In this reference guide we’re going to describe various aspects of Quarkus configuration. A Quarkus application and Quarkus itself (core and extensions) are both configured via the same mechanism that leverages the SmallRye Config API an implementation of the MicroProfile Config specification.

If you’re looking for information how to make a Quarkus extension configurable then see the Writing Your Own Extension guide.

The final configuration is the aggregation of the properties defined by all these sources. A configuration property lookup starts by the highest ordinal configuration source available and works it way down to other sources until a match is found. This means that any configuration property may override a value just by setting a different value in a higher ordinal config source. For example, a property configured using an environment property overrides the value provided using the application.properties file.

1.1. System properties

System properties can be handed to the application through the -D flag during startup. The following examples assign the value youshallnotpass to the attribute quarkus.datasource.password .

For a runner jar: java -Dquarkus.datasource.password=youshallnotpass -jar target/quarkus-app/quarkus-run.jar

For a native executable: ./target/myapp-runner -Dquarkus.datasource.password=youshallnotpass

For a runner jar: export QUARKUS_DATASOURCE_PASSWORD=youshallnotpass ; java -jar target/quarkus-app/quarkus-run.jar

For a native executable: export QUARKUS_DATASOURCE_PASSWORD=youshallnotpass ; ./target/myapp-runner

Environment variables names follow the conversion rules specified by MicroProfile Config . Config searches three environment variables for a given property name (e.g. foo.BAR.baz ):

A property with double quotes foo."bar".baz , replace each character that is neither alphanumeric nor _ with _ : FOO__BAR__BAZ

A property with dashes foo.bar-baz , replace each character that is neither alphanumeric nor _ with _ : FOO_BAR_BAZ

An indexed property foo.bar[0] or foo.bar[0].baz , replace each character that is neither alphanumeric nor _ with _ : FOO_BAR_0_ or FOO_BAR_0__BAZ

In some situations, looking up the exact property name is impossible. This is the case for configuration names that contain user defined path segments.

Applying the conversion rules for Environment Variables names, quarkus.datasource."datasource-name".jdbc.url becomes QUARKUS_DATASOURCE__DATASOURCE_NAME__JDBC_URL . The configuration will work as expected if both properties are available in the Config system.

If only QUARKUS_DATASOURCE__DATASOURCE_NAME__JDBC_URL is present, the Config system needs to reconvert the configuration name to its most likely dotted format. This works fine for fixed configuration segments, but not for names that contain dynamic segments. In this case, Quarkus is unable to determine if DATASOURCE_NAME should be converted to datasource.name or datasource-name (or any other special character separator).

For this reason, such properties always require their dotted version name in another source (the value can be left empty) to disambiguate the Environment Variable name. It will provide additional information to perform a two-way conversion and match the property names together.

# value can be left empty
quarkus.datasource."datasource-name".jdbc.url=

For dev mode, this file can be placed in the root of the project, but it is advised to not check it in to version control because it typically contains passwords, access tokens, API keys or other secrets.

1.4. Quarkus Application configuration file

The Quarkus Application configuration file is loaded from the classpath resources, for instance src/main/resources/application.properties , src/test/resources/application.properties or from a jar dependency that contains an application.properties entry. Each application.properties found is treated as a separate ConfigSource and follow the same rules as every other source (override per property). Additionally, the configuration file may also reside in $PWD/config/application.properties . The loading starts from the config folder and then classpath order ( application.properties files in the application sources will have priority on the classloader loading order).

application.properties
greeting.message=hello (1)
quarkus.http.port=9090 (2)
The config/application.properties is also available in dev mode. The file needs to be placed inside the build tool output directory ( target for Maven and build/classes/java/main for Gradle). Keep in mind however that any cleaning operation from the build tool like mvn clean or gradle clean will remove the config directory as well.

1.5. MicroProfile Config configuration file

The MicroProfile Config configuration file in src/main/resources/META-INF/microprofile-config.properties .

microprofile-config.properties
greeting.message=hello (1)
quarkus.http.port=9090 (2)
It works in the exact same way as Quarkus Application configuration file application.properties . Recommendation is to use Quarkus application.properties .

1.6. Locations

Additionally to the default config locations, Quarkus provides a way to scan additional locations for configuration properties files.

The quarkus.config.locations configuration property accepts multiple locations separated by a comma , and each must represent a valid URI . The supported URI schemes are:

All loaded sources use the same ordinal of the source that found the quarkus.config.locations configuration property. For instance, if quarkus.config.locations is set as a system property, then all loaded sources have their ordinals set to 400 (system properties use 400 as their ordinal). The ordinal may be overridden directly for each config source by setting the config_ordinal property and the ordinal value. The config_ordinal property only affects the ordinal of the source in which is being set. Sources are sorted first by their ordinal, then by location order, and finally by loading order.

