hive-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "xuhuan (JIRA)" <j...@apache.org>
Subject [jira] [Created] (HIVE-10234) Problems using custom inputformat
Date Tue, 07 Apr 2015 01:51:12 GMT
xuhuan created HIVE-10234:
-----------------------------

             Summary: Problems using custom inputformat
                 Key: HIVE-10234
                 URL: https://issues.apache.org/jira/browse/HIVE-10234
             Project: Hive
          Issue Type: Bug
          Components: File Formats
    Affects Versions: 0.13.1
         Environment: CentOS release 6.3
CDH-5.3.1-1.cdh5.3.1.p0.5
hive-0.13.1-cdh5.3.1
Inputformat: https://github.com/twitter/elephant-bird/blob/master/rcfile/src/main/java/com/twitter/elephantbird/mapreduce/input/RCFileProtobufInputFormat.java
            Reporter: xuhuan
            Priority: Critical


I apply the RCFileProtobufInputFormat from twitter's elephantbird project into hive to handle
Protobuf records which are written in a RCFile file format. What I have done is belowing:
1. compile the soure code of RCFileProtobufInputFormat and make a jar
2. add the jar to hive_aux_jar_path and update the config of server and client
3. create table using inputformat "com.twitter.elephantbird.mapreduce.input.RCFileProtobufInputFormat"
Creating table is ok.but when I excute a query like "select * from table test"(there is no
data in table test),I got these mistake: FAILED: SemanticException 1:14 Input format must
implement InputFormat. Error encountered near token test. 
I found some cases in stackoverflow,someones said that's because the custom inputformat class
implement the new API(org.apache.hadoop.marpreduce) and RCFileProtobufInputFormat class indeed
extends the new API. I'm wondering wheather a custom inputformat must extend the old API(org.apache.hadoop.marpred)?
Could this be the reason of my problem?




--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message