Read the FAQ first: https://github.com/edenhill/librdkafka/wiki/FAQ
Description
We met some fatal error with idempotence producer after kafka cluster was down, -- the producer would not be able to send any more message from then on!
The problem still exists even with latest librdkafka release (v1.8.2) -- thus not the same problem with #3577
Here's the log while error occurred,
[2021-10-19 18:47:34.885129]EMERG KafkaProducer[producer-140691259498496] FATAL | [thrd:127.0.0.1:29092/bootstrap]: Fatal error: Broker: Broker received an out of order sequence number: ProduceRequest for topicForTest [0] with 1 message(s) failed due to sequence desynchronization with broker 1 (PID{Id:8000,Epoch:0}, base seq 0, idemp state change 1383467ms ago, last partition error NOT_LEADER_FOR_PARTITION (actions Refresh,MsgNotPersisted, base seq 0..0, base msgid 1, 107ms ago)
Met error: Broker: Broker received an out of order sequence number [45] fatal | ProduceRequest for topicForTest [0] with 1 message(s) failed due to sequence desynchronization with broker 1 (PID{Id:8000,Epoch:0}, base seq 0, idemp state change 1383467ms ago, last partition error NOT_LEADER_FOR_PARTITION (actions Refresh,MsgNotPersisted, base seq 0..0, base msgid 1, 107ms ago)
Exception thrown by producer: 2021-10-19 18:47:34.885264: Broker: Broker received an out of order sequence number [45] (/home/winner/Repo/modern-cpp-kafka-3/include/./kafka/KafkaProducer.h:435)
(However, it seems no related log from broker side)
How to reproduce
- Start the Kafka cluster (v2.8.0)
- Start an idempotent producer (with librdkafka v1.8.2), send a few messages and then wait
- Kill the zookeeper and Kafka brokers processes
- Wait ~20 min
- Re-start the Kafka cluster
- Try to send a message with the previous producer, and probably fail with the "out-of-sequence-number" fatal error, -- the producer is not functional anymore.
Checklist
IMPORTANT: We will close issues where the checklist has not been completed.
Please provide the following information:
Read the FAQ first: https://github.com/edenhill/librdkafka/wiki/FAQ
Description
We met some fatal error with idempotence producer after kafka cluster was down, -- the producer would not be able to send any more message from then on!
The problem still exists even with latest librdkafka release (
v1.8.2) -- thus not the same problem with #3577Here's the log while error occurred,
(However, it seems no related log from broker side)
How to reproduce
Checklist
IMPORTANT: We will close issues where the checklist has not been completed.
Please provide the following information:
1.8.2kafka_2.13-2.8.0enable.idempotence=trueubuntu20.04 x64debug=..as necessary) from librdkafka