大数据学习之路之Hadoop

43次阅读

共计 15918 个字符，预计需要花费 40 分钟才能阅读完成。

Hadoop 介绍

一、简介

Hadoop 是一个开源的分布式计算平台，用于存储大数据，并使用 MapReduce 来处理。Hadoop 擅长于存储各种格式的庞大的数据，任意的格式甚至非结构化的处理。两个核心：

  • HDFS：Hadoop 分布式文件系统（Hadoop Distributed File System），具有高容错性和伸缩性，使用 java 开发
  • MapReduce：Google MapReduce 的开源实现，分布式编程模型使用户更方便的开发并行应用

使用 Hadoop 可以轻松的组织计算机资源，从而搭建自己的分布式计算平台，并且可以充分利用集群的计算 和存储能力，完成海量数据的处理。

二、Hadoop 的优势

  1. 高可靠性：Hadoop 按位存储和处理数据的能力具有很高的可靠性
  2. 高拓展性：Hadoop 是在可用的计算机集簇间分配数据完成计算任务的，这些集簇可以拓展到数以千计的节点中
  3. 高效性：Hadoop 能够在节点之间动态地移动数据，以保证各个节点的动态平衡，因此其处理速度非常快
  4. 高容错性：Hadoop 能够自动保存数据的多份副本，并且能够自动将失败的任务重新分配

三、关联项目

  • Common：为 Hadoop 及其子项目提供支持的常用工具，主要包括 FileSystem，RPC 和串行化库。
  • Avro：Avro 用于数据序列化的系统。提供了丰富的数据结构类型、快速可压缩的二进制格式、存储持久性数据的文件集、远程调用 RPC 的功能和简单的动态语言集成功能。
  • MapReduce：是一种编程模型，用于大规模数据集（大于 1TB）的并行运算。
  • HDFS：分布式文件系统。
  • YRAN：分布式资源管理。
  • Chukwa：开源的数据收集系统，用于监控和分析大型分布式系统的数据。
  • Hive：一个建立在 Hadoop 基础之上的数据仓库，它提供了一些用于对 Hadoop 文件中的数据集进行数据整理、特殊查询和分析存储的工具。Hive 提供一种结构化数据的机制，支持类似传统 RDBMS 的 SQL 语言的查询语言来帮助那些熟悉 SQL 的用户查询 Hadoop 中的数据，该查询语言成为 Hive SQL。
  • Hbase：一个分布式的、面向列的开源数据库，适合非结构化的数据存储。主要用于需要随机访问、实时读写的大数据。
  • Pig：是一个对大型数据集进行分析、评估的平台。Pig 最突出的优势是它的结构能够经受住高度并行化的检验。
  • Zookeeper：为分布式应用设计的协调服务，主要为用户提供同步、配置管理、分组和命令等服务。

四、编译安装 Hadoop

因为我是用的是 32 位系统，官方预编译版本只有 64 位的，无法使用，所以得编译源代码。

根据编译文件 BUILDING.txt 内容，安装 hadoop 之前需要保证有以下工具：

Hadoop 编译说明书

需要：Unix 系统
JDK1.8
maven 3.3 或更高
ProtoBuffer 2.5.0
CMake 3.1 或更新（如果需要编译本地代码）Zlib develop（如果需要编译本地代码）openssl devel(如果编译原生 hadoop-pipe，并获得最佳的 HDFS 加密性能)
Linux FUSE(用户空间的文件系统) 2.6 或更高（如果编译 fuse_dfs）第一次编译需要网络保持连接（获取所有的 maven 和 Hadoop 需要的依赖）Python（发布文档需要）bats（用于 shell 代码测试）Node.js / bower / Ember-cli(用于编译前端 UI 界面)
---------------------------------------------------------------------
获得具有所有工具的环境的最简单方法是通过 Docker 提供的配置。这就需要一本最近的 docker 版本 1.4.1 或者更高的可以正常工作的版本

