a00d9f17b1
wolff-1.10.4 handles message_too_large error code from Kafka. It tries to split the too-large batch into single-call batches (for EMQX, one call batch is always one message), and retry. In case a single-call batch is too large, EMQX should increment a failure counter.
405 B
405 B
Improve Kafka producer error handling for message_too_large.
Prior to this change, Kafka producers would retry sending oversized batches (message_too_large error) in hopes of a server side configuration fix (max.message.bytes).
Now, oversized messages are automatically split into single-message batches for retry. If a message still exceeds size limits, it will be dropped to maintain data flow.