Kafka does not automatically migrate the the existing partition data to new volumes. Only new partitions will placed on on new volumes.
For now, you can manually copy some the partition dirs(careful with checkpoint files) to new disk, or you can increase the partitions. Or we can just delete the old disk contents and add new disk to log.dirs, kafka replicates the the data from other brokers and places on available disks. This may take time depending on data size. JIRA related to auto migration: https://issues.apache.org/ jira/browse/KAFKA-1689 On Fri, Oct 21, 2016 at 3:01 AM, Jeremy Hansen <jer...@skidrow.la> wrote: > I’ve read several posts on this but it seemed to be old versions of > kafka. I wanted to get the latest proper way to do this. > > If I add a volume to my brokers running kafka 0.10.0 and I’ve adjusted > log.dirs with the new volume mount point, what do I have to do to rebalance > so kafka takes advantage of the new volume? Also, does it make sense to > increase partitions when adding an additional volume? > > Thank you > -jeremy
