Kishore Senji created KAFKA-3055: ------------------------------------ Summary: JsonConverter mangles schema during serialization (fromConnectData) Key: KAFKA-3055 URL: https://issues.apache.org/jira/browse/KAFKA-3055 Project: Kafka Issue Type: Bug Components: copycat Affects Versions: 0.9.0.0 Reporter: Kishore Senji Assignee: Ewen Cheslack-Postava
Test case is here: https://github.com/ksenji/kafka-connect-test/tree/master/src/test/java/org/apache/kafka/connect/json If Caching is disabled, it behaves correctly and JsonConverterWithNoCacheTest runs successfully. Otherwise the test JsonConverterTest fails. The reason is that the JsonConverter has a bug where it mangles the schema as it assigns all String fields with the same name (and similar for all Int32 fields) This is how the schema & payload gets serialized for the Person Struct (with caching disabled): {"schema":{"type":"struct","fields":[{"type":"string","optional":false,"field":"firstName"},{"type":"string","optional":false,"field":"lastName"},{"type":"string","optional":false,"field":"email"},{"type":"int32","optional":false,"field":"age"},{"type":"int32","optional":false,"field":"weightInKgs"}],"optional":false,"name":"Person"},"payload":{"firstName":"Eric","lastName":"Cartman","email":"eric.cart...@southpark.com","age":10,"weightInKgs":40}} where as when caching is enabled the same Struct gets serialized as (with caching enabled) : {"schema":{"type":"struct","fields":[{"type":"string","optional":false,"field":"email"},{"type":"string","optional":false,"field":"email"},{"type":"string","optional":false,"field":"email"},{"type":"int32","optional":false,"field":"weightInKgs"},{"type":"int32","optional":false,"field":"weightInKgs"}],"optional":false,"name":"Person"},"payload":{"firstName":"Eric","lastName":"Cartman","email":"eric.cart...@southpark.com","age":10,"weightInKgs":40}} As we can see all String fields became "email" and all int32 fields became "weightInKgs". -- This message was sent by Atlassian JIRA (v6.3.4#6332)