When we exclude a host from hadoop we have to manually run below commands once puppet has run
Once done, run puppet on both Master nodes (an-master1001 and an-master1002) to update the exclude file. Then run hdfs dfsadmin -refreshNodes command for each NameNode FS URI (run these command on an-master1001 for example):

sudo -u hdfs hdfs dfsadmin -fs hdfs://an-master1001.eqiad.wmnet:8020 -refreshNodes
sudo -u hdfs hdfs dfsadmin -fs hdfs://an-master1002.eqiad.wmnet:8020 -refreshNodes

Run this on each ResourceManager host (an-master1001 and an-master1002 for example):

sudo -u hdfs yarn rmadmin -refreshNodes

We should update puppet code to automatically trigger those commands if hosts.exclude file is updated