I have a 3 node Kafka cluster. Upgrading Kafka version from 0.11.0.2 to 0.11.0.3, I notice that Kafka sometimes truncate new message's topic. Here is the log:
[2018-07-30 09:19:39,612] WARN [ReplicaFetcherThread-0-101]: Based on follower's leader epoch, leader replied with an unknown offset in 7798-7. High watermark 0 will be used for truncation. (kafka.server.ReplicaFetcherThread) [2018-07-30 09:19:39,612] WARN [ReplicaFetcherThread-0-101]: Based on follower's leader epoch, leader replied with an unknown offset in 7798-22. High watermark 0 will be used for truncation. (kafka.server.ReplicaFetcherThread) [2018-07-30 09:19:39,612] WARN [ReplicaFetcherThread-0-101]: Based on follower's leader epoch, leader replied with an unknown offset in 7798-19. High watermark 0 will be used for truncation. (kafka.server.ReplicaFetcherThread) [2018-07-30 09:19:39,612] WARN [ReplicaFetcherThread-0-101]: Based on follower's leader epoch, leader replied with an unknown offset in 7798-4. High watermark 0 will be used for truncation. (kafka.server.ReplicaFetcherThread) [2018-07-30 09:19:39,613] WARN [ReplicaFetcherThread-0-101]: Based on follower's leader epoch, leader replied with an unknown offset in 7798-1. High watermark 0 will be used for truncation. (kafka.server.ReplicaFetcherThread) [2018-07-30 09:19:39,613] WARN [ReplicaFetcherThread-0-101]: Based on follower's leader epoch, leader replied with an unknown offset in 7798-16. High watermark 0 will be used for truncation. (kafka.server.ReplicaFetcherThread) [2018-07-30 09:19:39,613] WARN [ReplicaFetcherThread-0-101]: Based on follower's leader epoch, leader replied with an unknown offset in 7798-13. High watermark 0 will be used for truncation. (kafka.server.ReplicaFetcherThread) [2018-07-30 09:19:39,613] WARN [ReplicaFetcherThread-0-101]: Based on follower's leader epoch, leader replied with an unknown offset in 7798-10. High watermark 0 will be used for truncation. (kafka.server.ReplicaFetcherThread) [2018-07-30 09:19:39,615] WARN [ReplicaFetcherThread-0-102]: Based on follower's leader epoch, leader replied with an unknown offset in 7798-23. High watermark 0 will be used for truncation. (kafka.server.ReplicaFetcherThread) [2018-07-30 09:19:39,627] WARN [ReplicaFetcherThread-0-102]: Based on follower's leader epoch, leader replied with an unknown offset in 7798-8. High watermark 0 will be used for truncation. (kafka.server.ReplicaFetcherThread) [2018-07-30 09:19:39,627] WARN [ReplicaFetcherThread-0-102]: Based on follower's leader epoch, leader replied with an unknown offset in 7798-5. High watermark 0 will be used for truncation. (kafka.server.ReplicaFetcherThread) [2018-07-30 09:19:39,627] WARN [ReplicaFetcherThread-0-102]: Based on follower's leader epoch, leader replied with an unknown offset in 7798-20. High watermark 0 will be used for truncation. (kafka.server.ReplicaFetcherThread) [2018-07-30 09:19:39,627] WARN [ReplicaFetcherThread-0-102]: Based on follower's leader epoch, leader replied with an unknown offset in 7798-17. High watermark 0 will be used for truncation. (kafka.server.ReplicaFetcherThread) [2018-07-30 09:19:39,627] WARN [ReplicaFetcherThread-0-102]: Based on follower's leader epoch, leader replied with an unknown offset in 7798-2. High watermark 0 will be used for truncation. (kafka.server.ReplicaFetcherThread) [2018-07-30 09:19:39,627] WARN [ReplicaFetcherThread-0-102]: Based on follower's leader epoch, leader replied with an unknown offset in 7798-14. High watermark 0 will be used for truncation. (kafka.server.ReplicaFetcherThread) [2018-07-30 09:19:39,627] WARN [ReplicaFetcherThread-0-102]: Based on follower's leader epoch, leader replied with an unknown offset in 7798-11. High watermark 0 will be used for truncation. (kafka.server.ReplicaFetcherThread) [2018-07-30 09:19:39,628] INFO Truncating 7798-20 to 0 has no effect as the largest offset in the log is -1. (kafka.log.Log) [2018-07-30 09:19:39,628] INFO Truncating 7798-17 to 0 has no effect as the largest offset in the log is -1. (kafka.log.Log) [2018-07-30 09:19:39,628] INFO Truncating 7798-14 to 0 has no effect as the largest offset in the log is -1. (kafka.log.Log) [2018-07-30 09:19:39,628] INFO Truncating 7798-11 to 0 has no effect as the largest offset in the log is -1. (kafka.log.Log) [2018-07-30 09:19:39,628] INFO Truncating 7798-8 to 0 has no effect as the largest offset in the log is -1. (kafka.log.Log) [2018-07-30 09:19:39,628] INFO Truncating 7798-2 to 0 has no effect as the largest offset in the log is -1. (kafka.log.Log) [2018-07-30 09:19:39,628] INFO Truncating 7798-5 to 0 has no effect as the largest offset in the log is -1. (kafka.log.Log) [2018-07-30 09:19:39,628] INFO Truncating 7798-23 to 0 has no effect as the largest offset in the log is -1. (kafka.log.Log) [2018-07-30 09:19:39,613] INFO Truncating 7798-1 to 0 has no effect as the largest offset in the log is -1. (kafka.log.Log) [2018-07-30 09:19:39,637] INFO Truncating 7798-22 to 