关于return code X from org.apache.hadoop.hive.ql.exec.mr.MapredLocalTask

vase

浏览: 422225 次
性别:
来自: 北京

最近访客更多访客>>

sonicer

jyzbcs

xklc

paladin1988

博主相关

博客

微博

相册

留言

关于我

文章分类

社区版块

存档分类

博客分类：

Hive
Hadoop

不知道是不是hive-0.12版增强了local mode的原因，在之前版本运行好好的Hive-QL在这个版本上错误频频，折磨一天多以后终于定位到原因，把在内部的总结在这再记录下，希望对遇到同样问题的筒子们有所帮助。

部分一关于return code 1 from org.apache.hadoop.hive.ql.exec.mr.MapredLocalTask

Hive升级到0.12版之后，若干原来在0.10上执行正常的SQL会在新版上报错误 “return code 1 from org.apache.hadoop.hive.ql.exec.mr.MapredLocalTask”，查看hive执行日志，从中找到如下错误

Total MapReduce jobs = 1
java.io.IOException: Cannot run program "/data/opt/hadoop_cdh5/bin/hadoop" (in directory "/root"): error=13, 权限不够
        at java.lang.ProcessBuilder.start(ProcessBuilder.java:1041)
        at java.lang.Runtime.exec(Runtime.java:617)
        at java.lang.Runtime.exec(Runtime.java:450)
        at org.apache.hadoop.hive.ql.exec.mr.MapredLocalTask.execute(MapredLocalTask.java:253)
        at org.apache.hadoop.hive.ql.exec.Task.executeTask(Task.java:151)
        at org.apache.hadoop.hive.ql.exec.TaskRunner.runSequential(TaskRunner.java:65)
        at org.apache.hadoop.hive.ql.Driver.launchTask(Driver.java:1485)
        at org.apache.hadoop.hive.ql.Driver.execute(Driver.java:1263)
        at org.apache.hadoop.hive.ql.Driver.runInternal(Driver.java:1091)
        at org.apache.hadoop.hive.ql.Driver.run(Driver.java:931)
        at org.apache.hadoop.hive.ql.Driver.run(Driver.java:921)
        at org.apache.hadoop.hive.service.HiveServer$HiveServerHandler.execute(HiveServer.java:198)
        at org.apache.hadoop.hive.service.ThriftHive$Processor$execute.getResult(ThriftHive.java:644)
        at org.apache.hadoop.hive.service.ThriftHive$Processor$execute.getResult(ThriftHive.java:628)
        at org.apache.thrift.ProcessFunction.process(ProcessFunction.java:39)
        at org.apache.thrift.TBaseProcessor.process(TBaseProcessor.java:39)
        at org.apache.thrift.server.TThreadPoolServer$WorkerProcess.run(TThreadPoolServer.java:244)
        at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1145)
        at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:615)
        at java.lang.Thread.run(Thread.java:744)
Caused by: java.io.IOException: error=13, 权限不够
        at java.lang.UNIXProcess.forkAndExec(Native Method)
        at java.lang.UNIXProcess.<init>(UNIXProcess.java:135)
        at java.lang.ProcessImpl.start(ProcessImpl.java:130)
        at java.lang.ProcessBuilder.start(ProcessBuilder.java:1022)
        ... 19 more
FAILED: Execution Error, return code 1 from org.apache.hadoop.hive.ql.exec.mr.MapredLocalTask

从上边错误及下边报错的类MapredLocalTask可以看出跟本地任务有关

hive从0.7版以后，为了提高小数据的计算速度，增加了本地模式，即将hdfs上的数据拉到hiveserver本地进行计算，可以通过以下几个参数对相关行为进行设置

hive.exec.mode.local.auto=false
hive.exec.mode.local.auto.input.files.max=4
hive.exec.mode.local.auto.inputbytes.max=134217728

