broker 3个节点,其中一个连不上zookeeper。
保错:
FATAL Could not establish session with zookeeper (kafka.server.KafkaHealthcheck$SessionExpireListener)
[2017-11-01 04:12:46,591] INFO Partition [__consumer_offsets,47] on broker 4: Shrinking ISR for partition [__consumer_offsets,47] from 4,3,5 to 4 (kafka.cluster.Partition)
[2017-11-01 04:12:46,634] WARN Conditional update of path /brokers/topics/__consumer_offsets/partitions/47/state with data {"controller_epoch":137,"leader":4,"version":1,"leader_epoch":3041,"isr":[4]} and expected version 15909 failed due to null (kafka.utils.ZkUtils)
[2017-11-01 04:12:46,634] INFO Partition [__consumer_offsets,47] on broker 4: Cached zkVersion [15909] not equal to that in zookeeper, skip updating ISR (kafka.cluster.Partition)
[2017-11-01 04:12:46,634] INFO Partition [__consumer_offsets,41] on broker 4: Shrinking ISR for partition [__consumer_offsets,41] from 4,3,5 to 4 (kafka.cluster.Partition)
[2017-11-01 04:12:46,634] WARN Conditional update of path /brokers/topics/__consumer_offsets/partitions/41/state with data {"controller_epoch":137,"leader":4,"version":1,"leader_epoch":3040,"isr":[4]} and expected version 15923 failed due to null (kafka.utils.ZkUtils)
[2017-11-01 04:12:46,635] INFO Partition [__consumer_offsets,41] on broker 4: Cached zkVersion [15923] not equal to that in zookeeper, skip updating ISR (kafka.cluster.Partition)
一直重复报错,而且这些partition 数据重复消费旧数据
zookeeper什么版本?
zookeeper-3.5
kafka 0.10.0
把zk升到3.6或者3.9。
你的答案