Useful to find the missing host, when you've 100s of machines on AWS running TT and DN process and their dns names are autogenerated

hadoop_hosts_diff.sh

# This assumes that jq is installed on your machine

JT_MACHINE="jt-host"
NN_MACHINE="nn-host"

# Get all the TTs
curl "http://${JT_MACHINE}:50030/jmx?qry=hadoop:service=JobTracker,name=JobTrackerInfo" | jq -r .beans[].AliveNodesInfoJson | jq -r .[].hostname | sort > tasktrackers
# Get all the DNs
curl "http://${NN_MACHINE}:50070/jmx?qry=Hadoop:service=NameNode,name=NameNodeInfo" | jq -r .beans[].LiveNodes | jq -r 'keys | .[]' | sort > datanodes

diff tasktrackers datanodes | egrep "<|>"