Idle Historicals

Hi devs,
I am observing strange Druid behavior now.
I had small test cluster with 4 Historical nodes and 2TB dataset. It was fully loaded to historicals. I've provisioned 21 new machines with different configuration (more CPU and RAM, less SSD), added them to cluster.
Then previous 4 nodes were shut down.
As expected, data is being downloaded from deep storage to new Historical nodes (it is still in the process), but only 11 machines are used, other 10 are sitting idle.

Am I missing something in configuration?

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@druid.apache.org
For additional commands, e-mail: dev-help@druid.apache.org

Re: Idle Historicals

The problem is gone, when I reduced druid.server.maxSize on each Historical. So, 11 nodes was not enough to keep all the data, and now it is evenly distributed between 21 nodes.

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@druid.apache.org
For additional commands, e-mail: dev-help@druid.apache.org