在工作过程中,经常需要调整一些hadoop的参数配置,所以经常会遇到各种各样的问题。比如改了个配置怎么突然namenode起不来啦,加了个jar包怎么让hadoop的jvm加载啊,如何设定log目录啦等等,每次都需要仔细的查一遍启动脚本才能找到原因,费时又费力,因此专门总结了一下以便不时之需。
cloudera的hadoop的启动脚本写的异常复杂和零散,各种shell脚本分散在系统的各个角落,让人很无语。下面以namenode启动的过程为例说明hadoop的启动脚本的调用关系和各个脚本的作用。
hadoop启动的入口脚本是/etc/init.d/hadoop-hdfs-name,下面我们顺着启动namenode的顺序看看hadoop的启动调用过程。
/etc/init.d/hadoop-hdfs-namenode:
#1.加载/etc/default/hadoop /etc/default/hadoop-hdfs-namenode
#2.执行/usr/lib/hadoop/sbin/hadoop-daemon.sh启动namenode
cloudera启动namenode的用户为hdfs,默认的配置目录是/etc/hadoop/conf
start() { [ -x $EXEC_PATH ] || exit $ERROR_PROGRAM_NOT_INSTALLED [ -d $CONF_DIR ] || exit $ERROR_PROGRAM_NOT_CONFIGURED log_success_msg "Starting ${DESC}: " su -s /bin/bash $SVC_USER -c "$EXEC_PATH --config '$CONF_DIR' start $DAEMON_FLAGS" # Some processes are slow to start sleep $SLEEP_TIME checkstatusofproc RETVAL=$? [ $RETVAL -eq $RETVAL_SUCCESS ] && touch $LOCKFILE return $RETVAL }
/etc/default/hadoop /etc/default/hadoop-hdfs-namenode:
#1.配置logdir,piddir,user
/usr/lib/hadoop/sbin/hadoop-daemon.sh
#1.加载/usr/lib/hadoop/libexec/hadoop-config.sh
DEFAULT_LIBEXEC_DIR="$bin"/../libexec HADOOP_LIBEXEC_DIR=${HADOOP_LIBEXEC_DIR:-$DEFAULT_LIBEXEC_DIR} . $HADOOP_LIBEXEC_DIR/hadoop-config.sh
#2.加载hadoop-env.sh
if [ -f "${HADOOP_CONF_DIR}/hadoop-env.sh" ]; then . "${HADOOP_CONF_DIR}/hadoop-env.sh" fi
#3.指定log目录
# get log directory if [ "$HADOOP_LOG_DIR" = "" ]; then export HADOOP_LOG_DIR="$HADOOP_PREFIX/logs" fi
#4.补全log目录和log4j的logger等参数
export HADOOP_LOGFILE=hadoop-$HADOOP_IDENT_STRING-$command-$HOSTNAME.log export HADOOP_ROOT_LOGGER=${HADOOP_ROOT_LOGGER:-"INFO,RFA"} export HADOOP_SECURITY_LOGGER=${HADOOP_SECURITY_LOGGER:-"INFO,RFAS"} export HDFS_AUDIT_LOGGER=${HDFS_AUDIT_LOGGER:-"INFO,NullAppender"} log=$HADOOP_LOG_DIR/hadoop-$HADOOP_IDENT_STRING-$command-$HOSTNAME.out pid=$HADOOP_PID_DIR/hadoop-$HADOOP_IDENT_STRING-$command.pid HADOOP_STOP_TIMEOUT=${HADOOP_STOP_TIMEOUT:-5}
#5.调用/usr/lib/hadoop-hdfs/bin/hdfs
hadoop_rotate_log $log echo starting $command, logging to $log cd "$HADOOP_PREFIX" case $command in namenode|secondarynamenode|datanode|journalnode|dfs|dfsadmin|fsck|balancer|zkfc) if [ -z "$HADOOP_HDFS_HOME" ]; then hdfsScript="$HADOOP_PREFIX"/bin/hdfs else hdfsScript="$HADOOP_HDFS_HOME"/bin/hdfs fi nohup nice -n $HADOOP_NICENESS $hdfsScript --config $HADOOP_CONF_DIR $command "$@" > "$log" 2>&1 < /dev/null & ;; (*) nohup nice -n $HADOOP_NICENESS $hadoopScript --config $HADOOP_CONF_DIR $command "$@" > "$log" 2>&1 < /dev/null & ;; esac echo $! > $pid sleep 1; head "$log" sleep 3; if ! ps -p $! > /dev/null ; then exit 1 fi