在 Linux 上，你可以运行下面的命名安装 Docker
$ ./start-build-env.sh
接下来显示的提示是位于源树的已安装版本，并且已安装和配置了所有必需的测试和构建工具。请注意，在此 docker 环境中，您只能从您开始的位置访问 Hadoop 源树。因此如果你想运行
dev-support/bin/test-patch /path/to/my.patch
那么这个 patch 文件必须放在 hadoop 源树中。在 ubuntu 中清楚并安装所需的软件包：Oracle JDK 1.8 (首选)
  $ sudo apt-get purge openjdk*
  $ sudo apt-get install software-properties-common
  $ sudo add-apt-repository ppa:webupd8team/java
  $ sudo apt-get update
  $ sudo apt-get install oracle-java8-installer
Maven
  $ sudo apt-get -y install maven
本地依赖包
  $ sudo apt-get -y install build-essential autoconf automake libtool cmake zlib1g-dev pkg-config libssl-dev
ProtocolBuffer 2.5.0 (必须)
  $ sudo apt-get -y install protobuf-compiler
  
# 1. 下载源码
wget https://www.apache.org/dyn/closer.cgi/hadoop/common/hadoop-3.1.2/hadoop-3.1.2-src.tar.gz
# 2. 解压
tar -zxcf hadoop-3.1.2-src.tar.gz
cd hadoop-3.1.2-src
# 3.mvn 编译
mvn package -Pdist,native -DskipTests -Dtar

编译这个玩意儿断断续续用了 3 天时间，下面是遇到的问题总结记录一下。

问题 1：

mvn package -Pdist,native -DskipTests -Dtar 的时候编译失败：

[ERROR] Failed to execute goal org.codehaus.mojo:native-maven-plugin:1.0-alpha-8:javah (default) on project hadoop-common: Error running javah command: Error executing command line. Exit code:2 -> [Help 1]

解决：

vim hadoop-common-project/hadoop-common/pom.xml将 javah 的执行路径改为绝对路径

<javahPath>${env.JAVA_HOME}/bin/javah</javahPath>
改为
<javahPath>/usr/bin/javah</javahPath>
# 具体的路径需要对应你机器上的真实路径

问题 2：

mvn package -Pdist,native -DskipTests -Dtar 的时候编译失败：

[ERROR] Failed to execute goal org.apache.hadoop:hadoop-maven-plugins:3.1.2:cmake-compile (cmake-compile) on project hadoop-common: CMake failed with error code 1 -> [Help 1]
org.apache.maven.lifecycle.LifecycleExecutionException: Failed to execute goal org.apache.hadoop:hadoop-maven-plugins:3.1.2:cmake-compile (cmake-compile) on project hadoop-common: CMake failed with error code 1

解决：

cmake 版本不对，安装 cmake3.0 版本：

# download
wget https://cmake.org/files/v3.0/cmake-3.0.0.tar.gz
tar -zxvf cmake-3.0.0.tar.gz
cd cmake-3.0.0
./configure
make
sudo apt-get install checkinstall
sudo checkinstall
sudo make install
# 建立软链接
sudo ln -s bin/* /usr/bin/

还是不行。使用 mvn package -Pdist,native -DskipTests -Dtar -e -X 打印所有日志，可以找到：

[INFO] Running cmake /home/wangjun/software/hadoop-3.1.2-src/hadoop-common-project/hadoop-common/src -DGENERATED_JAVAH=/home/wangjun/software/hadoop-3.1.2-src/hadoop-common-project/hadoop-common/target/native/javah -DJVM_ARCH_DATA_MODEL=32 -DREQUIRE_BZIP2=false -DREQUIRE_ISAL=false -DREQUIRE_OPENSSL=false -DREQUIRE_SNAPPY=false -DREQUIRE_ZSTD=false -G Unix Makefiles
[INFO] with extra environment variables {}
[WARNING] Soft-float JVM detected
[WARNING] CMake Error at /home/wangjun/software/hadoop-3.1.2-src/hadoop-common-project/hadoop-common/HadoopCommon.cmake:182 (message):
[WARNING]   Soft-float dev libraries required (e.g.  'apt-get install libc6-dev-armel'
[WARNING]   on Debian/Ubuntu)
[WARNING] Call Stack (most recent call first):
[WARNING]   CMakeLists.txt:26 (include)
[WARNING] 
[WARNING] 
[WARNING] -- Configuring incomplete, errors occurred!
[WARNING] See also "/home/wangjun/software/hadoop-3.1.2-src/hadoop-common-project/hadoop-common/target/native/CMakeFiles/CMakeOutput.log".
[WARNING] See also "/home/wangjun/software/hadoop-3.1.2-src/hadoop-common-project/hadoop-common/target/native/CMakeFiles/CMakeError.log".

