[
https://issues.apache.org/jira/browse/HIVE-2118?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Ning Zhang updated HIVE-2118:
-----------------------------
Description: When hive.mapred.mode=nonstrict is set, Hive queries doesn't
need to have a predicate on partition column. This means the partition
expression in the PartitionPruner could be null or it could be null after
removing predicates on non-partition columns. We need to handle those cases
correctly in terms of whether the resulting partition should be inserted into
the "true_parts" or "unknown_parts". The differences is in that if it is
true_parts then "select *" queries will be translated to a non_MR job (a fetch
task), otherwise it will be translated to a MR job.
> Partition Pruning bug in the case of hive.mapred.mode=nonstrict
> ---------------------------------------------------------------
>
> Key: HIVE-2118
> URL: https://issues.apache.org/jira/browse/HIVE-2118
> Project: Hive
> Issue Type: Sub-task
> Reporter: Ning Zhang
> Assignee: Ning Zhang
>
> When hive.mapred.mode=nonstrict is set, Hive queries doesn't need to have a
> predicate on partition column. This means the partition expression in the
> PartitionPruner could be null or it could be null after removing predicates
> on non-partition columns. We need to handle those cases correctly in terms of
> whether the resulting partition should be inserted into the "true_parts" or
> "unknown_parts". The differences is in that if it is true_parts then "select
> *" queries will be translated to a non_MR job (a fetch task), otherwise it
> will be translated to a MR job.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
