Kafka Rest Proxy JSON schema validation

I want to produce to a kafka topic via Kafka Rest Proxy. I've created a JSON schema in the schema-registry and I would like that all messages are validated against the registered schema and rejected if they don't match the Schema.

My Schema

{
  "type": "object",
  "properties": {
    "foo": {
      "type": "string",
    },
    "bar": {
      "type": "number" 
    }
  }
}

This schema is registered correctly and assigned version 1. Then I try to produce a message with the wrong data type for both foo and bar but the message is accepted.

curl --location --request POST 'http://localhost:8082/topics/test' \
--header 'Content-Type: application/vnd.kafka.jsonschema.v2+json' \
--header 'Accept: application/vnd.kafka.v2+json' \
--data-raw '{
    "value_schema_id": 1,
    "records": [
        {
            "value": {
                "foo": 10,
                "bar":"not a number"
            }
        }
    ]
}'

Notice that I'm producing to the test topic that has a schema associated, but the faulty message get's accepted anyways. I also tried adding "value_schema_id": 1 to make sure the schema is referenced in the payload, but the faulty message is still accepted.

However, if I pass the JSON schema as value_schema it works as expected

{
    "value_schema": "{\"type\": \"object\",\"properties\": {\"foo\": {\"type\": \"string\"},\"bar\": {\"type\": \"number\"}}}",
    "records": [
        {
            "value": {
                "foo": "10",
                "bar": "1"
            }
        }
    ]
}

Response

{
    "error_code": 42203,
    "message": "Conversion of JSON to Object failed: Failed to convert JSON using JSON Schema: #/bar: expected type: Number, found: String"
}

Question: Is it possible to reference an existing schema id when producing messages without having to pass the whole JSON schema each time?

Yes, it's possible. You have to enable it though to either value, key or both as per below:

confluent.value.schema.validation=true 
confluent.key.schema.validation=true

Also, in your JSON request you can specify the ID of your schema. Take the below as an example, to illustrate what I'm saying:

{
  "key_schema_id": 1234,
  "value_schema_id: 56789,
  "records": [
    {
      "key": "key123",
      "value": "my_simple_string_value_example"
    }
  ]
}

Where:

key_schema_id

Is the ID of the key schema in Schema Registry. By setting an appropriate value to it will guarantee that only messages with a key compliant with the schema identified by that ID will be accepted.

Similarly:

value_schema_id

Is the ID of the value schema in Schema Registry. By setting an appropriate value to it will guarantee that only messages with a value compliant with the schema identified by that ID will be accepted.

Hope that helps to shed some light.

Cheers,

Eduardo Ponzoni