Super helpful to know it can be done, but is this the ambari-server conf file you edit or the ambari-agent file? I assume it's ambari-server but it would be nice to have that as a clear call out in the article. ... View more

FYI: This file seems to only be present on the hive server. ... View more

hdp 2.6.3 is missing this script. Any hints to where it was moved? ... View more

This doesn't work for hdp 2.6.3 ... View more

Maybe it's clearer to call it cluster-name instead of "identity-assertion" Apache Knox https://{gateway-host}:{gateway-port}/{gateway-path}/{cluster-name}/webhdfs ... View more

Created symlink of ranger kms conf to core site and hdfs site is a vagues statement. Could you explain a little more... I know how to create a symlink, but I don't know what you mean by "Created symlink of ranger kms conf to core site and hdfs site" ... View more

I think this really depends on the workload... I'd actually consider turning up the replication, given the following conditions: Data does not change frequently but is queried repeatedly. If you aren't writing constantly to a cluster and you have extra capacity why not consider increasing the replication factor to decrease the network traffic. If you have increased locality of data by spreading it wider across the cluster this could actually reduce traffic on the network. Yes, you'd pay a higher upfront cost for writing data, but if the workload is write once, read 1000 times, you may be better off increasing the replication factor. Thoughts? I want to acknowledge in a situation where you are doing some write heavy operations your article is on point. ... View more