I have a Hunk server that is connected to a HDP 2.x Yarn Cluster that is hosted on AWS EC-2. When I run searches using the virtual indexes the search returns results but never finalizes.
When a search job using a virtual index runs and a map reduce job is started, the mapreduce server opens an ephemeral port somewhere in the range of 32768 to 61000. This port is for the ipc server listener that we need to contact to get confirmation that the map reduce job completed. The fix is to open this port range within the security group (ec-2 firewall) to allow traffic from Hunk to the Hadoop cluster nodes.
When a search job using a virtual index runs and a map reduce job is started, the mapreduce server opens an ephemeral port somewhere in the range of 32768 to 61000. This port is for the ipc server listener that we need to contact to get confirmation that the map reduce job completed. The fix is to open this port range within the security group (ec-2 firewall) to allow traffic from Hunk to the Hadoop cluster nodes.