Middle manager insufficient memory error.

I believe that my configuration is already wrong but I would like to confirm with the community to understand the matter better. I have 4 middle managers (r3.2xlarge) running with a total of 28 slots. I’m using 17 slots. However, some of the tasks failed with this error.

OpenJDK 64-Bit Server VM warning: INFO: os::commit_memory(0x00007fd5140bf000, 506040320, 0) failed; error=‘Cannot allocate memory’ (errno=12)

This is an insane amount of memory to give peons:
druid.indexer.runner.javaOpts=-server -Xmx30G -XX:+UseG1GC -XX:MaxGCPauseMillis=100 -XX:+PrintGCDetails -XX:+PrintGCTimeStamps
druid.indexer.runner.maxZnodeBytes=524288

You don’t have enough available memory to support 28 peons each requiring 30G of heap.

Please read http://druid.io/docs/0.9.0/operations/performance-faq.html#what-should-i-set-my-jvm-heap

Also:
http://druid.io/docs/0.9.0/tutorials/cluster.html#tune-druid-processes-that-serve-queries

Thanks for confirming this!