1.7. Additional Config Sources

Quarkus provides additional extensions which cover other configuration formats and stores:

You can use @Inject @ConfigProperty or just @ConfigProperty . The @Inject annotation is not necessary for members annotated with @ConfigProperty . If you do not provide a value for this property, the application startup fails with jakarta.enterprise.inject.spi.DeploymentException: No config value of type [class java.lang.String] exists for: greeting.message . The default value is injected if the configuration does not provide a value for greeting.suffix . This property is optional - an empty Optional is injected if the configuration does not provide a value for greeting.name .

2.1. Default Values

If a property is associated with a default value (by way of the defaultValue attribute), and no configuration value is supplied for the property, then rather than throwing a jakarta.enterprise.inject.spi.DeploymentException , the default value will be used. The defaultValue value is expressed as a String , and uses the same conversion mechanism used to process configuration values. Several Built-in Converters already exist for primitives, boxed primitives, and other classes; for example:

Optional containers: java.util.Optional , java.util.OptionalInt , java.util.OptionalLong , and java.util.OptionalDouble

Java enum types

JSR 310 java.time.Duration

JDK networking java.net.SocketAddress , java.net.InetAddress , etc.

As you might expect, these converters are org.eclipse.microprofile.config.spi.Converter implementations. Therefore, these converters comply with the Microprofile or custom implementation providers expression rules like:

Boolean values will be true in cases "true", "1", "YES", "Y" "ON". Otherwise, value will be interpreted as false

For float and double values the fractional digits must be separated by a dot .

Note that when a combination of Optional* types and the defaultValue attribute are used, the defined defaultValue will still be used and if no value is given for the property, the Optional* will be present and populated with the converted default value. However, when the property is explicitly empty, the default value is not used and the Optional will be empty. Consider this example:

# missing value, optional property
greeting.name=

In this case, since greeting.name was configured to be Optional* up above, the corresponding property value will be an empty Optional and execution will continue normally. This would be the case even if there was a default value configured: the default value is not used if the property is explicitly cleared in the configuration.

On the other hand, this example:

# missing value, non-optional
greeting.suffix=

will result in a java.util.NoSuchElementException: SRCFG02004: Required property greeting.message not found on startup and the default value will not be assigned.

Below is an example of a Quarkus-supplied converter:

@ConfigProperty(name = "server.address", defaultValue = "192.168.1.1")
InetAddress serverAddress;

The org.eclipse.microprofile.config.ConfigProvider.getConfig() API allows to access the Config API programmatically. This API is mostly useful in situations where CDI injection is not available.

String databaseName = ConfigProvider.getConfig().getValue("database.name", String.class);
Optional<String> maybeDatabaseName = ConfigProvider.getConfig().getOptionalValue("database.name", String.class);
Do not use System.getProperty(String) or System.getEnv(String) to retrieve configuration values. These APIs are not configuration aware and do not support the features described in this guide.

We often need to configure our application differently depending on the target environment . For example, the local development environment may be different from the production environment.

Configuration Profiles allow for multiple configurations in the same file or separate files and select between them via a profile name.

4.1. Profile in the property name

To be able to set properties with the same name, each property needs to be prefixed with a percentage sign % followed by the profile name and a dot . in the syntax %{profile-name}.config.name :

application.properties
quarkus.http.port=9090
%dev.quarkus.http.port=8181

The Quarkus HTTP port will be 9090. If the dev profile is active it will be 8181.

Profiles in the .env file follow the syntax _{PROFILE}_CONFIG_KEY=value :

QUARKUS_HTTP_PORT=9090
_DEV_QUARKUS_HTTP_PORT=8181

If a profile does not define a value for a specific attribute, the default (no profile) value is used:

application.properties
bar=”hello”
baz=”bonjour”
%dev.bar=”hallo”

With the dev profile enabled, the property bar has the value hallo , but the property baz has the value bonjour . If the prod profile is enabled, bar has the value hello (as there is no specific value for the prod profile), and baz the value bonjour .

4.2. Default Profiles

By default, Quarkus provides three profiles, that activate automatically in certain conditions:

4.3. Custom Profiles

It is also possible to create additional profiles and activate them with the quarkus.profile configuration property. A single config property with the new profile name is the only requirement:

application.properties
quarkus.http.port=9090
%staging.quarkus.http.port=9999

Setting quarkus.profile to staging will activate the staging profile.

