首页 > 代码库 > Java之美[从菜鸟到高手演练]之Hadoop常用命令

Java之美[从菜鸟到高手演练]之Hadoop常用命令

作者:二青

邮箱:xtfggef@gmail.com     微博:http://weibo.com/xtfggef

这篇文章主要是讲一下位于bin下的hadoop命令,我们可以直接输入hadoop无任何参数看一下:

技术分享


用法就是:hadoop [---config confdir] COMMAND此处COMMAND就是下面列出来的那些,fs, version,jar 等等。

用户命令

fs

目前版本的hadoop已经摒弃了fs命令,取而代之的是hdfs dfs. 

Usage: hdfs dfs [GENERIC_OPTIONS] [COMMAND_OPTIONS], 熟悉linux命令的同学学这个很快,举个例子,创建目录,mkdir,在这里为:hdfs dfs -mkdir /test

解释:此处-mkdir即为GENERIC_OPTIONS,后面的/test为COMMAND_OPTIONS,默认路径为/,所以执行完上面的操作后,我们在根路径/下面创建了test目录。下面是我总结了一下hadoop中fs相关的命令,亦可见文件EXCEL。

参数作用示例返回值
appendToFile将一个或者多个本地
文件追加到目的文件
hdfs dfs -appendToFile localfile /user/hadoop/hadoopfileReturns 0 on success and 1 on error
cat输出文件hdfs dfs -cat file:///file3 /user/hadoop/file4Returns 0 on success and -1 on error
chgrp改变文件的分组hdfs dfs -chgrp [-R] GROUP URI [URI ...] 
chmod改变文件的权限hdfs dfs -chmod [-R] <MODE[,MODE]... | OCTALMODE> URI [URI ...] 
chown改变文件的拥有者hdfs dfs -chown [-R] [OWNER][:[GROUP]] URI [URI ] 
copyFromLocal从本地复制  
copyToLocal复制到本地  
count得到文件/目录等数目
追加参数-q, -h有不同的意义
hdfs dfs -count -q hdfs://nn1.example.com/file1Returns 0 on success and -1 on error
cp复制,参数-f,-phdfs dfs -cp /user/hadoop/file1 /user/hadoop/file2Returns 0 on success and -1 on error
du得到指定文件的大小hdfs dfs -du /test/hadoopReturns 0 on success and -1 on error.
dus已摒弃,和du类似  
expunge清空回收站hdfs dfs -expunge 
get复制文件到本地路径下hdfs dfs -get /user/hadoop/file localfileReturns 0 on success and -1 on error
getfacl显示文件或者目录的权限控制列表hdfs dfs -getfacl /file
hdfs dfs -getfacl -R /dir
Returns 0 on success and non-zero on error
getfattr显示文件或者目录的扩展属性hdfs dfs -getfattr -d /fileReturns 0 on success and non-zero on error
getmerge合并多个文件一个目标文件里hdfs dfs -getmerge <src> <localdst> [addnl] 
lslinux里一样hdfs dfs -ls /user/hadoop/file1Returns 0 on success and -1 on error
lsr等同于ls -R  
mkdir创建目录,-p创建多层目录hdfs dfs -mkdir /user/hadoop/dir1 /user/hadoop/dir2Returns 0 on success and -1 on error
moveFromLocal类似put,区别在于put完后删除
原文件
  
