Heavy data moving between historicals

Historical config:

druid.server.maxsize= 180Gb

heap size = 12 GB

direct memory = 30 GB

in coordinator console, all the datasources are available for querying (confirmed it in load status API)

the problem is, eventhough there are no new segments to load, the coordinator console keeps on showing that there are some segments to load.

As my understanding goes, this only happens when rebalancing condition occurs.

After some digging about rebalancing, i found that when a historical is being queried heavily co-ordinator will try to move the segments from that historical to some other historical. Is this the reason for my problem?

If so, the data moving costs are very high us. How can i deal with this problem?

Hi Kiran,

This is a known issue, as a workaround, you can throttle the segment movement across nodes by setting maxSegmentsToMove to a very low value. See https://github.com/apache/incubator-druid/issues/3354 for more details.