Our setup is a three node cluster, we have a Kafka 0.7.2 (with 10 partitions 
per node) and Zookeeper 3.3.4 on each node. When we first start up the cluster, 
there are no issues and things work just fine, but after a while things start 
acting a bit haywire. It looks like we start to miss messages that should be 
coming through the system. However on client restart, logging shows that only 
the partitions on broker 2 and 3 are being re-balanced (for the particular 
topic I'm connecting to). I'm confidant that there is nothing else connected to 
the cluster (as we've seen this a couple of times now). To make it more 
interesting, if I switch topics, the client will rebalance the partitions on 
all three brokers.

Anyone know of any reason why this could occur? 

Thanks,
Karl

Reply via email to