[HIVE-2869] Merging small files throws RuntimeException when hive.mergejob.maponly=false

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 0.8.0
• Component/s: Query Processor

Environment

CentOS release 5.5 (Final)

• Assignee: Unassigned
• Reporter: Shrijeet Paliwal
• Created: 13/Mar/12 22:43
• Updated: 03/Apr/12 22:03

Description

Hive Version: Hive 0.8 (last commit SHA b581a6192b8d4c544092679d05f45b2e50d42b45 )
Hadoop version : chd3u0

Trying to use the hive merge small file feature by setting all the necessary params.
Have disabled use of CombineHiveInputFormat since my input is compressed text.

hive> set mapred.min.split.size.per.node=1000000000;
hive> set mapred.min.split.size.per.rack=1000000000;
hive> set mapred.max.split.size=1000000000;
hive> set hive.merge.size.per.task=1000000000;
hive> set hive.merge.smallfiles.avgsize=1000000000;
hive> set hive.merge.size.smallfiles.avgsize=1000000000;
hive> set hive.merge.mapfiles=true;
hive> set hive.merge.mapredfiles=true;
hive> set hive.mergejob.maponly=false;

The plan decides to launch two MR jobs but after first job succeeds I get runt time error
"java.lang.RuntimeException: Plan invalid, Reason: Reducers == 0 but reduce operator specified"

How to reproduce :

• Creare tables as follows :

  --create input table
  create table tmp_notmerged (
    id                int,
    name              string
  )
  ROW FORMAT DELIMITED FIELDS TERMINATED BY '\t'
  STORED AS TEXTFILE;
  
  
  --create o/p table
  create table tmp_merged (
    id                int
  )
  STORED AS TEXTFILE;
  

• Load data into tmp_notmerged (find files attached in with this jira)

• set knobs and fire hive query

  set hive.merge.mapfiles=true;
  set hive.mergejob.maponly=false;
  insert overwrite table tmp_merged select id from tmp_notmerged;
  

• You should see error "java.lang.RuntimeException: Plan invalid, Reason: Reducers == 0 but reduce operator specified"

Proposed fix :

Patch is here : https://gist.github.com/2025303