查看 hadoop-common-project/hadoop-common/target/native/CMakeFiles/CMakeError.log 日志，看到报错：

gnu/stubs-soft.h: No such file or directory

解决方案：更改 hadoop-common-project/hadoop-common/HadoopCommon.cmake，将两处-mfloat-abi=softfp 改为-mfloat-abi=hard，参考：https://blog.csdn.net/wuyushe…。（最好是重新解压原始包更改完重新编译，要不然可能会出错）

这个改完又有了新问题，编译 Apache Hadoop MapReduce NativeTask 是报错

[WARNING] /usr/bin/ranlib libgtest.a
[WARNING] make[2]: Leaving directory '/home/wangjun/software/hadoop-3.1.2-src/hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-nativetask/target/native'
[WARNING] /usr/local/bin/cmake -E cmake_progress_report /home/wangjun/software/hadoop-3.1.2-src/hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-nativetask/target/native/CMakeFiles  1
[WARNING] [7%] Built target gtest
[WARNING] make[1]: Leaving directory '/home/wangjun/software/hadoop-3.1.2-src/hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-nativetask/target/native'
[WARNING] /tmp/ccpXG9td.s: Assembler messages:
[WARNING] /tmp/ccpXG9td.s:2040: Error: bad instruction `bswap r5'[WARNING] /tmp/ccpXG9td.s:2063: Error: bad instruction `bswap r1'
[WARNING] make[2]: *** [CMakeFiles/nativetask.dir/build.make:79: CMakeFiles/nativetask.dir/main/native/src/codec/BlockCodec.cc.o] Error 1
[WARNING] make[2]: *** Waiting for unfinished jobs....
[WARNING] make[1]: *** [CMakeFiles/Makefile2:96: CMakeFiles/nativetask.dir/all] Error 2
[WARNING] make[1]: *** Waiting for unfinished jobs....
[WARNING] /tmp/ccBbS5rL.s: Assembler messages:
[WARNING] /tmp/ccBbS5rL.s:1959: Error: bad instruction `bswap r5'[WARNING] /tmp/ccBbS5rL.s:1982: Error: bad instruction `bswap r1'
[WARNING] make[2]: *** [CMakeFiles/nativetask_static.dir/build.make:79: CMakeFiles/nativetask_static.dir/main/native/src/codec/BlockCodec.cc.o] Error 1
[WARNING] make[2]: *** Waiting for unfinished jobs....
[WARNING] /tmp/cc6DHbGO.s: Assembler messages:
[WARNING] /tmp/cc6DHbGO.s:979: Error: bad instruction `bswap r2'[WARNING] /tmp/cc6DHbGO.s:1003: Error: bad instruction `bswap r3'
[WARNING] make[2]: *** [CMakeFiles/nativetask_static.dir/build.make:125: CMakeFiles/nativetask_static.dir/main/native/src/codec/Lz4Codec.cc.o] Error 1
[WARNING] make[1]: *** [CMakeFiles/Makefile2:131: CMakeFiles/nativetask_static.dir/all] Error 2
[WARNING] make: *** [Makefile:77: all] Error 2

