feat(sink): support encode protobuf with confluent schema registry

[xiangjinwu]

I hereby agree to the terms of the RisingWave Labs, Inc. Contributor License Agreement.

What's changed and what's your intention?

When creating kafka sink with format plain encode protobuf, the schema definition can now come from confluent schema registry, in addition to previously supported http[s]:// and file://.

Checklist

• I have written necessary rustdoc comments
• I have added necessary unit tests and integration tests
• I have added test labels as necessary. See details.
• I have added fuzzing tests or opened an issue to track them. (Optional, recommended for new SQL features Sqlsmith: Sql feature generation #7934).
• My PR contains breaking changes. (If it deprecates some features, please create a tracking issue to remove them in the future).
• All checks passed in ./risedev check (or alias, ./risedev c)
• My PR changes performance-critical code. (Please run macro/micro-benchmarks and show the results.)

• My PR contains critical fixes that are necessary to be merged into the latest release. (Please check out the details)

Documentation

• My PR needs documentation updates. (Please use the Release note section below to summarize the impact on users)

Release note

Support confluent schema registry for kafka sink using format plain encode protobuf.

The following options are same as in protobuf/avro source and avro sink.

• schema.registry (instead of schema.location): required
• schema.registry.username
• schema.registry.password
• schema.registry.name.strategy: optional and defaults to topic_name_strategy

The following options are same as protobuf source, but different from avro source/sink:

• message
  • required for protobuf
  • (for avro, only required when name strategy is record_name_strategy or topic_record_name_strategy)
• key.message
  • NOT APPLICABLE for protobuf (because we only support format plain; there's no format upsert support)
  • (for avro, only required when name strategy is record_name_strategy or topic_record_name_strategy)

Example:

format plain encode protobuf (
  schema.registry = 'http://message_queue:8081',
  message = 'test.package.MessageA');

[xiangjinwu]

This can be pre-computed and is same for all messages.

[tabVersion]

generally LGTM, thanks

[tabVersion]

do we also need to handle the auth for schema registry here?

[xiangjinwu]

It is handled inside the unified SchemaLoader (#14642)

risingwave/src/connector/src/schema/loader.rs

Line 46 in ef7eb1d

let client_config = format_options.into();

But yes our e2e tests use redpanda without password, and cannot catch bugs on auth (e.g. #14755).

[xiangjinwu]

Seems to be an involved process. Leaving it as a separate issue.
https://docs.redpanda.com/23.2/manage/security/authentication/#configure-basic-authentication