zoukankan      html  css  js  c++  java
  • hadoop使用FAQ

    1.Hive任务内存溢出

    hive执行select day_id,count(1) from ti_94014_signal_physics_yyyymmddhh   group by day_id; Mapreduce报错
    报错:
     For more detailed output, check application tracking page:http://oc-data-152:8088/cluster/app/application_1467076091284_38724Then, click on links to logs of each attempt.
    Diagnostics: Container [pid=130260,containerID=container_e18_1467076091284_38724_01_000001] is running beyond physical memory limits. Current usage: 1.0 GB of 1 GB physical memory used; 1.8 GB of 2.1 GB virtual memory used. Killing container.
    Dump of the process-tree for container_e18_1467076091284_38724_01_000001 :
    |- PID PPID PGRPID SESSID CMD_NAME USER_MODE_TIME(MILLIS) SYSTEM_TIME(MILLIS) VMEM_USAGE(BYTES) RSSMEM_USAGE(PAGES) FULL_CMD_LINE
    |- 130260 130257 130260 130260 (bash) 0 0 115847168 362 /bin/bash -c /usr/jdk64/jdk1.7.0_67/bin/java -Djava.io.tmpdir=/data/data4/hadoop/yarn/local/usercache/ynhajob/appcache/application_1467076091284_38724/container_e18_1467076091284_38724_01_000001/tmp -Dlog4j.configuration=container-log4j.properties -Dyarn.app.container.log.dir=/data/data6/hadoop/yarn/log/application_1467076091284_38724/container_e18_1467076091284_38724_01_000001 -Dyarn.app.container.log.filesize=0 -Dhadoop.root.logger=INFO,CLA -Dhadoop.root.logfile=syslog -Dhdp.version=2.3.4.7-4 -Xmx819m -Dhdp.version=2.3.4.7-4 org.apache.hadoop.mapreduce.v2.app.MRAppMaster 1>/data/data6/hadoop/yarn/log/application_1467076091284_38724
     
    判断:Hive任务内存溢出
     
    修改参数:set yarn.app.mapreduce.am.resource.mb=2048;(原配置1024)
     
    set mapreduce.map.java.opts=-Xmx1638m;
    set mapreduce.map.memory.mb=2048;
    set mapreduce.job.reduce.slowstart.completedmaps=0.90  —这三个暂时不用修改,只是相关参数可以熟悉下
     
    说明:
    select day_id,count(1) from ti_94014_signal_physics_yyyymmddhh  
    insert into table tmp_liyangtest_20160828 select day_id,count(1) from ti_94014_signal_physics_yyyymmddhh group by day_id;
     
    这两个一个直接打印出结果,一个写入临时文件;insert方式并不能节约内存
    create table tmp_liyangtest_20160828 (a string,b string)
     
     

    2.执行setfacl报错

    [root@ai185 ~]# hadoop fs -setfacl -R -m user:aaa:rw- /app/aaa
    setfacl: The ACL operation has been rejected.  Support for ACLs has been disabled by setting dfs.namenode.acls.enabled to false.
     
    判断:
    设置访问控制列表,acls默认未开启,直接使用该命令会报错
     
    解决:
    开启acls,配置hdfs-site.xml
    [hadoop@localhost hadoop-2.5.2]$ vi etc/hadoop/hdfs-site.xml
    <property>
        <name>dfs.namenode.acls.enabled</name>
        <value>true</value>
    </property>
     
    重启hdfs
     

    3.dismission datanode节点很慢

     
    加以下参数:
    dfs.namenode.replication.max-streams   200
    dfs.namenode.replication.max-streams-hard-limit 400
    dfs.namenode.replication.work.multiplier.per.iteration 100
     

    4.slider 提交yarn任务失败,报failed to renew token TIMELINE_DELEGATION_TOKEN异常

    slider所在机器的timeline.enable配置改 false就成功 
     
     
  • 相关阅读:
    Effective Java 19 Use interfaces only to define types
    Effective Java 18 Prefer interfaces to abstract classes
    Effective Java 17 Design and document for inheritance or else prohibit it
    Effective Java 16 Favor composition over inheritance
    Effective Java 15 Minimize mutability
    Effective Java 14 In public classes, use accessor methods, not public fields
    Effective Java 13 Minimize the accessibility of classes and members
    Effective Java 12 Consider implementing Comparable
    sencha touch SortableList 的使用
    sencha touch dataview 中添加 button 等复杂布局并添加监听事件
  • 原文地址:https://www.cnblogs.com/brownyangyang/p/9145736.html
Copyright © 2011-2022 走看看