4.4. Profile aware files

In this case, properties for a specific profile may reside in an application-{profile}.properties named file. The previous example may be expressed as:

application.properties
quarkus.http.port=9090
%staging.quarkus.http.test-port=9091
application-staging.properties
quarkus.http.port=9190
quarkus.http.test-port=9191

In this style, the configuration names in the profile aware file do not need to be prefixed with the profile name.

Properties in the profile aware file have priority over profile aware properties defined in the main file.

Do not use profile aware files to set quarkus.profile or quarkus.test.profile . This will not work because the profile is required in advance to load the profile aware files.

A profile aware file is only loaded if the unprofiled application.properties is also available in the same location and the file extension matches between the files. This is required to keep a consistent loading order and pair all the resources together.

4.5. Parent Profile

A Parent Profile adds one level of hierarchy to the current profile. The configuration quarkus.config.profile.parent accepts a single profile name.

When the Parent Profile is active, if a property cannot be found in the current active Profile, the config lookup fallbacks to the Parent Profile. Consider:

quarkus.profile=dev
quarkus.config.profile.parent=common
%common.quarkus.http.port=9090
%dev.quarkus.http.ssl-port=9443
quarkus.http.port=8080
quarkus.http.ssl-port=8443

Do not use Profile aware files to set quarkus.config.profile.parent . This will not work because the profile is required in advance to load the profile aware files.

4.6. Multiple Profiles

Multiple Profiles may be active at the same time. The configuration quarkus.profile accepts a comma-separated list of profile names: quarkus.profile=common,dev . Both common and dev are separate profiles.

When multiple profiles are active, the rules for profile configuration are the same. If two profiles define the same configuration, then the last listed profile has priority. Consider:

application.properties
quarkus.profile=common,dev
my.prop=1234
%common.my.prop=1234
%dev.my.prop=5678
%common.commom.prop=common
%dev.dev.prop=dev
%test.test.prop=test

It is also possible to define multiple profile properties, with a comma-separated list of profile names. If the same property name exists in multiple profile properties then, the property name with the most specific profile wins:

application.properties
quarkus.profile=dev
%prod,dev.my.prop=1234
%dev.my.prop=5678
%prod,dev.another.prop=1234

Multiple profiles priority work in reverse order. With quarkus.profile=common,dev , Quarkus first checks the dev profile and then the common profile.

Quarkus provides property expressions expansion on configuration values. An expression string is a mix of plain strings and expression segments, which are wrapped by the sequence ${ …​ } .

These expressions are resolved when the property is read. So if the configuration property is build time the property expression will be resolved at build time. If the configuration property is overridable at runtime it will be resolved at runtime.

Consider:

application.properties
remote.host=quarkus.io
callable.url=https://${remote.host}/

The resolved value of the callable.url property is https://quarkus.io/ .

Another example would be defining different database servers by profile:

application.properties
%dev.quarkus.datasource.jdbc.url=jdbc:mysql://localhost:3306/mydatabase?useSSL=false
quarkus.datasource.jdbc.url=jdbc:mysql://remotehost:3306/mydatabase?useSSL=false

can be simplified to:

application.properties
%dev.application.server=localhost
application.server=remotehost
quarkus.datasource.jdbc.url=jdbc:mysql://${application.server}:3306/mydatabase?useSSL=false

Additionally, the Expression Expansion engine supports the following segments:

${expression:value} - Provides a default value after the : if the expansion doesn’t find a value.

${my.prop${compose}} - Composed expressions. Inner expressions are resolved first.

${my.prop}${my.prop} - Multiple expressions.

Expressions lookups are performed in all config sources. The expression values and expansion values may reside in different config sources.

5.1. With Environment Variables

Property Expressions also work with Environment Variables.

application.properties
remote.host=quarkus.io
application.host=${HOST:${remote.host}}

This will expand the HOST environment variable and use the value of the property remote.host as the default value if HOST is not set.

A secret configuration may be expressed as ${handler::value} , where the handler is the name of a io.smallrye.config.SecretKeysHandler to decode or decrypt the value . Consider:

application.properties
my.secret=${aes-gcm-nopadding::DJNrZ6LfpupFv6QbXyXhvzD8eVDnDa_kTliQBpuzTobDZxlg}
# the encryption key required to decode the secret. It can be set in any source.
smallrye.config.secret-handler.aes-gcm-nopadding.encryption-key=somearbitrarycrazystringthatdoesnotmatter

A lookup to my.secret will use the SecretKeysHandler name aes-gcm-nopadding to decode the value DJNrZ6LfpupFv6QbXyXhvzD8eVDnDa_kTliQBpuzTobDZxlg .

