0

I have tried to create a json_split function(UDF) using brickhouse.udf.json.JsonSplitUDF which was successful at beeline. But when I am trying to use the same function in my queries, it fails due to "unable to evaluate brickhouse.udf.json.JsonSplitUDF"! I have used the same class at Hive by adding explicitly it is working.

additional info: I have added the jars at hive.reloadable.aux.jars.path and did the reload at beeline. we are using Kerberos and Sentry authentication.

I would be glad to provide if any additional information required in order to assist me further.

I would greatly appreciate your time and earliest response.

Thanks Jagath

ehfeng
  • 3,807
  • 4
  • 33
  • 42
jagath
  • 238
  • 1
  • 6
  • Sentry will not let a Beeline session use a custom JAR, because the queries will run server-side under a privileged account (i.e. `hive`) and therefore a custom JAR might create a security breach. Whereas the legacy Hive CLI runs client-side under your own account, so you are free to run arbitrary code. – Samson Scharfrichter Dec 07 '16 at 17:44
  • But a Sentry admin may grant you the privilege to execute a specific JAR -- cf. https://www.cloudera.com/documentation/enterprise/5-8-x/topics/cm_mc_hive_udf.html under "With Sentry enabled" – Samson Scharfrichter Dec 07 '16 at 17:46

0 Answers0