moveToLocal目前没有实现  
mv移动文件hdfs dfs -mv /user/hadoop/file1 /user/hadoop/file2Returns 0 on success and -1 on error
put像目标目录推送文件hdfs dfs -put localfile /user/hadoop/hadoopfileReturns 0 on success and -1 on error
rm删除文件hdfs dfs -rm hdfs://nn.example.com/file /user/hadoop/emptydirReturns 0 on success and -1 on error
rmr类似于rm -r  
setfacl设置文件或者目录的权限控制列表hdfs dfs -setfacl -m user:hadoop:rw- /fileReturns 0 on success and non-zero on error
setfattr设置文件或者目录的扩展属性hdfs dfs -setfattr -n user.myAttr -v myValue /fileReturns 0 on success and non-zero on error
setrep改变文件和目录的复制因子hdfs dfs -setrep -w 3 /user/hadoop/dir1Returns 0 on success and -1 on error
stat返回路径信息hdfs dfs -stat pathExit Code: Returns 0 on success and -1 on error
tail输出文件的最后1千字节hdfs dfs -tail pathnameReturns 0 on success and -1 on error
test检查文件hdfs dfs -test -e filename 
text以文本方式输出文件hdfs dfs -text <src> 
touchz创建空文件hdfs dfs -touchz pathnameReturns 0 on success and -1 on error


version

hadoop version查看hadoop版本信息

jar

可以运行jar文件

credential

这个命令用来管理证书,密钥和一些其他的私有信息。

archive

1. 创建archive

Hadoop archive 是一种特定的文档格式,每一个archive会与相应的系统目录相对应,以*.har结尾。每一个archive文档包含一些信息(metadata),记录该文档的一些信息,如索引,数据等。使用如下命令去创建hadoop archive:hadoop archive -archiveName zoo.har -p /foo/bar -r 3 /outputdir

-archiveName:定义要创建的archive文档的名字。

-p: 定义要archive的目录,可以是绝对路径,也可是相对路径。

-r: 定义复制因子,默认是10。

最后一个是输出目录,我们来看一个例子:把/hadoop下的config文件夹打包成config.har并放到/archives下,执行如下命令:

hadoop archive -archiveName config.har -p /hadoop/hadoop/etc -r 1 hadoop /archives

root@master:~/hadoop/bin# ./hadoop archive -archiveName config.har -p /hadoop -r 1 config /archives
15/01/15 14:28:02 INFO client.RMProxy: Connecting to ResourceManager at master/192.168.1.118:8032
15/01/15 14:28:04 INFO client.RMProxy: Connecting to ResourceManager at master/192.168.1.118:8032
15/01/15 14:28:04 INFO client.RMProxy: Connecting to ResourceManager at master/192.168.1.118:8032
15/01/15 14:28:05 INFO mapreduce.JobSubmitter: number of splits:1
15/01/15 14:28:05 INFO mapreduce.JobSubmitter: Submitting tokens for job: job_1421303038566_0001
15/01/15 14:28:06 INFO impl.YarnClientImpl: Submitted application application_1421303038566_0001
15/01/15 14:28:06 INFO mapreduce.Job: The url to track the job: http://master:8088/proxy/application_1421303038566_0001/
15/01/15 14:28:06 INFO mapreduce.Job: Running job: job_1421303038566_0001
15/01/15 14:28:21 INFO mapreduce.Job: Job job_1421303038566_0001 running in uber mode : false
15/01/15 14:28:21 INFO mapreduce.Job:  map 0% reduce 0%
15/01/15 14:28:33 INFO mapreduce.Job:  map 100% reduce 0%
15/01/15 14:28:44 INFO mapreduce.Job:  map 100% reduce 100%
15/01/15 14:28:44 INFO mapreduce.Job: Job job_1421303038566_0001 completed successfully
15/01/15 14:28:44 INFO mapreduce.Job: Counters: 49
        File System Counters
                FILE: Number of bytes read=701
                FILE: Number of bytes written=215089
                FILE: Number of read operations=0
                FILE: Number of large read operations=0
                FILE: Number of write operations=0
                HDFS: Number of bytes read=23417
                HDFS: Number of bytes written=23343
                HDFS: Number of read operations=24
                HDFS: Number of large read operations=0
                HDFS: Number of write operations=7
        Job Counters
                Launched map tasks=1
                Launched reduce tasks=1
                Other local map tasks=1
                Total time spent by all maps in occupied slots (ms)=9767
                Total time spent by all reduces in occupied slots (ms)=7888
                Total time spent by all map tasks (ms)=9767
                Total time spent by all reduce tasks (ms)=7888
                Total vcore-seconds taken by all map tasks=9767
                Total vcore-seconds taken by all reduce tasks=7888
                Total megabyte-seconds taken by all map tasks=10001408
                Total megabyte-seconds taken by all reduce tasks=8077312
        Map-Reduce Framework
                Map input records=7
                Map output records=7
                Map output bytes=680
                Map output materialized bytes=701
                Input split bytes=116
                Combine input records=0
                Combine output records=0
                Reduce input groups=7
                Reduce shuffle bytes=701
                Reduce input records=7
                Reduce output records=0
                Spilled Records=14
                Shuffled Maps =1
                Failed Shuffles=0
                Merged Map outputs=1
                GC time elapsed (ms)=291
                CPU time spent (ms)=2090
                Physical memory (bytes) snapshot=306540544
                Virtual memory (bytes) snapshot=1320914944
                Total committed heap usage (bytes)=127242240
        Shuffle Errors
                BAD_ID=0
                CONNECTION=0
                IO_ERROR=0
                WRONG_LENGTH=0
                WRONG_MAP=0
                WRONG_REDUCE=0
        File Input Format Counters
                Bytes Read=632
        File Output Format Counters
                Bytes Written=0
root@master:~/hadoop/bin#

