[jira] [Created] (GIRAPH-814) Incorrect MapReduce application classpath processing

["Alexandre Fonseca (JIRA)" <ji...@apache.org>]

Alexandre Fonseca created GIRAPH-814:
----------------------------------------

             Summary: Incorrect MapReduce application classpath processing
                 Key: GIRAPH-814
                 URL: https://issues.apache.org/jira/browse/GIRAPH-814
             Project: Giraph
          Issue Type: Bug
    Affects Versions: 1.1.0
            Reporter: Alexandre Fonseca
         Attachments: GIRAPH-814.patch

*Symptom:*
Yarn ApplicationManager is unable to find mapred classes if user does not override mapreduce.application.classpath in mapred-site.xml:

{code}(GiraphApplicationMaster.java:main(442)) - GiraphApplicationMaster caught a t$
java.lang.NoClassDefFoundError: org/apache/hadoop/mapreduce/lib/output/TextOutputFormat{code}

*Culprit:*
Processing of default mapreduce application classpath in addLocalClasspathToEnv(..)  YarnUtils.java:180

*Reasoning:*
YarnConfiguration.DEFAULT_YARN_APPLICATION_CLASSPATH is defined as an array of strings as per YarnConfiguration.java:832:

{code}  public static final String[] DEFAULT_YARN_APPLICATION_CLASSPATH = {
      ApplicationConstants.Environment.HADOOP_CONF_DIR.$(),
      ApplicationConstants.Environment.HADOOP_COMMON_HOME.$()
          + "/share/hadoop/common/*",
      ApplicationConstants.Environment.HADOOP_COMMON_HOME.$()
          + "/share/hadoop/common/lib/*",
      ApplicationConstants.Environment.HADOOP_HDFS_HOME.$()
          + "/share/hadoop/hdfs/*",
      ApplicationConstants.Environment.HADOOP_HDFS_HOME.$()
          + "/share/hadoop/hdfs/lib/*",
      ApplicationConstants.Environment.HADOOP_YARN_HOME.$()
          + "/share/hadoop/yarn/*",
      ApplicationConstants.Environment.HADOOP_YARN_HOME.$()
          + "/share/hadoop/yarn/lib/*" };{code}

MRJobConfig.DEFAULT_MAPREDUCE_APPLICATION_CLASSPATH is defined as a comma separated string as per MRJobConfig.java:679:

{code}  public final String 
  DEFAULT_MAPREDUCE_APPLICATION_CLASSPATH = Shell.WINDOWS ?
      "%HADOOP_MAPRED_HOME%\\share\\hadoop\\mapreduce\\*," 
      + "%HADOOP_MAPRED_HOME%\\share\\hadoop\\mapreduce\\lib\\*" :
      "$HADOOP_MAPRED_HOME/share/hadoop/mapreduce/*,"
      + "$HADOOP_MAPRED_HOME/share/hadoop/mapreduce/lib/*";{code}

However, in YarnUtils.java:190, DEFAULT_MAPREDUCE_APPLICATION_CLASSPATH is treated as if it were an array of strings just as YARN_APPLICATION_CLASSPATH some lines before. This results in an incorrect classpath if the user relies on the default setting of MAPREDUCE_APPLICATION_CLASSPATH (notice the comma between the last 2 entries that should be a colon):

{code}13/12/09 21:54:56 INFO yarn.GiraphYarnClient: Environment for AM :
{CLASSPATH=${CLASSPATH}:./*:$HADOOP_CONF_DIR:
$HADOOP_COMMON_HOME/share/hadoop/common/*:
$HADOOP_COMMON_HOME/share/hadoop/common/lib/*:
$HADOOP_HDFS_HOME/share/hadoop/hdfs/*:
$HADOOP_HDFS_HOME/share/hadoop/hdfs/lib/*:
$HADOOP_YARN_HOME/share/hadoop/yarn/*:
$HADOOP_YARN_HOME/share/hadoop/yarn/lib/*:
$HADOOP_MAPRED_HOME/share/hadoop/mapreduce/*,
$HADOOP_MAPRED_HOME/share/hadoop/mapreduce/lib/*}{code}



--
This message was sent by Atlassian JIRA
(v6.1.4#6159)