Avro Schema Registry

DO NOT USE IN PRODUCTION

This is an implementation of the Confluent Schema Registry API (mostly) compatible with the Rails implementation by salsify/avro-schema-registry.

I kept the database schema used by salsify and my main goal was to replace the Rails service with this one. There's nothing wrong with theirs, I just happen to enjoy Rust more than Ruby. All thanks should go to salsify for their initial design and ideas on using PostgreSQL instead of Kafka as a backend.

Note: the backend used is PostgreSQL (schema compatible with salsify/avro-schema-registry PostgreSQL schema). The Confluent Schema Registry uses Kafka as a backend therefore we don't provide exactly the same semantics (but close enough).

Fingerprint

In salsify/avro-schema-registry two fingerprints are present (v1 and v2). In this implementation, we only make use of v2 and do NOT support v1.

If you are still using fingerprints with v1, please make sure you migrate first, before using this service as your API.

Endpoints

Endpoint	Method	Maturity
/compatibility/subjects/{subject}/versions/{version}	POST	Ready (NONE, BACKWARD, FORWARD, FULL); Unimplemented (TRANSITIVE)
/config	GET	Ready
/config	PUT	Ready
/config/{subject}	GET	Ready
/config/{subject}	PUT	Ready
/schemas/ids/{id}	GET	Ready
/subjects	GET	Ready
/subjects/{subject}	DELETE	Ready
/subjects/{subject}	POST	Ready
/subjects/{subject}/versions	GET	Ready
/subjects/{subject}/versions	POST	Ready
/subjects/{subject}/versions/latest	DELETE	Ready
/subjects/{subject}/versions/latest	GET	Ready
/subjects/{subject}/versions/{version}	DELETE	Ready
/subjects/{subject}/versions/{version}	GET	Ready
/subjects/{subject}/versions/latest/schema	GET	Ready
/subjects/{subject}/versions/{version}/schema	GET	Ready

Extra Endpoints

Endpoint	Method	Maturity
/_/health_check	GET	Incomplete
/_/metrics	GET	Ready

Build

cargo build --release

Run

This assumes you have a running PostgreSQL instance (versions 9.5 and above) and you've run the diesel setup (with its migrations).

1. Setup env (everything is controlled through environment variables)

export SENTRY_URL="http://sentry-url/id" \ # optional
    DEFAULT_HOST=127.0.0.1:8080 \ # optional (default is 127.0.0.1:8080)
    DATABASE_URL=postgres://postgres:@localhost:5432/diesel_testing \
    SCHEMA_REGISTRY_PASSWORD=silly_password

2. Run application

# If you haven't set PORT, it listens on the default 8080
cargo run # or the binary after running `cargo build`

3. Make a request

# If you assume the credentials above:
curl -v -H 'Authorization: Basic OnNpbGx5X3Bhc3N3b3Jk' -H 'Accept: application/json' http://127.0.0.1:8080/config

Tests

Unit

cargo test middleware

Integration

1. Setup testing environment variables

export RUST_TEST_THREADS=1 \
    DATABASE_URL=postgres://postgres:@localhost:5432/diesel_testing \
    SCHEMA_REGISTRY_PASSWORD=silly_password

2. Run test suite

cargo test speculate

Important

We don't ever return Error code 50003 -- Error while forwarding the request to the master. This is because this error is specific to Kafka.

Contributing

You are more than welcome to contribute to this project. Fork and make a Pull Request, or create an Issue if you see any problem.