hive-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Ashutosh Chauhan <hashut...@apache.org>
Subject Re: Review Request 56929: HIVE-12492
Date Thu, 23 Feb 2017 02:13:27 GMT

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/56929/#review166439
-----------------------------------------------------------




ql/src/java/org/apache/hadoop/hive/ql/optimizer/ConvertJoinMapJoin.java (line 523)
<https://reviews.apache.org/r/56929/#comment238430>

    Good to add some javadoc here for args:
    buckets : bucket count for BMJ conversion consideration or reduce count for DHJ conversion
consideration.
    maxSize: size threshold for map join conversion
    checkHashTableEntries: threshold for distinct keys in hash table for MJ. Ignored while
called from convertToDHJ.



ql/src/java/org/apache/hadoop/hive/ql/optimizer/ConvertJoinMapJoin.java (line 984)
<https://reviews.apache.org/r/56929/#comment238389>

    max < 1 will be more robust.



ql/src/java/org/apache/hadoop/hive/ql/optimizer/ConvertJoinMapJoin.java (line 996)
<https://reviews.apache.org/r/56929/#comment238390>

    LOG.debug()



ql/src/java/org/apache/hadoop/hive/ql/optimizer/ConvertJoinMapJoin.java (line 1003)
<https://reviews.apache.org/r/56929/#comment238392>

    LOG.debug



ql/src/java/org/apache/hadoop/hive/ql/optimizer/ConvertJoinMapJoin.java (line 1006)
<https://reviews.apache.org/r/56929/#comment238446>

    LOG.debug (Table small enough for consideration of broadcast edge but we estimate too
many distinct keys in its HT, so not converting to MJ)



ql/src/java/org/apache/hadoop/hive/ql/optimizer/stats/annotation/StatsRulesProcFactory.java
(line 1416)
<https://reviews.apache.org/r/56929/#comment238431>

    return null;



ql/src/test/queries/clientpositive/tez_dynpart_hashjoin_1.q (lines 55 - 105)
<https://reviews.apache.org/r/56929/#comment238445>

    I think these changes are redundant. Lets keep these tests as is. I was just trying to
give a reference on how plans look like for DHJ.



ql/src/test/results/clientpositive/llap/join_max_hashtable.q.out (line 18)
<https://reviews.apache.org/r/56929/#comment238439>

    This should have been 
    Reducer 3 <- Map 1 (CUSTOM_SIMPLE_EDGE), Map 2 (SIMPLE_EDGE) or 
    Reducer 3 <- Map 1 (CUSTOM_SIMPLE_EDGE), Map 2 (CUSTOM_SIMPLE_EDGE) 
    if logic added in patch kicks in. 
    Currently this looks like a regular map join.



ql/src/test/results/clientpositive/llap/join_max_hashtable.q.out (line 95)
<https://reviews.apache.org/r/56929/#comment238442>

    This should have been 
    Reducer 3 <- Map 1 (CUSTOM_SIMPLE_EDGE), Map 2 (SIMPLE_EDGE) or 
    Reducer 3 <- Map 1 (CUSTOM_SIMPLE_EDGE), Map 2 (CUSTOM_SIMPLE_EDGE) 
    if logic added in patch kicks in. 
    Currently this looks like a regular map join.



ql/src/test/results/clientpositive/llap/join_max_hashtable.q.out (line 172)
<https://reviews.apache.org/r/56929/#comment238441>

    This should have been 
    Reducer 3 <- Map 1 (CUSTOM_SIMPLE_EDGE), Map 2 (SIMPLE_EDGE) or 
    Reducer 3 <- Map 1 (CUSTOM_SIMPLE_EDGE), Map 2 (CUSTOM_SIMPLE_EDGE) 
    if logic added in patch kicks in. 
    Currently this looks like a regular map join.


- Ashutosh Chauhan


On Feb. 22, 2017, 3:47 p.m., Jesús Camacho Rodríguez wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/56929/
> -----------------------------------------------------------
> 
> (Updated Feb. 22, 2017, 3:47 p.m.)
> 
> 
> Review request for hive and Ashutosh Chauhan.
> 
> 
> Bugs: HIVE-12492
>     https://issues.apache.org/jira/browse/HIVE-12492
> 
> 
> Repository: hive-git
> 
> 
> Description
> -------
> 
> HIVE-12492
> 
> 
> Diffs
> -----
> 
>   common/src/java/org/apache/hadoop/hive/conf/HiveConf.java 3777fa96ba8eac79e07e454ee437fb05583158c5

>   itests/src/test/resources/testconfiguration.properties 4a69bcc9d4147950dc38955353575ecd01e781d8

>   ql/src/java/org/apache/hadoop/hive/ql/optimizer/ConvertJoinMapJoin.java 93e36316770bb5a2d1d60e1317fbddb1eec9c4a2

>   ql/src/java/org/apache/hadoop/hive/ql/optimizer/stats/annotation/StatsRulesProcFactory.java
61f13749a86473a2dc30eeb5ca86142e6c6599f4 
>   ql/src/java/org/apache/hadoop/hive/ql/stats/StatsUtils.java 0da7ea434e0d75c1d07044ec5dfe7de673e22768

>   ql/src/test/queries/clientpositive/join_max_hashtable.q PRE-CREATION 
>   ql/src/test/queries/clientpositive/tez_dynpart_hashjoin_1.q aa331f20096a0a63144bb164aafbccf0edce4625

>   ql/src/test/results/clientpositive/llap/join_max_hashtable.q.out PRE-CREATION 
>   ql/src/test/results/clientpositive/llap/tez_dynpart_hashjoin_1.q.out 33350a75ed4b9a6fc4ab34110d84f44161d980ec

> 
> Diff: https://reviews.apache.org/r/56929/diff/
> 
> 
> Testing
> -------
> 
> 
> Thanks,
> 
> Jesús Camacho Rodríguez
> 
>


Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message