看错误应该是指令问题，google 一番后，找到解决方案：https://issues.apache.org/jir…。

编辑 primitives.h 文件，根据 https://issues.apache.org/jir… 里面的 git log 修改后重新编译。

经历了 3 天的折磨，终于成功了！来，看看成功后的显示：

[INFO] No site descriptor found: nothing to attach.
[INFO] ------------------------------------------------------------------------
[INFO] Reactor Summary for Apache Hadoop Main 3.1.2:
[INFO] 
[INFO] Apache Hadoop Main ................................. SUCCESS [3.532 s]
[INFO] Apache Hadoop Build Tools .......................... SUCCESS [6.274 s]
[INFO] Apache Hadoop Project POM .......................... SUCCESS [3.668 s]
[INFO] Apache Hadoop Annotations .......................... SUCCESS [5.743 s]
[INFO] Apache Hadoop Assemblies ........................... SUCCESS [1.739 s]
[INFO] Apache Hadoop Project Dist POM ..................... SUCCESS [4.782 s]
[INFO] Apache Hadoop Maven Plugins ........................ SUCCESS [10.777 s]
[INFO] Apache Hadoop MiniKDC .............................. SUCCESS [5.156 s]
[INFO] Apache Hadoop Auth ................................. SUCCESS [18.468 s]
[INFO] Apache Hadoop Auth Examples ........................ SUCCESS [8.293 s]
[INFO] Apache Hadoop Common ............................... SUCCESS [03:15 min]
[INFO] Apache Hadoop NFS .................................. SUCCESS [14.700 s]
[INFO] Apache Hadoop KMS .................................. SUCCESS [15.340 s]
[INFO] Apache Hadoop Common Project ....................... SUCCESS [0.876 s]
[INFO] Apache Hadoop HDFS Client .......................... SUCCESS [46.540 s]
[INFO] Apache Hadoop HDFS ................................. SUCCESS [02:34 min]
[INFO] Apache Hadoop HDFS Native Client ................... SUCCESS [12.125 s]
[INFO] Apache Hadoop HttpFS ............................... SUCCESS [20.005 s]
[INFO] Apache Hadoop HDFS-NFS ............................. SUCCESS [8.934 s]
[INFO] Apache Hadoop HDFS-RBF ............................. SUCCESS [01:08 min]
[INFO] Apache Hadoop HDFS Project ......................... SUCCESS [0.892 s]
[INFO] Apache Hadoop YARN ................................. SUCCESS [0.879 s]
[INFO] Apache Hadoop YARN API ............................. SUCCESS [25.531 s]
[INFO] Apache Hadoop YARN Common .......................... SUCCESS [01:57 min]
[INFO] Apache Hadoop YARN Registry ........................ SUCCESS [14.521 s]
[INFO] Apache Hadoop YARN Server .......................... SUCCESS [0.920 s]
[INFO] Apache Hadoop YARN Server Common ................... SUCCESS [23.432 s]
[INFO] Apache Hadoop YARN NodeManager ..................... SUCCESS [28.782 s]
[INFO] Apache Hadoop YARN Web Proxy ....................... SUCCESS [9.515 s]
[INFO] Apache Hadoop YARN ApplicationHistoryService ....... SUCCESS [14.077 s]
[INFO] Apache Hadoop YARN Timeline Service ................ SUCCESS [12.728 s]
[INFO] Apache Hadoop YARN ResourceManager ................. SUCCESS [51.338 s]
[INFO] Apache Hadoop YARN Server Tests .................... SUCCESS [8.675 s]
[INFO] Apache Hadoop YARN Client .......................... SUCCESS [13.937 s]
[INFO] Apache Hadoop YARN SharedCacheManager .............. SUCCESS [10.853 s]
[INFO] Apache Hadoop YARN Timeline Plugin Storage ......... SUCCESS [12.546 s]
[INFO] Apache Hadoop YARN TimelineService HBase Backend ... SUCCESS [1.069 s]
[INFO] Apache Hadoop YARN TimelineService HBase Common .... SUCCESS [17.176 s]
[INFO] Apache Hadoop YARN TimelineService HBase Client .... SUCCESS [15.662 s]
[INFO] Apache Hadoop YARN TimelineService HBase Servers ... SUCCESS [0.901 s]
[INFO] Apache Hadoop YARN TimelineService HBase Server 1.2  SUCCESS [17.512 s]
[INFO] Apache Hadoop YARN TimelineService HBase tests ..... SUCCESS [17.327 s]
[INFO] Apache Hadoop YARN Router .......................... SUCCESS [14.430 s]
[INFO] Apache Hadoop YARN Applications .................... SUCCESS [1.990 s]
[INFO] Apache Hadoop YARN DistributedShell ................ SUCCESS [10.400 s]
[INFO] Apache Hadoop YARN Unmanaged Am Launcher ........... SUCCESS [7.210 s]
[INFO] Apache Hadoop MapReduce Client ..................... SUCCESS [2.549 s]
[INFO] Apache Hadoop MapReduce Core ....................... SUCCESS [38.022 s]
