Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 4 additions & 0 deletions CONTRIBUTING.md
Original file line number Diff line number Diff line change
Expand Up @@ -13,6 +13,10 @@ Check [Adding a New Instrumentation](docs/add_new_instrumentation.md) for instru

Check [How Instrumentations Work](docs/how_instrumentations_work.md) for a deep dive into how instrumentations work.

## Adding Configurations

Check [Adding a New Configuration](docs/add_new_configurations.md) for instructions on adding a new configuration.

## Code contributions

### Development environment quick check
Expand Down
59 changes: 59 additions & 0 deletions docs/add_new_configurations.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,59 @@
# Add a New Configuration

This doc will walk through how to properly add and document a new [configuration](https://docs.datadoghq.com/tracing/trace_collection/library_config/java/) in the Java Library.

## Where Configurations Live

All configurations in the Java Library are defined in the package `dd-trace-api/src/main/java/datadog/trace/api/config`.
Configurations are separated into different files based on the product they are related to. e.g. `CiVisiblity` related configurations live in `CiVisibilityConfig.java`, `Tracer` related in `TracerConfig.java`, etc.
Default values for configurations live in `ConfigDefaults.java`.

Configuration values are read and stored in `Config.java`, which utilizes helper methods in `ConfigProvider.java` to fetch the final value for a configuration after searching through all Configuration Sources and determining the final value based on Source priority. Below is the list of Sources that are queried for assigning Configuration values in order from highest to lowest priority:
1. System Properties: JVM System Properties
2. Stable Config - Fleet Automation: Now known as Declarative Configuration, this source reads a list of configurations through a `.yaml` file on Fleet Automation to take effect on all instances on a host.
3. CI Environment Variables: Source for Configurations related to the `CiVisibility` product.
4. Environment Variables: JVM Environment Variables
5. Properties File: By defining a filepath in `DD_TRACE_CONFIG`/`trace.config`, users can define Configuration key/value pairs in the file
6. OpenTelemetry Environment Variables: A list of OpenTelemetry Configurations that are supported in the Java Library. See [OtelEnvironmentConfigSource.java](../utils/config-utils/src/main/java/datadog/trace/bootstrap/config/provider/OtelEnvironmentConfigSource.java) for a list of supported OpenTelemetry Configurations.
7. Stable Config - Local File: Now known as Declarative Configuration, this source reads a list of configurations through a local `.yaml` file on set by the user to take effect on all instances on a host.
8. Captured Environment Variables: Auto-detects values for certain Configurations. Essentially setting "default" values for specific Configurations.

Additionally, `Config.java` also includes getters that can be used in other classes to get the value of a configuration. These getters should be the only method used to query the value of a configuration. Do **NOT** use `ConfigProvider.java` to re-query the values of a configuration.

## Adding a Configuration

In order to properly add a new configuration in the library, follow the below steps.
1. Determine whether this configuration exists in another tracing library in the [Feature Parity Dashboard](https://feature-parity.us1.prod.dog/#/configurations?viewType=configurations). Developers can search by Environment Variable name or description of the configuration.
1. If the configuration exists in a separate tracing library, reuse the name of the existing configuration. If the configuration already exists in the Java Library, utilize that instead.
2. Add the definition of the configuration to the appropriate "Config" class based on its related product.
1. When choosing a configuration definition, consider separating words by `.` and compound words by `-`. Note that `dd.` or `DD_` should not belong in the configuration definition as the base definitions are normalized to include those prefixes when querying the varying Configuration Sources. (e.g. `public static final String DOGSTATSD_START_DELAY = "dogstatsd.start-delay"`)
3. If a non-null default value for the configuration exists, define the value in a static field in `ConfigDefaults.java`.
4. Create a local field in `Config.java` to represent the configuration. In the constructor of `Config.java`, call the appropriate helper from `ConfigProvider.java` to query and assign the value of the configuration based off what datatype the Configuration expects to store. (e.g. `ConfigProvider::getString`, `ConfigProvider::getBoolean`, etc.)
1. This field should be final and not changed during runtime. If the value of a configuration needs to be changed, it can be done through a Snapshot with Dynamic Configuration. See [DynamicConfig.java](../internal-api/src/main/java/datadog/trace/api/DynamicConfig.java).
5. Create a getter for the field in `Config.java` to allow other classes to access the value of the configuration.
6. Add the configuration to the `toString()` method of `Config.java` for logging purposes.
7. Add the Environment Variable name of the configuration to the `supportedConfigurations` key of `metadata/supported-configurations.json` in the format of `ENV_VAR: ["VERSION", ...]`. If the configuration already existed in another library, add the version listed on the Feature Parity Dashboard. If introducing a new configuration, provide a version of `A`.
1. If there are aliases of the Environment Variable, add them to the `aliases` key of the file.

See below for the format of the `supported-configurations.json` file.
```
{
"supportedConfigurations": {
"DD_ENV_VAR": ["A"],
"DD_TEST_VAR": ["A"]
},
"aliases": {
"DD_ENV_VAR": ["DD_ENV_ALIAS"]
},
"deprecations": {
}
}
```
8. If the configuration is unique to all tracing libraries, add it to the [Feature Parity Dashboard](https://feature-parity.us1.prod.dog/#/configurations?viewType=configurations). This ensures that we have good documentation of all configurations supported in the library.
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Suggested change
8. If the configuration is unique to all tracing libraries, add it to the [Feature Parity Dashboard](https://feature-parity.us1.prod.dog/#/configurations?viewType=configurations). This ensures that we have good documentation of all configurations supported in the library.
8. If the configuration is unique to this library, add it to the [Feature Parity Dashboard](https://feature-parity.us1.prod.dog/#/configurations?viewType=configurations). This ensures that we have good documentation of all configurations supported in the library.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This was intentional. If we are adding a config to Java that exists in other libraries, we expect it to already be documented in the FPD


For details on adding environment variables to `metadata/supported-configurations.json` or the Feature Parity Dashboard, refer to this [document](https://datadoghq.atlassian.net/wiki/spaces/APMINT/pages/5372248222/APM+-+Centralized+Configuration+Config+inversion#dd-trace-java).

## Verifying the Configuration

To verify a configuration has been correctly added, developers can modify existing test cases in `ConfigTest.groovy` to set the value of the configuration with various sources and confirm the internal value is correctly set in `Config.java`.
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Would be sick to have a new test populate for the new config automatically... maybe someday. 😄

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Not sure that "automated generated tests" will have real value if it's only testing "getter" behavior to comply with the code coverage policy.
To me, it make sense to write tests when there is a specific logic at play, like this will be enabled if both X and Y are enable and Z contains xxx.

On thing I would like at mid term is to have:

  • Config declaration per product -- like we have today TracerConfig, CiVisibilityConfig
  • Such config would declare keys, aliases, type (and parser if needed)
  • This config declaration would be used to generate the Config class(es).

It does not have to be a god class with all config from it, it can be split by products too.
But this would get rid of the contructor / getter / toString / testing boilerplate...

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Sounds like this could be something Config Inversion/Registry could be very useful for with our new v2 format. :)

Optionally, new test cases can be added for testing specific to the behavior of a configuration.
Loading