0 has no effect as the largest offset in the log is -1. (kafka.log.Log) [2018-07-30 09:19:39,637] INFO Truncating 7798-16 to 0 has no effect as the largest offset in the log is -1. (kafka.log.Log) [2018-07-30 09:19:39,637] INFO Truncating 7798-19 to 0 has no effect as the largest offset in the log is -1. (kafka.log.Log) [2018-07-30 09:19:39,637] INFO Truncating 7798-13 to 0 has no effect as the largest offset in the log is -1. (kafka.log.Log) [2018-07-30 09:19:39,637] INFO Truncating 7798-10 to 0 has no effect as the largest offset in the log is -1. (kafka.log.Log) [2018-07-30 09:19:39,637] INFO Truncating 7798-7 to 0 has no effect as the largest offset in the log is -1. (kafka.log.Log) [2018-07-30 09:19:39,637] INFO Truncating 7798-4 to 0 has no effect as the largest offset in the log is -1. (kafka.log.Log) [2018-07-30 09:19:39,708] ERROR [ReplicaFetcherThread-0-102]: Error for partition [7798,8] to broker 102:org.apache.kafka.common.errors.UnknownTopicOrPartitionException: This server does not host this topic-partition. (kafka.server.ReplicaFetcherThread) [2018-07-30 09:19:39,708] ERROR [ReplicaFetcherThread-0-102]: Error for partition [7798,2] to broker 102:org.apache.kafka.common.errors.UnknownTopicOrPartitionException: This server does not host this topic-partition. (kafka.server.ReplicaFetcherThread) [2018-07-30 09:19:39,708] ERROR [ReplicaFetcherThread-0-102]: Error for partition [7798,23] to broker 102:org.apache.kafka.common.errors.UnknownTopicOrPartitionException: This server does not host this topic-partition. (kafka.server.ReplicaFetcherThread) [2018-07-30 09:19:39,708] ERROR [ReplicaFetcherThread-0-102]: Error for partition [7798,17] to broker 102:org.apache.kafka.common.errors.UnknownTopicOrPartitionException: This server does not host this topic-partition. (kafka.server.ReplicaFetcherThread) [2018-07-30 09:19:39,708] ERROR [ReplicaFetcherThread-0-102]: Error for partition [7798,5] to broker 102:org.apache.kafka.common.errors.UnknownTopicOrPartitionException: This server does not host this topic-partition. (kafka.server.ReplicaFetcherThread) [2018-07-30 09:19:39,708] ERROR [ReplicaFetcherThread-0-102]: Error for partition [7798,14] to broker 102:org.apache.kafka.common.errors.UnknownTopicOrPartitionException: This server does not host this topic-partition. (kafka.server.ReplicaFetcherThread) [2018-07-30 09:19:39,708] ERROR [ReplicaFetcherThread-0-102]: Error for partition [7798,11] to broker 102:org.apache.kafka.common.errors.UnknownTopicOrPartitionException: This server does not host this topic-partition. (kafka.server.ReplicaFetcherThread) [2018-07-30 09:19:39,709] ERROR [ReplicaFetcherThread-0-102]: Error for partition [7798,20] to broker 102:org.apache.kafka.common.errors.UnknownTopicOrPartitionException: This server does not host this topic-partition. (kafka.server.ReplicaFetcherThread) [2018-07-30 09:19:39,714] ERROR [ReplicaFetcherThread-0-101]: Error for partition [7798,7] to broker 101:org.apache.kafka.common.errors.UnknownTopicOrPartitionException: This server does not host this topic-partition. (kafka.server.ReplicaFetcherThread) [2018-07-30 09:19:39,714] ERROR [ReplicaFetcherThread-0-101]: Error for partition [7798,13] to broker 101:org.apache.kafka.common.errors.UnknownTopicOrPartitionException: This server does not host this topic-partition. (kafka.server.ReplicaFetcherThread) [2018-07-30 09:19:39,714] ERROR [ReplicaFetcherThread-0-101]: Error for partition [7798,22] to broker 101:org.apache.kafka.common.errors.UnknownTopicOrPartitionException: This server does not host this topic-partition. (kafka.server.ReplicaFetcherThread) [2018-07-30 09:19:39,715] ERROR [ReplicaFetcherThread-0-101]: Error for partition [7798,4] to broker 101:org.apache.kafka.common.errors.UnknownTopicOrPartitionException: This server does not host this topic-partition. (kafka.server.ReplicaFetcherThread) [2018-07-30 09:19:39,715] ERROR [ReplicaFetcherThread-0-101]: Error for partition [7798,10] to broker 101:org.apache.kafka.common.errors.UnknownTopicOrPartitionException: This server does not host this topic-partition. (kafka.server.ReplicaFetcherThread) [2018-07-30 09:19:39,715] ERROR [ReplicaFetcherThread-0-101]: Error for partition [7798,19] to broker 101:org.apache.kafka.common.errors.UnknownTopicOrPartitionException: This server does not host this topic-partition. (kafka.server.ReplicaFetcherThread) [2018-07-30 09:19:39,715] ERROR [ReplicaFetcherThread-0-101]: Error for partition [7798,1] to broker 101:org.apache.kafka.common.errors.UnknownTopicOrPartitionException: This server does not host this topic-partition. (kafka.server.ReplicaFetcherThread) t seems at the first Kafka save the message and after that, it started to truncate messages, right away! I guess it's because of resolving this bug <https://issues.apache.org/jira/browse/KAFKA-6857> in Kafka version 11.0.3. This scenario happened even for __consumer_offset topic. Is there any way to solve the problem?
