This {section} explains the configuration methods for {project_name} and how to start and apply the preferred configuration. It includes configuration guidelines for optimizing {project_name} for faster startup and low memory footprint.
When an option is set in more than one source, the one that comes first in the list determines the value for that option. For example, the value for an option set by a command-line parameter has a higher priority than an environment variable for the same option.
If `--db-url=cliValue` had not been used, the applied value would be `KC_DB_URL=envVarValue`. If the value were not applied by either the command line or an environment variable, `db-url=confFileValue` would be used. If none of the previous values were applied, the value `kc.db-url=keystoreValue` would be used due to the lowest priority among the available configuration sources.
The configuration uses a _unified-per-source_ format, which simplifies translation of a key/value pair from one configuration source to another. Note that these formats apply to spi options as well.
Command-line parameter format:: Values for the command-line use the `--_<key-with-dashes>_=_<value>_` format. For some values, an `-_<abbreviation>_=_<value>_` shorthand also exists.
KeyStore configuration file format:: Values that go into the KeyStore configuration file use the `kc._<key-with-dashes>_` format. `_<value>_` is then a password stored in the KeyStore.
Alternatively, see <@links.server id="all-config"/> for all server options.
=== Formats for environment variables
You can use placeholders to resolve an environment specific value from environment variables inside the `keycloak.conf` file by using the `${r"++${ENV_VAR}++"}` syntax:
By default, the server always fetches configuration options from the `conf/keycloak.conf` file. For a new installation, this file holds only commented settings as an idea of what you want to set when running in production.
=== Setting sensitive options using a Java KeyStore file
Thanks to Keystore Configuration Source you can directly load properties from a Java KeyStore using the `[--config-keystore]` and `[--config-keystore-password]` options.
Optionally, you can specify the KeyStore type using the `[--config-keystore-type]` option. By default, the KeyStore type is `PKCS12`.
The secrets in a KeyStore need to be stored using the `PBE` (password-based encryption) key algorithm, where a key is derived from a KeyStore password.
You can generate such a KeyStore using the following `keytool` command:
After executing the command, you will be prompted to *Enter the password to be stored*, which represents a value of the `kc.db-password` property above.
When the KeyStore is created, you can start the server using the following parameters:
However, for a specific behavior or capability that is missing in the {project_name} configuration, you can use properties from the underlying Quarkus framework.
If possible, avoid using properties directly from Quarkus, because they are unsupported by {project_name}. If your need is essential, consider opening an https://github.com/keycloak/keycloak/issues/new?assignees=&labels=kind%2Fenhancement%2Cstatus%2Ftriage&template=enhancement.yml[enhancement request] first. This approach helps us improve the configuration of {project_name} to fit your needs.
. Create a `quarkus.properties` file in the `conf` directory.
. Define the required properties in that file.
+
You can use only a https://github.com/keycloak/keycloak/blob/main/quarkus/runtime/pom.xml#L17[subset] of the Quarkus extensions that are defined in the https://quarkus.io/guides/all-config[Quarkus documentation]. Also, note these differences for Quarkus properties:
* A lock icon for a Quarkus property in the https://quarkus.io/guides/all-config[Quarkus documentation] indicates a build time property. You run the `build` command to apply this property. For details about the build command, see the subsequent sections on optimizing {project_name}.
* No lock icon for a property in the Quarkus guide indicates a runtime property for Quarkus and {project_name}.
Note that some Quarkus properties are already mapped in the {project_name} configuration, such as `quarkus.http.port` and similar essential properties. If the property is used by {project_Name}, defining that property key in `quarkus.properties` has no effect. The {project_Name} configuration value takes precedence over the Quarkus property value.
{project_name} depends upon Quarkus and MicroProfile for processing configuration values. Be aware that value expressions are supported. For example, `$\{some_key}` evaluates to the value of `some_key`.
To disable expression evaluation, the `\` character functions as an escape character. In particular, it must be used to escape the usage of `$` characters when they appear to define an expression or are repeated. For example, if you want the configuration value `my$$password`, use `my\$\$password` instead. Note that the `\` character requires additional escaping or quoting when using most unix shells, or when it appears in properties files.