For more information, please check SmallRye Config Secret Keys documentation.

SmallRye Config may provide handlers not fully supported by Quarkus. Currently, only smallrye-config-crypto is supported.

The default config source from Quarkus provides a random UUID value. It generates the UUID at startup time. So, the value changes between startups, including reloads in dev mode.

You can access the generated value using the quarkus.uuid property. Use expressions to access it: ${quarkus.uuid} . For example, it can be useful to configure a Kafka client with a unique consumer group:

mp.messaging.incoming.prices.group.id=${quarkus.uuid}

Run time properties which are optional, and which have had values set at build time or which have a default value, may be explicitly cleared by assigning an empty string to the property. Note that this will only affect runtime properties, and will only work with properties whose values are not required.

application.properties
remote.host=quarkus.io

A lookup to remote.host with -Dremote.host= will throw an Exception, because the system property cleared the value.

A config value which contains unescaped commas may be converted to Collection . This works for simple cases, but it becomes cumbersome and limited for more advanced cases.

Indexed Properties provide a way to use indexes in config property names to map specific elements in a Collection type. Since the indexed element is part of the property name and not contained in the value, this can also be used to map complex object types as Collection elements. Consider:

application.properties
my.collection=dog,cat,turtle
my.indexed.collection[0]=dog
my.indexed.collection[1]=cat
my.indexed.collection[2]=turtle

The indexed property syntax uses the property name and square brackets `[ ] with an index in between.

A call to Config#getValues("my.collection", String.class) , will automatically create and convert a List<String> that contains the values dog , cat and turtle . A call to Config#getValues("my.indexed.collection", String.class) returns the exact same result. If the same property name exists in both forms (regular and indexed), the regular value has priority.

The indexed property is sorted by their index before being added to the target Collection . Any gaps contained in the indexes do not resolve to the target Collection , which means that the Collection result will store all values without any gaps.

Quarkus itself is configured via the same mechanism as your application. Quarkus reserves the quarkus. namespace for its own configuration. For example to configure the HTTP server port you can set quarkus.http.port in application.properties . All the Quarkus configuration properties are documented and searchable .

As mentioned above, properties prefixed with quarkus. are effectively reserved for configuring Quarkus itself and its extensions. Therefore, the quarkus. prefix should never be used for application specific properties.

10.1. Build Time configuration

Some Quarkus configurations only take effect during build time, meaning it is not possible to change them at runtime. These configurations are still available at runtime but as read-only and have no effect in Quarkus behaviour. A change to any of these configurations requires a rebuild of the application itself to reflect changes of such properties.

However, some extensions do define properties overridable at runtime . A simple example is the database URL, username and password which is only known specifically in your target environment, so they can be set and influence the application behaviour at runtime.

Given that configuration sources usually provide more options than actually used during the build, it might be useful to know which configuration options have actually been used during a Quarkus build process.

12.1. Dumping build time configuration options read during the build

Setting quarkus.config-tracking.enabled to true will enable a configuration interceptor that will record every configuration option that was read during the build process along with their values. The resulting report will be stored in ${project.basedir}/.quarkus/quarkus-prod-config-dump by default. The target file could be configured using the following options:

quarkus.config-tracking.directory - directory in which the configuration dump should be stored, the default is ${project.basedir}/.quarkus

quarkus.config-tracking.file-prefix - file name prefix, the default value is quarkus

quarkus.config-tracking.file-suffix - file name suffix, the default value is -config-dump

quarkus.config-tracking.file - path to a file in which the configuration dump should be stored. This option supersedes the file-prefix and file-suffix options. Also supersedes the value of quarkus.config-tracking.directory , unless the value is a relative path.

The prod part of the quarkus-prod-config-dump file name refers to the Quarkus build mode, indicating that the dump was taken for the production build.

The reason ${project.basedir}/.quarkus directory was chosen as the default location was to make it easy to track build time configuration changes between builds and use that as an indicator to build output caching tools (such as Apache Maven Build Cache and Develocity Build Cache ) whether the application binary has to be re-built.

12.1.1. Filtering configuration options

Configuration tracker could be instructed to exclude some of the options from the report by configuring quarkus.config-tracking.exclude with a comma-separated list of configuration option names that should be filtered out.

12.1.2. Path values

Configuration options with absolute path values that begin with a user home directory are, by default, transformed with Unix home directory alias '~' replacing the user home directory part and using / as a path element separator.

This transformation can be disabled by setting quarkus.config-tracking.use-user-home-alias-in-paths to false .

