Search code examples
hadoophortonworks-data-platformhortonworks-sandboxapache-ranger

apache ranger with hive plugin - what objects need to be persisted


I have installed ranger and the ranger hive plugin, created some policies to enable authorisation of hive objects to certain users. Its working well. However if my machine where ranger is installed (an EC2) goes down then I assume I will need to recreate policies? I have a MySQL RDS (multi az) with regular snapshots so the db data should be available but what else do I need to backup (ie to s3 and then restore to new EC2 if initial EC2 dies)? I assume some json files under /etc/ranger/. Anything else?

env

  • apache ranger 1.0
  • hive 2.1.1
  • hadoop 2.8.3

Note: not using hortonworks


Solution

  • If you want that your ranger server should be always up, deploy your admin/user server behind the ELB or using service discovery.

    Also, mysql replication is enough to populate the entire same workspace.

    /etc/ranger/{CLUSTER_NAME}/policycache/{.json} -- this contains the authorisation policies which will be applicable, even if your ranger server goes down.