Hi, I would like to start the voting on KIP-126. The KIP is intended to solve the problem that RecordTooLargeExceptions are thrown from the producer due to inaccurate estimation of the compression ratio. The solution is to split and resend the over sized batches if possible. A new metric is introduced to the producer to show the batch split rate.
The KIP wiki is following: https://cwiki.apache.org/confluence/display/KAFKA/KIP-126 +-+Allow+KafkaProducer+to+batch+based+on+uncompressed+size We have been running a producer with this patch for some time in our mirror maker and it looks working fine. Thanks, Jiangjie (Becket) Qin
