Last updated May 31, 2023

This document describes the general behavior of Heroku as it relates to the recognition and execution of Scala applications. For a more detailed explanation of how to deploy an application, see Getting Started with Scala on Heroku.

Activation

Heroku recognizes Scala applications when files matching any of the following patterns are found:

• /*.sbt
• /project/*.scala
• /project/build.properties
• /.sbt/*.scala

When a deployed application is recognized as a Scala application, Heroku responds with -----> Scala app detected.

$ git push heroku master
-----> Scala app detected

If an application also contains a /conf/application.conf file, a Play 2.0 application is detected and Heroku responds with -----> Play 2.0 app detected. See the Heroku Play Support article for more information.

You can deploy Scala applications that use Maven as well, but they’re treated as Java applications, so Heroku Java Support applies.

Environment

The following environment variables are set:

• PATH: .sbt_home/bin:/usr/local/bin:/usr/bin:/bin
• PORT: The web process binds to this HTTP port
• DATABASE_URL: URL of the Heroku Postgres database connection

When a Java process is started on your dyno, the following Java options are automatically picked up:

• -Dfile.encoding=UTF-8

These options are configured as part of the environment variable JAVA_TOOL_OPTIONS. It’s intended to augment a command line in environments where the command line can’t be accessed or modified. If you must override these settings, you can either define your preferred options in the Procfile command, which takes precedence, or you can set your own JAVA_TOOL_OPTIONS config var.

Adjusting Environment for a Dyno Size

When a new dyno size is selected, the following settings are automatically added to JAVA_TOOL_OPTIONS:

• eco, basic, or standard-1x: -Xmx350m -Xss512k
• standard-2x: -Xmx768m
• performance-m: -Xmx2g
• performance-l: -Xmx12g

Monitoring Resource Usage

You can use additional JVM flags to monitor resource usage in a dyno. The following flags are recommended for monitoring resource usage:

-XX:+PrintGCDetails -XX:+PrintGCDateStamps -XX:+PrintTenuringDistribution -XX:+UseConcMarkSweepGC

See the troubleshooting article for more information about tuning a JVM process.

Build Behavior

Applications must include a /project/build.properties file with the sbt.version property specifying a version of SBT between 0.11.0 and 1.x. SBT release candidates, betas, and other pre-release versions are allowed.

The Heroku Scala buildpack runs sbt compile stage to build the application. Applications must include a stage task, which performs any tasks needed to prepare an application to be run in-place. For example, Typesafe’s sbt-native-packager adds a stage task to SBT that generates start scripts for an application. To use the plugin, create a /project/plugins.sbt file containing:

addSbtPlugin("com.typesafe.sbt" % "sbt-native-packager" % "1.7.6")

Or you can write a custom stage task by putting something like this in your build.sbt:

val stage = taskKey[Unit]("Stage task")

val Stage = config("stage")

stage := {
  (packageWar in Compile).value
  (update in Stage).value.allFiles.foreach { f =>
    if (f.getName.matches("webapp-runner-[0-9\\.]+.jar")) {
      println("copying " + f.getName)
      IO.copyFile(f, baseDirectory.value / "target" / "webapp-runner.jar")
    }
  }
}

You can test your task by running sbt compile stage locally.

Clean Builds

In some cases, builds must clean artifacts before compiling. If a clean build is necessary, configure builds to perform clean by setting SBT_CLEAN=true:

$ heroku config:set SBT_CLEAN=true
Setting config vars and restarting example-app... done, v17
SBT_CLEAN: true

All subsequent deploys use the clean task. To remove the clean task, unset SBT_CLEAN:

$ heroku config:unset SBT_CLEAN
Unsetting SBT_CLEAN and restarting example-app... done, v18

Runtime Behavior

By default, Scala applications are launched with a start script generated by the sbt-native-packager:

web: target/universal/stage/bin/appname

If an application isn’t using sbt-native-packager or must be launched in a different way, you can include a custom Procfile in the root of the project specifying a different entry for the web process.

The Play framework automatically generates a start script for you, so no additional plugins are needed.

Supported JDK Versions

Heroku currently uses OpenJDK 8 to run your application. OpenJDK version 7 is available. Depending on what major version you select, the latest available update of that Java runtime is used each time you deploy your app. For a list of current versions, see the Java support article.

The JDK that your app uses is included in the slug, which affects your slug size.

Specifying a Java Version

You can specify a Java version by adding a file called system.properties to your application.

Set a property java.runtime.version in the file:

java.runtime.version=1.8

1.8 is the default version. To use Java 8, you don’t need this file at all. You can specify JDK 7 by setting this property to “1.7”. For details on how to set specific update versions, see the Java support article.

Postgres Auto-Provisioning

A mini Heroku Postgres database automatically provisions on the first deploy of your Scala applications. This auto-provisioning populates the DATABASE_URL environment variable.

If you don’t want the Postgres add-on, remove it by running:

$ heroku addons:destroy DATABASE --app example-app