[INFO] Apache Hadoop MapReduce Common ..................... SUCCESS [35.908 s]
[INFO] Apache Hadoop MapReduce Shuffle .................... SUCCESS [15.180 s]
[INFO] Apache Hadoop MapReduce App ........................ SUCCESS [18.915 s]
[INFO] Apache Hadoop MapReduce HistoryServer .............. SUCCESS [15.852 s]
[INFO] Apache Hadoop MapReduce JobClient .................. SUCCESS [12.987 s]
[INFO] Apache Hadoop Mini-Cluster ......................... SUCCESS [12.106 s]
[INFO] Apache Hadoop YARN Services ........................ SUCCESS [1.812 s]
[INFO] Apache Hadoop YARN Services Core ................... SUCCESS [8.685 s]
[INFO] Apache Hadoop YARN Services API .................... SUCCESS [9.236 s]
[INFO] Apache Hadoop YARN Site ............................ SUCCESS [0.859 s]
[INFO] Apache Hadoop YARN UI .............................. SUCCESS [0.840 s]
[INFO] Apache Hadoop YARN Project ......................... SUCCESS [34.971 s]
[INFO] Apache Hadoop MapReduce HistoryServer Plugins ...... SUCCESS [7.376 s]
[INFO] Apache Hadoop MapReduce NativeTask ................. SUCCESS [02:07 min]
[INFO] Apache Hadoop MapReduce Uploader ................... SUCCESS [9.915 s]
[INFO] Apache Hadoop MapReduce Examples ................... SUCCESS [14.651 s]
[INFO] Apache Hadoop MapReduce ............................ SUCCESS [15.959 s]
[INFO] Apache Hadoop MapReduce Streaming .................. SUCCESS [11.747 s]
[INFO] Apache Hadoop Distributed Copy ..................... SUCCESS [16.314 s]
[INFO] Apache Hadoop Archives ............................. SUCCESS [7.115 s]
[INFO] Apache Hadoop Archive Logs ......................... SUCCESS [8.686 s]
[INFO] Apache Hadoop Rumen ................................ SUCCESS [12.413 s]
[INFO] Apache Hadoop Gridmix .............................. SUCCESS [10.490 s]
[INFO] Apache Hadoop Data Join ............................ SUCCESS [7.894 s]
[INFO] Apache Hadoop Extras ............................... SUCCESS [7.098 s]
[INFO] Apache Hadoop Pipes ................................ SUCCESS [19.457 s]
[INFO] Apache Hadoop OpenStack support .................... SUCCESS [12.452 s]
[INFO] Apache Hadoop Amazon Web Services support .......... SUCCESS [04:55 min]
[INFO] Apache Hadoop Kafka Library support ................ SUCCESS [36.248 s]
[INFO] Apache Hadoop Azure support ........................ SUCCESS [43.752 s]
[INFO] Apache Hadoop Aliyun OSS support ................... SUCCESS [34.905 s]
[INFO] Apache Hadoop Client Aggregator .................... SUCCESS [17.099 s]
[INFO] Apache Hadoop Scheduler Load Simulator ............. SUCCESS [18.819 s]
[INFO] Apache Hadoop Resource Estimator Service ........... SUCCESS [29.363 s]
[INFO] Apache Hadoop Azure Data Lake support .............. SUCCESS [30.145 s]
[INFO] Apache Hadoop Image Generation Tool ................ SUCCESS [8.970 s]
[INFO] Apache Hadoop Tools Dist ........................... SUCCESS [46.265 s]
[INFO] Apache Hadoop Tools ................................ SUCCESS [0.883 s]
[INFO] Apache Hadoop Client API ........................... SUCCESS [08:41 min]
[INFO] Apache Hadoop Client Runtime ....................... SUCCESS [06:39 min]
[INFO] Apache Hadoop Client Packaging Invariants .......... SUCCESS [4.040 s]
[INFO] Apache Hadoop Client Test Minicluster .............. SUCCESS [13:29 min]
[INFO] Apache Hadoop Client Packaging Invariants for Test . SUCCESS [1.937 s]
[INFO] Apache Hadoop Client Packaging Integration Tests ... SUCCESS [1.865 s]
[INFO] Apache Hadoop Distribution ......................... SUCCESS [01:56 min]
[INFO] Apache Hadoop Client Modules ....................... SUCCESS [5.050 s]
[INFO] Apache Hadoop Cloud Storage ........................ SUCCESS [6.457 s]
[INFO] Apache Hadoop Cloud Storage Project ................ SUCCESS [0.829 s]
[INFO] ------------------------------------------------------------------------
[INFO] BUILD SUCCESS
[INFO] ------------------------------------------------------------------------
[INFO] Total time:  01:06 h
[INFO] Finished at: 2019-09-03T14:14:45+08:00
[INFO] ------------------------------------------------------------------------

