學(xué)習(xí)筆記--hadoop

Hadoop

hadoop-1.2.1.tar.gz jdk-6u32-linux-x64.bin

Useradd  -u 900 hadoop

Mv jdk1.6.0_32  /home/hadoop

Mv hadoop-1.2.1.tar.gz/home/hadoop

Chown hadoop.hadoop /home/hadoop -R

Su -hadoop

Ln -s jdk1.6.0_32 java

Tar zxf hadoop-1.2.1.tar.gz hadoop-1.2.1

Ln -s hadoop-1.2.1 hadoop

更改環(huán)境變量：

Vim /hadoop/conf/hadoop-env.sh

Cd /hadoop

Mkdir input

Cp conf/*.xml input

Bin/hadoop jar hadoop-examples-1.2.1.jar grep input output ‘dfs[a-z.]+’

設(shè)置無密碼登陸：

Ssh-keygen

Ssh-copy-id 172.25.60.1

保證 master 到所有的 slave 節(jié)點(diǎn)都實(shí)現(xiàn)無密碼登錄

Cd ~/hadoop/conf

Vim slaves ---->172.25.60.1

Vim masters---->172.25.60.1

Vim core-site.xml  在configuration中間添加以下內(nèi)容

<configuration>

        <property>

                <name>fs.default.name</name>

                        <value>hdfs://172.25.60.1:9000</value>

                                </property>

</configuration>

Vim hdfs-site.xml 在configuration中間添加以下內(nèi)容

<configuration>

        <property>

                <name>dfs.replication</name>

                        <value>1</value>

                                </property>

</configuration>

Vim mapred-site.xml 在configuration中間添加以下內(nèi)容

<configuration>

        <property>

                <name>mapred.job.tracker</name>

                        <value>172.25.60.1:9001</value>

                                </property>

</configuration>

格式化一個新的分布式文件系統(tǒng):

$ bin/hadoop namenode  -format

啟動 Hadoop 守護(hù)進(jìn)程:

$ bin/start-all.sh

在各個節(jié)點(diǎn)查看 hadoop 進(jìn)程:

$ jps

Hadoop 守護(hù)進(jìn)程的日志寫入到 ${HADOOP_HOME}/logs 目錄

瀏覽 NameNode 和 JobTracker 的網(wǎng)絡(luò)接口,它們的地址默認(rèn)為:

NameNode – http://172.25.60.1:50070/

JobTracker – http://172.25.60.1:50030/

將輸入文件拷貝到分布式文件系統(tǒng):

$ bin/hadoop fs -put conf input

運(yùn)行發(fā)行版提供的示例程序:

$ bin/hadoop jar hadoop-*-examples.jar grep input output 'dfs[a-z.]+'

查看輸出文件:

將輸出文件從分布式文件系統(tǒng)拷貝到本地文件系統(tǒng)查看:

$ bin/hadoop fs -get output output

$ cat output/*

或者

在分布式文件系統(tǒng)上查看輸出文件:

$ bin/hadoop fs -cat output/*

完成全部操作后,停止守護(hù)進(jìn)程:

$ bin/stop-all.sh

完全分布式（三個節(jié)點(diǎn)）server1 server2 server4：

在三個節(jié)點(diǎn)上安裝 rpcbind nfs-utils 并打開rpcind nfs服務(wù)

Vim /etc/exports

/home/hadoop    *(rw,all_squash,anonuid=900,anongid=900)

在slave2 4上添加用戶  useradd -u 900 hadoop

mount  172.25.60.1:/home/hadoop/  /home/hadoop/

在1上分別進(jìn)行ssh 連接 ssh 172.25.60.2ssh272.25.60.4

在master上：vim  ~/hadoop/conf

Vim slaves

172.25.60.2

172.25.60.4

Vim hdfs-site.xml

<Value>1</value> ------> <Value>2</value>

(刪除tmp------>格式化---->bin/start-dfs.sh----->bin/hadoop fs -put conf/ input----->bin/start-mapred.sh

bin/hadoop jar hadoop-examples-1.2.1.jar grep input output 'dfs[a-z.]+')

bin/hadoop  dfsadmin-report：查看節(jié)點(diǎn)運(yùn)行狀態(tài)

bin/hadoop fs -ls :查看輸出文件

添加一個節(jié)點(diǎn)文件：

在線添加節(jié)點(diǎn)：

添加用戶 useradd -u 900 hadoop

mount 172.25.60.1:/home/hadoop/home/hadoop

su - hadoop

vim slaves加入該節(jié)點(diǎn) ----->>172.25.60.5

bin/hadoop-daemon.sh start datanode

bin/hadoop-daemon.sh start tasktracker

在線刪除節(jié)點(diǎn)：

先做數(shù)據(jù)遷移：

在server上：vim mapred-site.xml

<property>

<name>dfs.hosts.exclude</name>

<value>/home/hadoop/hadoop/conf/hostexclude</value>

</property>

Vim hadoop/hadoop/conf/hostexclude------->172.25.60.4

Bin/hadoop dfsadmin -refreshNodes  ####刷新節(jié)點(diǎn)

回收站功能：

vimcore-site.xml添加以下:

<property>

<name>fs.trash.interval</name>

<value>1440</value>      1440=60*24

</property>

實(shí)驗(yàn)：bin/hadoop fs -rm input/hadoop-env.sh

bin/hadoop fs -ls input  查看是否刪除

bin/hadoop fs -ls 此時新增目錄 .Trash

bin/hadoop fs -ls .Trash/Current/user/hadoop/input

將此文件移回原目錄即可恢復(fù)

bin/hadoop fs -mv .Trash/Current/user/hadoop/input/hadoop-env.sh input   

優(yōu)化:

更新hadoop至2.6版本

刪除之前的鏈接，解壓hadoop-2.6.4.tar.gz   jdk-7u79-linux-x64.tar.gz到hadoop家目錄，并更改權(quán)限為hadoop.hadoop進(jìn)入hadoop用戶，鏈接成hadoop和java，進(jìn)入hadoop/etc/hadoop/

vim  hadoop-env.sh    export JAVA_HOME=/home/hadoop/java

cd /hadoop/etc/hadoop

vim  core-site.xml

<property>

                <name>fs.defaultFS</name>

                <value>hdfs://172.25.60.1:9000</value>

                        </property>

vim hdfs-site.xml

  <property>

                <name>dfs.replication</name>

                <value>2</value>

                        </property>

vim yarn-env.sh

# some Java parameters

export JAVA_HOME=/home/hadoop/java

cp  mapred-site.xml.template  mapred-site.xml

     <property>

                <name>mapreduce.framework.name</name>

                <value>yarn</value>

                        </property>

vim yarn-site.xml

         <property>

                <name>yarn.nodemanager.aux-services</name>

                <value>mapreduce_shuffle</value>

        </property>

vim slaves  

172.25.60.4

172.25.60.5

bin/hdfs namenode -format

sbin/start-dfs.sh

bin/hdfs dfs -mkdir /user

bin/hdfs dfs -mkdir /user/hadoop

bin/hdfs dfs -put etc/hadoop input

sbin/start-yarn.sh

bin/hadoop jar share/hadoop/mapreduce/hadoop-mapreduce-examples-2.6.4.jar grep input output 'dfs[a-z.]+'

訪問 172.25.60.1:50070   172.25.60.1:8088

##########替換lib下文件為64位(不更改的話啟動時會有warn警告)

mv   hadoop-native-64-2.6.0.tar   /home/hadoop/hadoop/lib/native

tarxf  hadoop-native-64-2.6.0.tar

###########指定節(jié)點(diǎn)目錄