可以看到namenode的sysout输出到$log中,即log=$HADOOP_LOG_DIR/hadoop-$HADOOP_IDENT_STRING-$command-$HOSTNAME.out
/usr/lib/hadoop/libexec/hadoop-config.sh
#1.加载/usr/lib/hadoop/libexec/hadoop-layout.sh
hadoop-layout.sh主要描述了hadoop的lib的文件夹结构,主要内容如下
HADOOP_COMMON_DIR="./" HADOOP_COMMON_LIB_JARS_DIR="lib" HADOOP_COMMON_LIB_NATIVE_DIR="lib/native" HDFS_DIR="./" HDFS_LIB_JARS_DIR="lib" YARN_DIR="./" YARN_LIB_JARS_DIR="lib" MAPRED_DIR="./" MAPRED_LIB_JARS_DIR="lib" HADOOP_LIBEXEC_DIR=${HADOOP_LIBEXEC_DIR:-"/usr/lib/hadoop/libexec"} HADOOP_CONF_DIR=${HADOOP_CONF_DIR:-"/etc/hadoop/conf"} HADOOP_COMMON_HOME=${HADOOP_COMMON_HOME:-"/usr/lib/hadoop"} HADOOP_HDFS_HOME=${HADOOP_HDFS_HOME:-"/usr/lib/hadoop-hdfs"} HADOOP_MAPRED_HOME=${HADOOP_MAPRED_HOME:-"/usr/lib/hadoop-0.20-mapreduce"} YARN_HOME=${YARN_HOME:-"/usr/lib/hadoop-yarn"}
#2.指定HDFS和YARN的lib
HADOOP_COMMON_DIR=${HADOOP_COMMON_DIR:-"share/hadoop/common"} HADOOP_COMMON_LIB_JARS_DIR=${HADOOP_COMMON_LIB_JARS_DIR:-"share/hadoop/common/lib"} HADOOP_COMMON_LIB_NATIVE_DIR=${HADOOP_COMMON_LIB_NATIVE_DIR:-"lib/native"} HDFS_DIR=${HDFS_DIR:-"share/hadoop/hdfs"} HDFS_LIB_JARS_DIR=${HDFS_LIB_JARS_DIR:-"share/hadoop/hdfs/lib"} YARN_DIR=${YARN_DIR:-"share/hadoop/yarn"} YARN_LIB_JARS_DIR=${YARN_LIB_JARS_DIR:-"share/hadoop/yarn/lib"} MAPRED_DIR=${MAPRED_DIR:-"share/hadoop/mapreduce"} MAPRED_LIB_JARS_DIR=${MAPRED_LIB_JARS_DIR:-"share/hadoop/mapreduce/lib"} # the root of the Hadoop installation # See HADOOP-6255 for directory structure layout HADOOP_DEFAULT_PREFIX=$(cd -P -- "$common_bin"/.. && pwd -P) HADOOP_PREFIX=${HADOOP_PREFIX:-$HADOOP_DEFAULT_PREFIX} export HADOOP_PREFIX
#3.对slave文件判断。但cdh的hadoop不是依靠slave来启动集群的,而是要用户自己写集群启动脚本(也许是为了逼用户用他的CloudManager。。。)
#4.再次指定env文件
if [ -f "${HADOOP_CONF_DIR}/hadoop-env.sh" ]; then . "${HADOOP_CONF_DIR}/hadoop-env.sh" fi
#5.指定java home
# Attempt to set JAVA_HOME if it is not set if [[ -z $JAVA_HOME ]]; then # On OSX use java_home (or /Library for older versions) if [ "Darwin" == "$(uname -s)" ]; then if [ -x /usr/libexec/java_home ]; then export JAVA_HOME=($(/usr/libexec/java_home)) else export JAVA_HOME=(/Library/Java/Home) fi fi # Bail if we did not detect it if [[ -z $JAVA_HOME ]]; then echo "Error: JAVA_HOME is not set and could not be found." 1>&2 exit 1 fi fi
#6.指定Java程序启动的heapsize,如果用户在hadoop-env.sh中指定了HADOOP_HEAPSIZE字段则会覆盖默认值1000m
