Kevin Tseng created FLINK-31951:
-----------------------------------
Summary: Mix schema record source creates corrupt record
Key: FLINK-31951
URL: https://issues.apache.org/jira/browse/FLINK-31951
Project: Flink
Issue Type: Bug
Components: Formats (JSON, Avro, Parquet, ORC, SequenceFile)
Affects Versions: 1.17.0, 1.18.0
Reporter: Kevin Tseng
This seems to be an unexpected side effect with how AvroDeserializationSchema
class was written.
Sometimes we do not have control over what record comes through a Kafka Topic.
In current implementation, if AvroDeserializationSchema encountered a record
byte array that does not conform to the specified Schema / SpecificRecord type,
it will cause future record to be deserialized incorrectly.
Origin of the issue is with how
{code:java}
AvroDeserializationSchema.deserialize{code}
handles exception, and how
{code:java}
AvroDeserializationSchema.checkAvroInitialized{code}
handles initialization of Decoder object
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
