这也是可行的:

export LD_LIBRARY_PATH=/usr/lib/hadoop/lib/native

对于那些在OSX上通过Homebrew安装Hadoop的用户，请按照以下步骤替换路径和Hadoop版本

wget http://www.eu.apache.org/dist/hadoop/common/hadoop-2.7.1/hadoop-2.7.1-src.tar.gz
tar xvf hadoop-2.7.1-src.tar.gz
cd hadoop-2.7.1-src
mvn package -Pdist,native -DskipTests -Dtar
mv lib /usr/local/Cellar/hadoop/2.7.1/

然后更新hadoop-env.sh

export HADOOP_OPTS="$HADOOP_OPTS -Djava.net.preferIPv4Stack=true -Djava.security.krb5.realm= -Djava.security.krb5.kdc= -Djava.library.path=/usr/local/Cellar/hadoop/2.7.1/lib/native"

除了@zhutoulala接受的答案之外，这里是一个更新，使其能够在ARMHF平台(树莓派3模型B)上使用最新的稳定版本(2.8)。 首先，我可以确认您必须将本机库重新编译为64位ARM，这里基于设置一些环境变量的其他答案将不起作用。正如Hadoop文档中所指出的，预构建的本机库是32位的。

第一个链接(http://www.ercoppa.org/posts/how-to-compile-apache-hadoop-on-ubuntu-linux.html)中给出的高级步骤是正确的。 在这个网址http://www.instructables.com/id/Native-Hadoop-260-Build-on-Pi/上，你可以获得更多关于树莓派的详细信息，但不包括Hadoop 2.8版。

以下是我对Hadoop 2.8的建议:

there is still no protobuf package on latest Raspbian so you must compile it yourself and version must be exactly protobuf 2.5 (https://protobuf.googlecode.com/files/protobuf-2.5.0.tar.gz) CMake file patching method must be changed. Moreovere, files to patch are not the same. Unfortunately, there is no accepted patch on JIRA specific to 2.8. On this URL (https://issues.apache.org/jira/browse/HADOOP-9320) you must copy and paste Andreas Muttscheller proposed patch on your namenode : :hadoop-2.8.0-src/hadoop-common-project/hadoop-common $ touch HADOOP-9320-v2.8.patch :hadoop-2.8.0-src/hadoop-common-project/hadoop-common $ vim HADOOP-9320-v2.8.patch #copy and paste proposed patch given here : https://issues.apache.org/jira/browse/HADOOP-9320?focusedCommentId=16018862&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-16018862 :hadoop-2.8.0-src/hadoop-common-project/hadoop-common $ patch < HADOOP-9320-v2.8.patch patching file HadoopCommon.cmake patching file HadoopJNI.cmake :hadoop-2.8.0-src/hadoop-common-project/hadoop-common $ cd ../.. :hadoop-2.8.0-src $ sudo mvn package -Pdist,native -DskipTests -Dtar

一旦构建成功:

    :hadoop-2.8.0-src/hadoop-dist/target/hadoop-2.8.0/lib/native $ tar -cvf nativelibs.tar *

用这个归档文件的内容替换Hadoop安装的lib/native目录的内容。运行Hadoop时的警告消息应该消失。

export JAVA_HOME=/home/hadoop/software/java/jdk1.7.0_80
export HADOOP_HOME=/usr/local/hadoop
export HADOOP_COMMON_LIB_NATIVE_DIR=$HADOOP_HOME/lib/native
export HADOOP_OPTS="-Djava.library.path=$HADOOP_COMMON_LIB_NATIVE_DIR"

经过KotiI建议的持续研究，这个问题得到了解决。

hduser@ubuntu:~$ cd /usr/local/hadoop

hduser@ubuntu:/usr/local/hadoop$ ls

bin  include  libexec      logs        README.txt  share
etc  lib      LICENSE.txt  NOTICE.txt  sbin

hduser@ubuntu:/usr/local/hadoop$ cd lib

hduser@ubuntu:/usr/local/hadoop/lib$ ls
native

hduser@ubuntu:/usr/local/hadoop/lib$ cd native/

hduser@ubuntu:/usr/local/hadoop/lib/native$ ls

libhadoop.a       libhadoop.so        libhadooputils.a  libhdfs.so
libhadooppipes.a  libhadoop.so.1.0.0  libhdfs.a         libhdfs.so.0.0.0

hduser@ubuntu:/usr/local/hadoop/lib/native$ sudo mv * ../

干杯

这个答案混合了@chromeeagle的分析和这个链接(Nan-Xiao)。

对于那些其他解决方案根本不起作用的人，请遵循以下步骤:

Edit the file $HADOOP_HOME/etc/hadoop/log4j.properties (credits to @chromeeagle). Add the line at the end: log4j.logger.org.apache.hadoop.util.NativeCodeLoader=DEBUG Launch your spark/pyspark shell. You will see additional log information regarding the native library not loading. In my case I had the following error: Failed to load native-hadoop with error: java.lang.UnsatisfiedLinkError: no hadoop in java.library.path To fix this specific problem, add the Hadoop native library path to the LD_LIBRARY_PATH environment variable in your user's profile: export LD_LIBRARY_PATH="$HADOOP_HOME/lib/native:$LD_LIBRARY_PATH"

希望这能有所帮助。我在几个HADOOP安装中遇到了这个问题，它在两个上都有效。