Posts: 12
Registered: ‎07-05-2018
Accepted Solution


i am trying to translate the query which is in sql working on oracle.

the query goes like this,

select MAX(NANVL(Column name,0)) OVER (partition by x, y) as z from table name 

  I would like to do the same using impala-shell on kudu-table because when i performed i got the following exception

ERROR: AnalysisException: default.nanvl() unknown
Thanks in advance

Posts: 487
Topics: 14
Kudos: 82
Solutions: 43
Registered: ‎09-02-2016

Re: impala-shell



Not sure which version of impala, CDH you are using


NANVL in-build function is available in oracle but impala does not support (as of CDH 5.14 version)


Below is the link for CDH 5.14 shows all the available conditional functions, which has NVL, NVL2 but not NANVL. You may need to write your custom function to meet this part



Posts: 12
Registered: ‎07-05-2018

Re: impala-shell

thank you for the solution, i have a sql query which has group by roll up clause and it goes like following.

from tablename group by rollup(decode(-----some expression here-----),decode(some expression here),decode (some expression here))

as i said i have to modify the query where the query work on kudu using impala shell,

when i performed the query i got the follwing exception,


ERROR: AnalysisException: default.rollup() unknown


please help me with what has to be done in this case


thank you