Hi,
Running disk balance is very bad via CC.
I'm running JBOD cluster with 10 brokers, 4 disks on each machine.
even when setting num.concurrent.intra.broker.partition.movements=1 my cluster IO goes to 100% on couple of brokers, leading to offline partitions and service degradation...
Now, i would like to ask:
Why CC doesn't support logdir movement throttle?
Why i can't run disk balance to only 1 broker at a time ?