Official Sentry Relay

The Sentry Relay is a service that pushes some functionality from the Sentry SDKs as well as the Sentry server into a proxy process.

Documentation

• Product documentation can be found at: https://docs.sentry.io/meta/relay/.
• Code and development documentation can be found at: https://getsentry.github.io/relay/.

License

Like Sentry, Relay is licensed under the BSL. See the LICENSE file and this forum post for more information.

Development

To build Relay, we require the latest stable Rust. The crate is split into a workspace with multiple features, so when running building or running tests always make sure to pass the --all and --all-features flags. The processing feature additionally requires a C compiler and CMake.

We use VSCode for development. This repository contains settings files configuring code style, linters, and useful features. When opening the project for the first time, make sure to install the Recommended Extensions, as they will allow editor assist during coding.

The root of the repository contains a Makefile with useful commands for development:

• make check: Runs code formatting checks and linters. This is useful before opening a pull request.
• make test: Runs unit tests, integration tests and Python package tests (see below for more information).
• make all: Runs all checks and tests. This runs most of the tasks that are also performed in CI.
• make clean: Removes all build artifacts, the virtualenv and cached files.

For a lot of tests you will need Redis and Kafka running in their respective default configuration. sentry devservices from sentry does this for you.

Building and Running

The easiest way to rebuild and run Relay is using cargo. Depending on the configuration, you may need to have a local instance of Sentry running.

# Initialize Relay for the first time
cargo run --all-features -- config init

# Rebuild and run with all features
cargo run --all-features -- run

The standard build commands are also available as make targets. Note that release builds still generate debug information.

# Build without optimizations in debug mode.
make build

# Build with release optimizations and debug information.
make release

For quickly verifying that Relay compiles after making some changes, you can also use cargo check:

cargo check --all --all-features

Features

By default, Relay compiles without processing mode. This is the configuration used for Relays operating as proxys. There are two optional features:

• processing: Enables event processing and ingestion functionality. This allows to enable processing in config. When enabled, Relay will produce events into a Kafka topic instead of forwarding to the configured upstream. Also, it will perform full event normalization, filtering, and rate limiting.

• ssl: Enables SSL support in the Server.

To enable a feature, pass it to the cargo invocation. For example, to run tests across all workspace crates with the processing feature enabled, run:

cargo run --features=processing

Tests

The test suite comprises unit tests, an integration test suite and a separate test suite for the Python package. Unit tests are implemented as part of the Rust crates and can be run via:

# Tests for default features
make test-rust

# Run Rust tests for all features
make test-rust-all

The integration test suite requires python. By default, the integration test suite will create a virtualenv, build the Relay binary with processing enabled, and run a set of integration tests:

# Create a new virtualenv, build Relay and run integration tests
make test-integration

# Build and run a single test manually
make build
.venv/bin/pytest tests/integration -k <test_name>

Linting

We use rustfmt and clippy from the latest stable channel for code formatting and linting. To make sure that these tools are set up correctly and running with the right configuration, use the following make targets:

# Format the entire codebase
make format

# Run clippy on the entire codebase
make lint

Python and C-ABI

Potentially, new functionality also needs to be added to the Python package. This first requires to expose new functions in the C ABI. For this, refer to the Relay C-ABI readme.

We highly recommend to develop and test the python package in a virtual environment. Once the ABI has been updated and tested, ensure the virtualenv is active and install the package, which builds the native library. There are two ways to install this:

# Install the release build, recommended:
pip install --editable ./py

# Install the debug build, faster installation but much slower runtime:
RELAY_DEBUG=1 pip install --editable ./py

For testing, we use ubiquitous pytest. Again, ensure that your virtualenv is active and the latest version of the native library has been installed. Then, run:

# Create a new virtualenv, install the release build and run tests
make test-python

# Run a single test manually
.venv/bin/pytest py/tests -k <test_name>

Development Server

If you have systemfd and cargo-watch installed, the make devserver command can auto-reload Relay:

cargo install systemfd cargo-watch
make devserver

SSL

The repository contains a SSL-certificate + private key for development purposes. It comes in two formats: Once as a (.pem, .cert)-pair, once as .pfx (PKCS #12) file.

The password for the .pfx file is password.

Usage with Sentry

To develop Relay with an existing Sentry devserver, Sentry onpremise installation, or Sentry SaaS, configure the upstream to the URL of the Sentry server in .relay/config.yml. For example, in local development set relay.upstream to http://localhost:8000/.

To test processing mode with a local development Sentry, use this configuration:

relay:
  # Point to your Sentry devserver URL:
  upstream: http://localhost:8000/
  # Listen to a port other than 3000:
  port: 3001
logging:
  # Enable full logging and backraces:
  level: trace
  enable_backtraces: true
limits:
  # Speed up shutdown on ^C
  shutdown_timeout: 0
processing:
  # Enable processing mode with store normalization and post data to Kafka:
  enabled: true
  kafka_config:
    - { name: "bootstrap.servers", value: "127.0.0.1:9092" }
    - { name: "message.max.bytes", value: 2097176 }
  redis: "redis://127.0.0.1"

Note that the Sentry devserver also starts a Relay in processing mode on port 3000 with similar configuration. That Relay does not interfere with your development build. To ensure SDKs send to your development instance, update the port in the DSN:

http://<key>@localhost:3001/<id>

Release Management

We use craft to release new versions. There are two separate projects to publish:

• Relay binary is released from the root folder. Run craft prepare and craft publish in that directory to create a release build and publish it, respectively. We use Calendar Versioning and coordinate releases with Sentry.

• Relay Python library along with the C-ABI are released from the py/ subfolder. Change into that directory and run craft prepare and craft publish. We use Semantic Versioning and release during the development cycle.