysis
===> BUILDING
[Installing OpenJDK 17.0.13]
[Building Scala project]
Running: sbt compile stage
Downloading sbt launcher for 1.7.2:
From https://repo1.maven.org/maven2/org/scala-sbt/sbt-launch/1.7.2/sbt-launch-1.7.2.jar
To /layers/heroku_sbt/sbt-extras/launchers/1.7.2/sbt-launch.jar
Downloading sbt launcher 1.7.2 md5 hash:
From https://repo1.maven.org/maven2/org/scala-sbt/sbt-launch/1.7.2/sbt-launch-1.7.2.jar.md5
To /layers/heroku_sbt/sbt-extras/launchers/1.7.2/sbt-launch.jar.md5
/layers/heroku_sbt/sbt-extras/launchers/1.7.2/sbt-launch.jar: OK
Picked up JAVA_TOOL_OPTIONS: -Dfile.encoding=UTF-8
[info] [launcher] getting org.scala-sbt sbt 1.7.2 (this may take some time)...
[info] [launcher] getting Scala 2.12.16 (for sbt)...
[info] welcome to sbt 1.7.2 (Azul Systems, Inc. Java 17.0.13)
[info] loading global plugins from /layers/heroku_sbt/sbt-global/plugins
[info] compiling 1 Scala source to /layers/heroku_sbt/sbt-global/plugins/target/scala-2.12/sbt-1.0/classes ...
[info] Non-compiled module 'compiler-bridge_2.12' for Scala 2.12.16. Compiling...
[info] Compilation completed in 8.191s.
[info] done compiling
[info] loading settings for project workspace-build from plugins.sbt ...
[info] loading project definition from /workspace/project
[info] loading settings for project root from build.sbt ...
[info] __ __
[info] \ \ ____ / /____ _ __ __
[info] \ \ / __ \ / // __ `// / / /
[info] / / / /_/ // // /_/ // /_/ /
[info] /_/ / .___//_/ \__,_/ \__, /
[info] /_/ /____/
[info]
[info] Version 2.8.19 running Java 17.0.13
[info]
[info] Play is run entirely by the community. Please consider contributing and/or donating:
[info] https://www.playframework.com/sponsors
[info]
[info] Running Play on Java 17 is experimental. Tweaks are necessary:
[info] https://github.com/playframework/playframework/releases/2.8.15
[info]
[info] compiling 8 Scala sources and 1 Java source to /workspace/target/scala-2.13/classes ...
[info] Non-compiled module 'compiler-bridge_2.13' for Scala 2.13.10. Compiling...
[info] Compilation completed in 7.85s.
[info] done compiling
[success] Total time: 13 s, completed Dec 23, 2024, 12:34:57 AM
[info] Wrote /workspace/target/scala-2.13/scala-getting-started_2.13-1.0-SNAPSHOT.pom
[success] Total time: 1 s, completed Dec 23, 2024, 12:34:58 AM
## Procfile Buildpack
- Processes from `Procfile`
- web: `target/universal/stage/bin/scala-getting-started -Dhttp.port=${PORT}`
- Done (finished in < 0.1s)
===> EXPORTING
Adding layer 'heroku/jvm:openjdk'
Adding layer 'heroku/jvm:runtime'
Adding layer 'buildpacksio/lifecycle:launch.sbom'
Added 1/1 app layer(s)
Adding layer 'buildpacksio/lifecycle:launcher'
Adding layer 'buildpacksio/lifecycle:config'
Adding layer 'buildpacksio/lifecycle:process-types'
Adding label 'io.buildpacks.lifecycle.metadata'
Adding label 'io.buildpacks.build.metadata'
Adding label 'io.buildpacks.project.metadata'
Setting default process type 'web'
Saving my-image-name...
*** Images (0809177c056c):
my-image-name
Adding cache layer 'heroku/jvm:openjdk'
Adding cache layer 'heroku/sbt:coursier-home'
Adding cache layer 'heroku/sbt:ivy-home'
Adding cache layer 'heroku/sbt:sbt-boot'
Adding cache layer 'heroku/sbt:sbt-extras'
Successfully built image 'my-image-name'
Note
Your output may differ.
Verify that you see “Successfully built image my-image-name” at the end of the output. And verify that the image is present locally:
$ docker image ls --format "table {{.ID}}\t{{.Repository}}\t{{.Tag}}" | grep my-image-name
0809177c056c my-image-name latest
Note
Skip ahead if you want to run the application first and get into the details later.
When you run pack build
with a builder, each buildpack runs a detection script to determine if it should be eligible to build the application. In our case the heroku/scala
buildpack found a build.sbt
file. As a result, the buildpack has enough information to install Java dependencies. You can view a list of the buildpacks used in the build output:
===> DETECTING
3 of 4 buildpacks participating
heroku/jvm 6.0.4
heroku/sbt 6.0.4
heroku/procfile 3.2.0
===> RESTORING
After the detect phase, each buildpack will execute. Buildpacks can inspect your project, install files to disk, run commands, write environment variables, and more. You can see some examples of that in the output above. For example, the Java buildpack installs dependencies from the build.sbt
automatically:
Running: sbt compile stage
If you’re familiar with Dockerfile you might know that many commands in a Dockerfile will create a layer. Buildpacks also use layers, but the CNB buildpack API provides for fine grained control over what exactly is in these layers and how they’re composed. Unlike Dockerfile, all images produced by CNBs can be rebased. The CNB api also improves on many of the pitfalls outlined in the satirical article Write a Good Dockerfile in 19 'Easy' Steps.