其中第一个为不启用本地模式，第二个参数表示文件数小于4时使用本地模式，第三个参数表示文件大小小于128m时采用本地模式

默认为不启用本地模式；在启用的情况下，满足第二、三个条件中的任意一个都会使用本地模式。

在之前我们用过的0.8.1、0.10版上都未遇到过上述错误，怀疑是现在0.12版本的问题突然导致上述错误。任务是在root用户下通过crontab调用的，进入shell后先启动hiveserver，所以默认工作目录其实是/root；为了能正常读写hdfs上的文件，hiveserver在启动时切换到了hdfs用户，一旦遇到上述两种满足启用本地模式的情况，hdfs用户试图向当前工作目录/root拉取数据，必然没有权限从而导致以上错误。

理清问题所在就好办了，我们可以先创建一个目录，把用户、用户组授权给hdfs，进入shell后，先切换工作目录，然后再启动hiveserver即可。如hdfs的home目录/home/hdfs

然后在任务shell的公共配置文件conf/kettle.conf中增加一行切换目录脚本即可解决以上问题

cd /home/hdfs

部分二关于return code 2 from org.apache.hadoop.hive.ql.exec.mr.MapredLocalTask

类似上一篇中return code 1的问题，这个也是跟hive本地任务有关系。

从hive的日志中可以找到出错时本地日志文件，如下：

查看日志文件内容

2014-07-10 11:50:37,606 INFO  mr.ExecDriver (SessionState.java:printInfo(417)) - Execution log at: /tmp/hdfs/hdfs_20140710114949_ab4d1d02-0637-4abd-9e45-2a27c5d740d9.log
2014-07-10 11:50:37,711 WARN  conf.Configuration (Configuration.java:loadProperty(2358)) - file:/tmp/hdfs/hive_2014-07-10_11-49-37_877_2428431256361163465-1/-local-10009/jobconf
2014-07-10 11:50:37,720 WARN  conf.Configuration (Configuration.java:loadProperty(2358)) - file:/tmp/hdfs/hive_2014-07-10_11-49-37_877_2428431256361163465-1/-local-10009/jobconf
2014-07-10 11:50:37,798 INFO  log.PerfLogger (PerfLogger.java:PerfLogBegin(97)) - <PERFLOG method=deserializePlan from=org.apache.hadoop.hive.ql.exec.Utilities>
2014-07-10 11:50:37,798 INFO  exec.Utilities (Utilities.java:deserializePlan(732)) - Deserializing MapredLocalWork via kryo
2014-07-10 11:50:38,043 INFO  log.PerfLogger (PerfLogger.java:PerfLogEnd(124)) - </PERFLOG method=deserializePlan start=1404964237798 end=1404964238043 duration=245 from=org.apa
2014-07-10 11:50:38,050 INFO  mr.MapredLocalTask (SessionState.java:printInfo(417)) - 2014-07-10 11:50:38   Starting to launch local task to process map join;  maximum memory =
2014-07-10 11:50:38,059 INFO  mr.MapredLocalTask (MapredLocalTask.java:initializeOperators(389)) - fetchoperator for t2:t_tmp_user_first_login created
2014-07-10 11:50:38,198 INFO  exec.TableScanOperator (Operator.java:initialize(338)) - Initializing Self 0 TS
2014-07-10 11:50:38,198 INFO  exec.TableScanOperator (Operator.java:initializeChildren(403)) - Operator 0 TS initialized
2014-07-10 11:50:38,199 INFO  exec.TableScanOperator (Operator.java:initializeChildren(407)) - Initializing children of 0 TS
2014-07-10 11:50:38,199 INFO  exec.SelectOperator (Operator.java:initialize(442)) - Initializing child 1 SEL
2014-07-10 11:50:38,199 INFO  exec.SelectOperator (Operator.java:initialize(338)) - Initializing Self 1 SEL
2014-07-10 11:50:38,605 ERROR mr.MapredLocalTask (MapredLocalTask.java:executeFromChildJVM(324)) - Hive Runtime Error: Map local work failed
java.lang.RuntimeException: java.lang.ClassNotFoundException: com.renren.hive.date.GetWeekISO
