Thousand of CLOSE_WAIT connections to HDFS Datanode

In IPv6 environments after a while drill has a lot of connections to HDFS Datanode in CLOSE_WAIT state.

To fix this we add '-Djava.net.preferIPv4Stack=true -Djava.net.preferIPv6Addresses=false' to our drill configuration and the problem vanished.

we saw https://community.cloudera.com/t5/Support-Questions/Over-40000-CLOSE-WAIT-after-hbase-cluster-restart/m-p/6403/highlight/true#M51847 from cloudera that helped.