Hi, Shri,
So, it seems that the duplicates are introduced by the publisher.
Currently, producer retries can introduce duplicates. We are trying to
eliminate that in the EOS work (KIP-98). We plan to have a beta version of
that available in 0.11.0.
Hi, Jeff,
We have a known issue of recovering fr
The kafka-console-producer.sh defaults to acks=1 so just be careful with
using those tools for too much debugging. Your output is helpful though.
https://github.com/apache/kafka/blob/5a2fcdd6d480e9f003cc49a59d5952ba4c515a71/core/src/main/scala/kafka/tools/ConsoleProducer.scala#L185
-hans
On Wed,
Just to add, I see below behavior repeat with even command line console
producer and consumer that come with Kafka.
Thanks,
Shri
__
Shrikant Patel | 817.367.4302
Enterprise Architecture Team
PDX-NHIN
-Original Message-
From: Shrikant Pat
Thanks Jeff, Onur, Jun, Hans. I am learning a lot from your response.
Just to summarize briefly my steps, 5 node Kafka and ZK cluster.
1. ZK cluster has all node working. Consumer is down.
2. Bring down majority of ZK nodes.
3. Thing are functional no issue (no dup or lost message)
4. Now first ka
