none
Issue filtering string columns in Cloudera Hive tables

    General discussion

  • Hi,

    we're using Cloudera 4.3 with Hive 0.11 on a six node Hadoop Cluster.
    PowerPivot 64bit and ODBC Connector for Hive is installed.

    When trying to apply a filter on a string column (like a date field) in a table in database "myschema", error message

    ERROR [HY000] [Cloudera][Hardy] (22) Error from ThriftHiveClient: Query returned non-zero code: 40000, cause: FAILED: ParseException line 1:9 cannot recognize input near '[' 'myschema' ']' in select clause

    is displayed.

    So the statement looks like this:

    SELECT  [myschema].mycolumn1, ... from [myschema].mytable
    Filtering on a integer column works properly.

    Using other ODBC Connectors (MapR, Hortonworks, ...) had no positive effect.

    Switching to Native Query in ODBC Connector shows no improvement.

    Other tools like Tableau, Qlikview worked properly.

    So the problem is the query creation in PowerPivot. The square brackets seem to be at least part of the issue.

    Without the capability to apply filters on any not-integer column usability of PowerPivot with Hive is drastically reduced.

    A fix of this issue is highly appreciated.

    Best regards

    Zap


    Tuesday, October 22, 2013 11:24 AM

All replies