For example, bash single quotes preserve the single backslash `--db-password='my\$\$password'`. Also, with bash double quotes, you need an additional backslash `--db-password="my\\$\\$password"`. Similarly in a properties file, backslash characters must also be escaped: `kc.db-password=my\\$\\$password`
Use development mode to try out {project_name} for the first time to get it up and running quickly. This mode offers convenient defaults for developers, such as for developing a new {project_name} theme.
Without further configuration, this command will not start {project_name} and show you an error instead. This response is done on purpose, because {project_name} follows a _secure by default_ principle. Production mode expects a hostname to be set up and an HTTPS/TLS setup to be available when started.
By default, example configuration options for the production mode are commented out in the default `conf/keycloak.conf` file. These options give you an idea about the main configuration to consider when running {project_name} in production.
You can create the initial admin user by using the web frontend, which you access using a local connection (localhost). You can instead create this user by using environment variables. Set `KC_BOOTSTRAP_ADMIN_USERNAME=_<username>_` for the initial admin username and `KC_BOOTSTRAP_ADMIN_PASSWORD=_<password>_` for the initial admin password.
Once the first user with administrative rights exists, you can use the Admin Console or the command line tool `kcadm.[sh|bat]` to create additional users.
If the initial administrator already exists and the environment variables are still present at startup, an error message stating the failed creation of the initial administrator is shown in the logs. {project_name} ignores the values and starts up correctly.
We recommend optimizing {project_name} to provide faster startup and better memory consumption before deploying {project_name} in a production environment. This section describes how to apply {project_name} optimizations for the best performance and runtime behavior.
This `build` command performs a set of optimizations for the startup and runtime behavior. The build process can take a few seconds. Especially when running {project_name} in containerized environments such as Kubernetes or OpenShift, startup time is important. To avoid losing that time, run a `build` explicitly before starting up, such as a separate step in a CI/CD pipeline.
This command shows `build options` that you enter. {project_name} distinguishes between **build options**, that are usable when running the `build` command, and **configuration options**, that are usable when starting up the server.
For a non-optimized startup of {project_name}, this distinction has no effect. However, if you run a build before the startup, only a subset of options is available to the build command. The restriction is due to the build options getting persisted into an optimized {project_name} image. For example, configuration for credentials such as `db-password` (which is a configuration option) must not get persisted for security reasons.
All build options are persisted in a plain text. Do not store any sensitive data as the build options. This applies across all the available configuration sources, including the KeyStore Config Source. Hence, we also do not recommend to store any build options in a Java keystore. Also, when it comes to the configuration options, we recommend to use the KeyStore Config Source primarily for storing sensitive data. For non-sensitive data you can use the remaining configuration sources.
To find available build options, see https://www.keycloak.org/server/all-config?f=build[All configuration page with build options selected] or enter the following command:
The `--optimized` parameter tells {project_name} to assume a pre-built, already optimized {project_name} image is used. As a result, {project_name} avoids checking for and running a build directly at startup, which saves time.
You can enter all configuration options at startup; these options are the ones in <@links.server id="all-config"/> that are **not** marked with a tool icon.
* If a build option is found at startup with a value that is equal to the value used when entering the `build`, that option gets silently ignored when you use the `--optimized` parameter.
* If that option has a different value than the value used when a build was entered, a warning appears in the logs and the previously built value is used. For this value to take effect, run a new `build` before starting.
You can achieve most optimizations to startup and runtime behavior by using the `build` command. Also, by using the `keycloak.conf` file as a configuration source, you avoid some steps at startup that would otherwise require command line parameters, such as initializing the CLI itself. As a result, the server starts up even faster.
{project_name} uses the Quarkus framework and a re-augmentation/mutable-jar approach under the covers. This process is started when a `build` command is run.
* A new closed-world assumption about installed providers is created, meaning that no need exists to re-create the registry and initialize the factories at every {project_name} startup.
* Configuration files are pre-parsed to reduce I/O when starting the server.
* Database specific resources are configured and prepared to run against a certain database vendor.
* By persisting build options into the server image, the server does not perform any additional step to interpret configuration options and (re)configure itself.