# some Java parameters JAVA_HEAP_MAX=-Xmx1000m # check envvars which might override default args if [ "$HADOOP_HEAPSIZE" != "" ]; then #echo "run with heapsize $HADOOP_HEAPSIZE" JAVA_HEAP_MAX="-Xmx""$HADOOP_HEAPSIZE""m" #echo $JAVA_HEAP_MAX fi
#7.指定程序的classpath,一大串代码,总结下就是
HADOOP_CONF_DIR+HADOOP_CLASSPATH+HADOOP_COMMON_DIR+HADOOP_COMMON_LIB_JARS_DIR+
HADOOP_COMMON_LIB_JARS_DIR+HADOOP_COMMON_LIB_NATIVE_DIR+HDFS_DIR+HDFS_LIB_JARS_DIR
+YARN_DIR+YARN_LIB_JARS_DIR+MAPRED_DIR+MAPRED_LIB_JARS_DIR
有一个要注意的,hadoop比较贴心的提供了HADOOP_USER_CLASSPATH_FIRST属性,如何设置了,
则HADOOP_CLASSPATH(用户自定义classpath)会在hadoop自身的jar包前加载,用来解决用户
想最先加载自定义的jar包情况。
#8.指定HADOOP_OPTS,-Dhadoop.log.dir这些类似参数会在conf下的log4j配置中用到
HADOOP_OPTS="$HADOOP_OPTS -Dhadoop.log.dir=$HADOOP_LOG_DIR"
HADOOP_OPTS="$HADOOP_OPTS -Dhadoop.log.file=$HADOOP_LOGFILE"
HADOOP_OPTS="$HADOOP_OPTS -Dhadoop.home.dir=$HADOOP_PREFIX"
HADOOP_OPTS="$HADOOP_OPTS -Dhadoop.id.str=$HADOOP_IDENT_STRING"
HADOOP_OPTS="$HADOOP_OPTS -Dhadoop.root.logger=${HADOOP_ROOT_LOGGER:-INFO,console}"
if [ "x$JAVA_LIBRARY_PATH" != "x" ]; then
HADOOP_OPTS="$HADOOP_OPTS -Djava.library.path=$JAVA_LIBRARY_PATH"
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$JAVA_LIBRARY_PATH
fi
HADOOP_OPTS="$HADOOP_OPTS -Dhadoop.policy.file=$HADOOP_POLICYFILE"
# Disable ipv6 as it can cause issues
HADOOP_OPTS="$HADOOP_OPTS -Djava.net.preferIPv4Stack=true"
/usr/lib/hadoop-hdfs/bin/hdfs
#1.加载/usr/lib/hadoop/libexec/hdfs-config.sh,但好像没啥作用
#2.根据启动参数指定java的启动mainclass:
if [ "$COMMAND" = "namenode" ] ; then CLASS='org.apache.hadoop.hdfs.server.namenode.NameNode' HADOOP_OPTS="$HADOOP_OPTS $HADOOP_NAMENODE_OPTS"
#3.启动Java程序 exec "$JAVA" -Dproc_$COMMAND $JAVA_HEAP_MAX $HADOOP_OPTS $CLASS "$@"
最后介绍几个配置的小例子。
1.如何指定hadoop的log目录:
从启动脚本中看几个配置的优先级排序是hadoop-env.sh>hadoop-config.sh>/etc/default/hadoop,因此我们如果想指定hadoop的log目录只需在hadoop-env.sh中添加一行:
export HADOOP_LOG_DIR=xxxxx
2.如何添加自己的jar包到hadoop中被namenode,datanode使用
export HADOOP_CLASSPATH=xxxxx
3.如何单独设定namenode的java heapsize。
比如想设置namenode10G,datanode1G,这个就有点意思了。如果直接指定HADOOP_HEAPSIZE那么此参数会作用于namenode,datanode,而单独在namenode的参数中指定也会有点小问题哦,不过基本是可以使用的。
总之,由于hadoop的启动脚本极其多而且琐碎,再加上hbase hive的启动脚本都是类似的结构,导致在添加修改一些配置时会产生很多莫名的问题,大家也可以在使用的过程中细细体会啦