12.1.3. Hashing recorded configuration values

Configuration values can be hashed using SHA-512 algorithm before they are written to a file. Configuration option names whose values should be hashed can be configured in quarkus.config-tracking.hash-options as a comma separated list.

12.2. Tracking build time configuration changes between builds

While quarkus.config-tracking.enabled enables effective build time configuration report generation, there is also a way to check whether the values stored in that report have changed before the next build of the project is launched.

Maven projects could add the following goal to their quarkus-maven-plugin configuration:

      <plugin>
        <groupId>${quarkus.platform.group-id}</groupId>
        <artifactId>quarkus-maven-plugin</artifactId>
        <version>${quarkus.platform.version}</version>
        <extensions>true</extensions>
        <executions>
          <execution>
            <id>track-prod-config-changes</id>
            <phase>process-resources</phase>
            <goals>
              <goal>track-config-changes</goal>
            </goals>
          </execution>
          <!-- other executions would follow below -->

The track-config-changes goal looks for ${project.basedir}/.quarkus/quarkus-prod-config-dump (file name and directory are configurable) and, if the file already exists, checks whether the values stored in the config dump have changed. It will log the changed options and save the current values of each of the options present in ${project.basedir}/.quarkus/quarkus-prod-config-dump in ${project.basedir}/target/quarkus-prod-config.check (the target file name and location can be configured). If the build time configuration has not changed since the last build both ${project.basedir}/.quarkus/quarkus-prod-config-dump and ${project.basedir}/.quarkus/quarkus-prod-config-dump will be identical.

12.2.1. Dump Quarkus application dependencies

In addition to dumping configuration values, track-config-changes goal also dumps all the Quarkus application dependencies, including Quarkus build time dependencies. This file could be used to check whether Quarkus build classpath has changed since the previous run, for instance together with Develocity’s ability to checksum a classpath. By default, the list of dependencies will be stored under target/quarkus-prod-dependencies.txt file. A different location could be configured using plugin parameters.

12.2.2. Dump current build configuration when the recorded configuration isn’t found

By default, track-config-changes looks for the configuration recorded during previous build and does nothing if it’s not found. Enabling dumpCurrentWhenRecordedUnavailable parameter will make it dump the current build configuration options taking into account quarkus.config-tracking.* configuration.

Unlike the build configuration options recorded during the quarkus:build goal, configuration options saved by quarkus:track-config-changes with dumpCurrentWhenRecordedUnavailable enabled will include all the build configuration options exposed by a org.eclipse.microprofile.config.Config instance. Which means this report may include some build configuration options that will not be used by the Quarkus application build process but also may be missing some build configuration options since MicroProfile Config specification allows configuration sources not to expose all the property names they provide to users.

Quarkus collects the config property values injected in CDI beans during static initialization phase . The collected values are then compared with their runtime initialization counterparts and if a mismatch is detected the application startup fails. How can it happen? For example, let’s have a CDI bean org.acme.MyBean . MyBean injects a @ConfigProperty of name foo and is initialized during the native build. The config property does not exist during the native build and so the default value bar is used. But later, when the application is started the property is defined with a system property: -Dfoo=baz . This would lead to inconsistent state and unexpected behavior. Therefore, Quarkus would fail in this situation by default.

package org.acme;
import jakarta.enterprise.context.ApplicationScoped;
import jakarta.enterprise.event.Observes;
import jakarta.enterprise.context.Initialized;
import org.eclipse.microprofile.config.inject.ConfigProperty;
@ApplicationScoped
public class MyBean {
   @ConfigProperty(name = "foo", defaultValue = "bar") (1)
   String foo;
    void onInit(@Observes @Initialized(ApplicationScoped.class) Object event) { (2)
       // this observer method is notified during STATIC_INIT...
import jakarta.enterprise.context.ApplicationScoped;
import jakarta.enterprise.event.Observes;
import jakarta.enterprise.context.Initialized;
import org.eclipse.microprofile.config.inject.ConfigProperty;
import io.quarkus.runtime.annotations.StaticInitSafe;
@ApplicationScoped
public class MyBeanNoFailure {
   @StaticInitSafe (1)
   @ConfigProperty(name = "foo", defaultValue = "bar")
   String foo;
    void onInit(@Observes @Initialized(ApplicationScoped.class) Object event) {
       // this observer method is notified during STATIC_INIT...

Accepts a single configuration profile name. If a configuration property cannot be found in the current active profile, the config performs the same lookup in the profile set by this configuration.

Environment variable: QUARKUS_CONFIG_PROFILE_PARENT