In HDP clusters, adding the hook "org.apache.atlas.hive.hook.HiveHook" to the property "hive.exec.post.hooks" causes namespace creation to timeout. The Hive database gets created for the namespace but the operation times out and does not move on.
Attached a file 'remove_hive_exec_hooks.diff', which gives an idea on how we might be able to override hive exec hooks configurations, so as to not execute the Hive Atlas hook from explore service.
I am not sure whether that is ok: if we modify the hive config to ignore the atlas hook, then we are bypassing corporate policies. Atlas is set up with a purpose: to track all access to data. If we bypass that, then we violate that purpose.
Can we instead find a way to locate the missing Atlas jars and load them in explore service?