hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Yi Liang (JIRA)" <j...@apache.org>
Subject [jira] [Comment Edited] (HBASE-18175) Add hbase-spark integration test into hbase-it
Date Mon, 19 Jun 2017 04:29:01 GMT

    [ https://issues.apache.org/jira/browse/HBASE-18175?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16053466#comment-16053466
] 

Yi Liang edited comment on HBASE-18175 at 6/19/17 4:28 AM:
-----------------------------------------------------------

Hi Sean,
   In the new patch, I add a submodule called hbase-spark-it under hbase-spark module, the
parent of hbase-spark-it is hbase-spark, not hbase, this can make hbase-spark module more
independent. 
Now the hbase-spark looks like

---hbase-spark
-----------src
-----------pom
-----------hbase-spark-it
---------------src
---------------pom


And I also have some thoughts on HBase-16179,  only solve that jira then we can push hbase-spark
into hbase-2.0


And I saw your comments in that jira, there might be some requirement like, we need to consider
spark-1.6, spark-2.0, and also scala 2.10, scala 2.11

I remember I try both the scala 2.10 and 2.11 for hbase-spark, it seems that there is no code
change for scala version change; the code differences are only in different spark versions.

I think we can have hbase-spark module structure as below

- hbase-spark
--------hbase-spark-1.6
------------src (diff code for 1.6)
------------pom
--------hbase-spark-2.0
--------hbase-spark-it
------------src 
------------pom

And user can change scala version on their own for both spark1.6 and spark2.0, What do you
think of this idea?
Thanks  


was (Author: easyliangjob):
Hi Sean,
   In the new patch, I add a submodule called hbase-spark-it under hbase-spark module, the
parent of hbase-spark-it is hbase-spark, not hbase, this can make hbase-spark module more
independent. 
Now the hbase-spark looks like

    - hbase-spark
           -hbase-spark-it
                -src
                -pom
          -src
          -pom


And I also have some thoughts on HBase-16179,  only solve that jira then we can push hbase-spark
into hbase-2.0


And I saw your comments in that jira, there might be some requirement like, we need to consider
spark-1.6, spark-2.0, and also scala 2.10, scala 2.11

I remember I try both the scala 2.10 and 2.11 for hbase-spark, it seems that there is no code
change for scala version change; the code differences are only in different spark versions.

I think we can have hbase-spark module structure as below

- hbase-spark
       -hbase-spark-1.6
             -src (diff code for 1.6)
             -pom
       -hbase-spark-2.0
       -hbase-spark-it
             -src 
             -pom

And user can change scala version on their own for both spark1.6 and spark2.0, What do you
think of this idea?
Thanks  

> Add hbase-spark integration test into hbase-it
> ----------------------------------------------
>
>                 Key: HBASE-18175
>                 URL: https://issues.apache.org/jira/browse/HBASE-18175
>             Project: HBase
>          Issue Type: Test
>          Components: spark
>            Reporter: Yi Liang
>            Assignee: Yi Liang
>            Priority: Critical
>             Fix For: 2.0.0
>
>         Attachments: hbase-18175-v1.patch
>
>
> After HBASE-17574, all test under hbase-spark are regarded as unit test, and this jira
will add integration test of hbase-spark into hbase-it.  This patch run same tests as mapreduce.IntegrationTestBulkLoad,
just change mapreduce to spark.  
> test in Maven:
> mvn verify -Dit.test=IntegrationTestSparkBulkLoad
> test on cluster:
> spark-submit --class org.apache.hadoop.hbase.spark.IntegrationTestSparkBulkLoad HBASE_HOME/lib/hbase-it-2.0.0-SNAPSHOT-tests.jar
-Dhbase.spark.bulkload.chainlength=500000 -m slowDeterministic



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

Mime
View raw message