Even though we used pack
and CNBs to build our image, it can be run with your favorite tools like any other OCI image. We will be using the docker
command line to run our image.
By default, images will be booted into a web server configuration. You can launch the app we just built by running:
$ docker run -it --rm --env PORT=5006 -p 5006:5006 my-image-name
Picked up JAVA_TOOL_OPTIONS: -XX:MaxRAMPercentage=80.0 -Dfile.encoding=UTF-8
2024-12-23 00:35:09 �[34mINFO �[0;39m �[36mplay.api.db.DefaultDBApi�[0;39m �[35m�[0;39m Database [default] initialized
2024-12-23 00:35:09 �[34mINFO �[0;39m �[36mplay.api.db.HikariCPConnectionPool�[0;39m �[35m�[0;39m Creating Pool for datasource 'default'
2024-12-23 00:35:10 �[34mINFO �[0;39m �[36mplay.api.Play�[0;39m �[35m�[0;39m Application started (Prod) (no global state)
2024-12-23 00:35:10 �[34mINFO �[0;39m �[36mplay.core.server.AkkaHttpServer�[0;39m �[35m�[0;39m Listening for HTTP on /[0:0:0:0:0:0:0:0]:5006
Now when you visit http://localhost:5006 you should see a working web application:
Don't forget to stop the docker container when you're done.
Here's a quick breakdown of that command we just ran:
docker run
Create and run a new container from an image.-it
Makes the container interactive and allocates a TTY.--rm
Automatically remove the container when it exits.--env PORT=5006
Creates an environment variable namedPORT
and sets it to5006
this is needed so the application inside the container knows what port to bind the web server.-p 5006:5006
Publishes a container's port(s) to the host. This is what allows requests from your machine to be received by the container.my-image-name
The name of the image you want to use for the application.
So far, we've downloaded an application via git and run a single command pack build
to generate an image, and then we can use that image as if it was generated via a Dockerfile via the docker run
command.
In addition to running the image as a web server, you can access the container's terminal interactively. In a new terminal window try running this command:
$ docker run -it --rm my-image-name bash
Now you can inspect the container interactively. For example, you can see the files on disk with ls
:
$ ls -A
.env
.g8
.git
.github
.gitignore
LICENSE
Procfile
Procfile.windows
README.md
app
app.json
build.sbt
build.sc
build_output.txt
conf
project
public
system.properties
target
And anything else you would typically do via an interactive container session.
Note
Skip this section if you want to try building your application with CNBs and learn about container structure later.
If you’re an advanced Dockerfile
user you might be interested in learning more about the internal structure of the image on disk. You can access the image disk interactively by using the bash
docker command above.
If you view the root directory /
you’ll see there is a layers
folder. Every buildpack that executes gets a unique folder:
$ docker run --rm my-image-name "ls /layers"
config
heroku_jvm
sbom
Individual buildpacks can compose multiple layers from their buildpack directory. For example you can see that java
binary is present within that buildpack layer directory:
$ docker run --rm my-image-name "which java"
/layers/heroku_jvm/openjdk/bin/java
OCI images are represented as sequential modifications to disk. By scoping buildpack disk modifications to their own directory, the CNB API guarantees that changes to a layer in one buildpack will not affect the contents of disk to another layer. This means that OCI images produced by CNBs are rebaseable by default, while those produced by Dockerfile are not.
We saw before how the image booted a web server by default. This is accomplished using an entrypoint. In another terminal outside of the running container you can view that entrypoint:
$ docker inspect my-image-name | grep '"Entrypoint": \[' -A2
"Entrypoint": [
"/cnb/process/web"
],
From within the image, you can see that file on disk:
$ docker run --rm my-image-name "ls /cnb/process/"
web
While you might not need this level of detail to build and run an application with Cloud Native Buildpacks, it is useful to understand how they’re structured if you ever want to write your own buildpack.
So far we've learned that CNBs are a declarative interface for producing OCI images (like docker). They aim to be no to low configuration and once built, you can interact with them like any other image.
For the next step, we encourage you to try running pack
with the Heroku builder against your application and let us know how it went. We encourage you to share your experience by opening a discussion and walking us through what happened:
- What went well?
- What could be better?
- Do you have any questions?
We are actively working on our Cloud Native Buildpacks and want to hear about your experience. The documentation below covers some intermediate-level topics that you might find helpful.
Most buildpacks rely on existing community standards to allow you to configure your application declaratively. They can also implement custom logic based on file contents on disk or environment variables present at build time.
The Procfile
is a configuration file format that was introduced by Heroku in 2011, you can now use this behavior on your CNB-powered application via the heroku/procfile
, which like the rest of the buildpacks in our builder is open source. The heroku/procfile
buildpack allows you to configure your web startup process.
This is the web
entry in the getting started guide's Procfile
:
web: target/universal/stage/bin/scala-getting-started -Dhttp.port=${PORT}
By including this file and using heroku/procfile
buildpack, your application will receive a default web process. You can configure this behavior by changing the contents of that file.