    at org.apache.hadoop.hive.ql.udf.generic.GenericUDFBridge.getUdfClass(GenericUDFBridge.java:132)
    at org.apache.hadoop.hive.ql.exec.FunctionRegistry.isStateful(FunctionRegistry.java:1474)
    at org.apache.hadoop.hive.ql.exec.FunctionRegistry.isDeterministic(FunctionRegistry.java:1437)
    at org.apache.hadoop.hive.ql.exec.ExprNodeGenericFuncEvaluator.isDeterministic(ExprNodeGenericFuncEvaluator.java:132)
    at org.apache.hadoop.hive.ql.exec.ExprNodeEvaluatorFactory.iterate(ExprNodeEvaluatorFactory.java:83)
    at org.apache.hadoop.hive.ql.exec.ExprNodeEvaluatorFactory.toCachedEval(ExprNodeEvaluatorFactory.java:73)
    at org.apache.hadoop.hive.ql.exec.SelectOperator.initializeOp(SelectOperator.java:59)
    at org.apache.hadoop.hive.ql.exec.Operator.initialize(Operator.java:377)
    at org.apache.hadoop.hive.ql.exec.Operator.initialize(Operator.java:453)
    at org.apache.hadoop.hive.ql.exec.Operator.initializeChildren(Operator.java:409)
    at org.apache.hadoop.hive.ql.exec.TableScanOperator.initializeOp(TableScanOperator.java:188)
    at org.apache.hadoop.hive.ql.exec.Operator.initialize(Operator.java:377)
    at org.apache.hadoop.hive.ql.exec.mr.MapredLocalTask.initializeOperators(MapredLocalTask.java:408)
    at org.apache.hadoop.hive.ql.exec.mr.MapredLocalTask.executeFromChildJVM(MapredLocalTask.java:302)
    at org.apache.hadoop.hive.ql.exec.mr.ExecDriver.main(ExecDriver.java:728)
    at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
    at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:57)
    at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
    at java.lang.reflect.Method.invoke(Method.java:606)
    at org.apache.hadoop.util.RunJar.main(RunJar.java:212)
Caused by: java.lang.ClassNotFoundException: com.renren.hive.date.GetWeekISO
    at java.net.URLClassLoader$1.run(URLClassLoader.java:366)
    at java.net.URLClassLoader$1.run(URLClassLoader.java:355)
    at java.security.AccessController.doPrivileged(Native Method)
    at java.net.URLClassLoader.findClass(URLClassLoader.java:354)
    at java.lang.ClassLoader.loadClass(ClassLoader.java:425)
    at java.lang.ClassLoader.loadClass(ClassLoader.java:358)
    at java.lang.Class.forName0(Native Method)
    at java.lang.Class.forName(Class.java:270)
    at org.apache.hadoop.hive.ql.udf.generic.GenericUDFBridge.getUdfClass(GenericUDFBridge.java:130)

由上可知，这次是找不到UDF的类（如遇到其他情况，需要具体问题具体分析），虽然在进入hive的时候通过add jar语句将自定义函数的jar包添加到hadoop集群，但在本地模式时确找不到了。定位到问题就好解决了：既然是local模式找不到udf jar包，说明在add jar步骤只是向当前job在hdfs上的工作目录下添加了，无视本地工作目录；那么我们就直接把udf的jar包copy到hive的lib目录下，测试正常。

该问题在之前用过的hive 0.10、0.8.1中都未遇到过，初步猜测跟0.12版的bug有关，具体原因就需要花时间翻代码对照前后版本的变动了

从org.apache.hadoop.hive.ql.exec.mr.MapredLocalTask代码中看，还有return code 3的情况，现在幸运的尚未遇到，遇到后再补记录