编译完成后的内容在 hadoop-dist 里面。感受一下为了编译这个玩意儿尝试了多少个版本：

cmake-3.0.0         
cmake-3.3.0       
hadoop-2.7.7-src.tar.gz  
hadoop-2.9.2-src         
hadoop-3.1.2-src                   
protobuf-2.5.0    
cmake-3.1.0         
hadoop-2.8.5-src     
hadoop-2.9.2-src.tar.gz  
hadoop-3.1.2-src.tar.gz  
cmake-3.1.0.tar.gz  
hadoop-2.7.7-src  
hadoop-2.8.5-src.tar.gz  
hadoop-3.1.2             
hadoop-3.1.2.tar.gz

五、启动运行 hadoop

将 hadoop-dist/target 里面的 hadoop-3.1.2.tar.gz 拷贝到你要安装的位置，解压。

# 进入 bin 目录，启动前先格式化 HDFS 系统
cd hadoop-3.1.2/bin
./hdfs namenode -format
......
......
2019-09-03 14:35:53,356 INFO namenode.NameNode: SHUTDOWN_MSG: 
/************************************************************
SHUTDOWN_MSG: Shutting down NameNode at raspberrypi/127.0.1.1
************************************************************/
# 启动所有服务
cd ../sbin/
./start-all.sh
WARNING: Attempting to start all Apache Hadoop daemons as wangjun in 10 seconds.
WARNING: This is not a recommended production deployment configuration.
WARNING: Use CTRL-C to abort.
Starting namenodes on [raspberrypi]
Starting datanodes
Starting secondary namenodes [raspberrypi]
Starting resourcemanager
Starting nodemanagers

访问 8088 端口 http://localhost:8088 就可以看到 hadoop 的管理界面了！

hadoop 的 web 界面：

# All Applications
http://localhost:8088
# DataNode Information
http://localhost:9864
# Namenode Information
http://localhost:9870
# node
http://localhost:8042
# SecondaryNamenode information
http://localhost:9868

问题 1：启动时报错：

$ ./start-all.sh 
WARNING: Attempting to start all Apache Hadoop daemons as wangjun in 10 seconds.
WARNING: This is not a recommended production deployment configuration.
WARNING: Use CTRL-C to abort.
Starting namenodes on [raspberrypi]
raspberrypi: ERROR: JAVA_HOME is not set and could not be found.
Starting datanodes
localhost: ERROR: JAVA_HOME is not set and could not be found.
Starting secondary namenodes [raspberrypi]
raspberrypi: ERROR: JAVA_HOME is not set and could not be found.
Starting resourcemanager
Starting nodemanagers
localhost: ERROR: JAVA_HOME is not set and could not be found.

解决方案：

vim ./etc/hadoop/hadoop-env.sh
# export JAVA_HOME=
改为具体的 java 安装路径，比如
export JAVA_HOME=/usr/lib/jvm/java-8-openjdk-armhf

正文完
发表至： hadoop
2019-09-19
 0
Hadoop-YARN调度性能优化实践
为何零基础也可学习大数据技术
关于hadoop:史上最全Hadoop-核心-HDFS-分布式文件系统详解上万字建议收藏
Hadoop分布式集群的搭建