This message was deleted.

It’s how Java apps work. You will see them using the available memory (up to Xmx) for different stuff, like caching. If you start the cluster and run a couple queries you will see a high usage as well. Tune the memory limits using Basic Cluster Tuning directives and dont mind ‘how much memory its using right now’ as it will pretty much be all available at any given time.

But for 570MB of data why its taking this much resource is confusing me. Today when we tested with around 19GB of parquet file load (multiple files) with 1+ 4 worker, after reading the data in stage1 the communication between the services gine and it got cancelled and failed automatically.