[
https://issues.apache.org/jira/browse/HIVE-18148?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16288814#comment-16288814
]
liyunzhang commented on HIVE-18148:
-----------------------------------
[~lirui]: tried to use the example you provided.
{code}
set hive.spark.dynamic.partition.pruning=true;
explain select * from src join part1 on src.key=part1.p join part2 on
src.value=part2.q;
{code}
But in my env( latest build: 095e6bf8988a03875bc9340b2ab82d5d13c4cb3c), the
physical plan before SparkCompiler#removeNestedDPP is
{code}
TS[0]-FIL[22]-SEL[2]-RS[9]-MAPJOIN[32]-MAPJOIN[31]-SEL[15]-FS[16]
TS[3]-FIL[23]-SEL[5]-MAPJOIN[32]
TS[6]-FIL[24]-SEL[8]-RS[13]-MAPJOIN[31]
{code}
there is no dpp operator for removeNestedDPP to traverse because HIVE-17087.
SparkMapJoinOptimizer#convertJoinMapJoin removes dpp operator when there is
mapjoin operator.
So did you reproduce the NPE when setting
{{hive.auto.convert.join.noconditionaltask}} as false? BTW, how does Hive on
Tez deal with this kind of nested dpp case?
> NPE in SparkDynamicPartitionPruningResolver
> -------------------------------------------
>
> Key: HIVE-18148
> URL: https://issues.apache.org/jira/browse/HIVE-18148
> Project: Hive
> Issue Type: Bug
> Components: Spark
> Reporter: Rui Li
> Assignee: Rui Li
> Attachments: HIVE-18148.1.patch
>
>
> The stack trace is:
> {noformat}
> 2017-11-27T10:32:38,752 ERROR [e6c8aab5-ddd2-461d-b185-a7597c3e7519 main]
> ql.Driver: FAILED: NullPointerException null
> java.lang.NullPointerException
> at
> org.apache.hadoop.hive.ql.optimizer.physical.SparkDynamicPartitionPruningResolver$SparkDynamicPartitionPruningDispatcher.dispatch(SparkDynamicPartitionPruningResolver.java:100)
> at
> org.apache.hadoop.hive.ql.lib.TaskGraphWalker.dispatch(TaskGraphWalker.java:111)
> at
> org.apache.hadoop.hive.ql.lib.TaskGraphWalker.walk(TaskGraphWalker.java:180)
> at
> org.apache.hadoop.hive.ql.lib.TaskGraphWalker.startWalking(TaskGraphWalker.java:125)
> at
> org.apache.hadoop.hive.ql.optimizer.physical.SparkDynamicPartitionPruningResolver.resolve(SparkDynamicPartitionPruningResolver.java:74)
> at
> org.apache.hadoop.hive.ql.parse.spark.SparkCompiler.optimizeTaskPlan(SparkCompiler.java:568)
> {noformat}
> At this stage, there shouldn't be a DPP sink whose target map work is null.
> The root cause seems to be a malformed operator tree generated by
> SplitOpTreeForDPP.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
