It seems this should help (coming in 1.1 release): https://cwiki.apache.org/confluence/display/KAFKA/KIP-113%3A+Support+replicas+movement+between+log+directories
On Wed, Dec 13, 2017 at 11:37 PM, ajay chaudhary < juniora...@yahoo.com.invalid> wrote: > Hi Team, > This is Ajay working with Yodlee India. We are trying to setup Kafka > cluster for streaming and we need some clarifications. > Could you please help us to understand this behavior. > How do we add a capacity to existing Kafka cluster. Let us assume, we have > Kafka cluster with 3 brokers and all brokers has single mount point > allocated(log.dirs). we create a topic with some partitions and after some > days mount point becomes 100% full may be due to unexpected data growth. > how do we add space to Kafka cluster so that existing partitions (say > partition 0 where it originally resides is full) make use of newly > provisioned space? > how would new messages be handled coming to this partition? Please help us > understand this? > Regards,Ajay chaudhary >