如何在hive中设置mapreduce任务的个数为1

Question

我试着在hive-

set hive.exec.reducers.max = 1;
set mapred.reduce.tasks = 1;
from flat_json
insert overwrite table aggr_pgm_measure PARTITION(dt='${START_TIME}')
reduce  log_time,
 req_id, ac_id, client_key, rulename, categoryname, bsid, visitorid, visitorgroupid, visitortargetid, targetpopulationid, windowsessionid, eventseq, event_code, eventstarttime
 using '${SCRIPT_LOC}/aggregator.pl' as 
 metric_id, metric_value, aggr_type, rule_name, category_name; 

尽管将max number和reduce task number设置为1，但我看到2个map reduce task生成。请见下文-

 > set hive.exec.reducers.max = 1;
hive>  set mapred.reduce.tasks = 1;
hive>
    > from flat_json
    > insert overwrite table aggr_pgm_measure PARTITION(dt='${START_TIME}')
    > reduce  log_time,
    >  req_id, ac_id, client_key, rulename, categoryname, bsid, visitorid, visitorgroupid, visitortargetid, targetpopulationid, windowsessionid, eventseq, event_code, eventstarttime
    >  using '${SCRIPT_LOC}/aggregator.pl' as
    >  metric_id, metric_value, aggr_type, rule_name, category_name;
converting to local s3://dsp-emr-test/anurag/dsp-test/60mins/script/aggregator.pl
Added resource: /mnt/var/lib/hive_07_1/downloaded_resources/aggregator.pl
Total MapReduce jobs = 2
Launching Job 1 out of 2
Number of reduce tasks is set to 0 since there's no reduce operator
Starting Job = job_201112270825_0009, Tracking URL = http://ip-10-85-66-9.ec2.internal:9100/jobdetails.jsp?jobid=job_201112270825_0009
Kill Command = /home/hadoop/.versions/0.20.205/libexec/../bin/hadoop job  -Dmapred.job.tracker=10.85.66.9:9001 -kill job_201112270825_0009
2011-12-27 10:30:03,542 Stage-1 map = 0%,  reduce = 0%

Answer 1

你认为相关的两件事其实并不相关。您正在设置reduce 任务的数量，而不是MapReduce 作业的数量。Hive将把你的查询转换成几个MapReduce作业，就像需要做的那样。每个MapReduce作业由多个map 任务和reduce 任务组成。

设置的是任务的最大数目。这意味着，每个MapReduce作业将受到它可以启动的任务数量的限制。不过，您仍然需要运行两个作业。对于Hive的MapReduce作业数量，你无能为力。为了执行查询，它需要运行每个阶段。