浏览目录结构,我们看到了生成的har包:

技术分享


用命令查看,并且copy到本地,得到如下:

技术分享


我们发现har包看上去就是一个文件夹,不过比较特殊的是,它里面包含metadata,记录文件的基本信息和数据,而且有专门的特殊用途。


2. 查找archive

我们可以像操作普通文件系统一下去操作archive,唯一不同的是查找路径URI,这个需要注意一下。

har://scheme-hostname:port/archivepath/fileinarchive
har:///archivepath/fileinarchive

root@master:~/hadoop/bin# ./hdfs dfs -ls har:///archives/config.har
Found 1 items
drwxr-xr-x   - root supergroup          0 2015-01-15 14:26 har:///archives/config.har/config

3. 解压缩archive

hdfs dfs -cp har:///user/zoo/foo.har/dir1 hdfs:/user/zoo/newdir
hadoop distcp har:///user/zoo/foo.har/dir1 hdfs:/user/zoo/newdir

官方文档

distcp

distcp是一个文件和目录复制命令,用于集群之间已经集群内节点之间的文件复制,详情见官方文档。

fsck

fsck是一个文件系统检查工具,用来检查各类问题,比如文件块丢失等。

用法:hdfs fsck [GENERIC_OPTIONS] <path> [-list-corruptfileblocks | [-move | -delete | -openforwrite] [-files [-blocks [-locations | -racks]]]] [-includeSnapshots]

COMMAND_OPTIONDescription
pathStart checking from this path.
-moveMove corrupted files to /lost+found
-deleteDelete corrupted files.
-filesPrint out files being checked.
-openforwritePrint out files opened for write.
-includeSnapshotsInclude snapshot data if the given path indicates a snapshottable directory or there are snapshottable directories under it.
-list-corruptfileblocksPrint out list of missing blocks and files they belong to.
-blocksPrint out block report.
-locationsPrint out locations for every block.
-racksPrint out network topology for data-node locations.

这是一个独立的命令,不属于dfs. 看个例子:检查/下面的文件目录情况。

技术分享

fetchdt

从namenode上拿到授权符,暂时没有用到,后面更新。

job

与map reduce的job进行交互。

Usage: mapred job | [GENERIC_OPTIONS] | [-submit <job-file>] | [-status <job-id>] | [-counter <job-id> <group-name> <counter-name>] | [-kill <job-id>] | [-events <job-id> <from-event-#> <#-of-events>] | [-history [all] <jobOutputDir>] | [-list [all]] | [-kill-task <task-id>] | [-fail-task <task-id>] | [-set-priority <job-id> <priority>]

COMMAND_OPTIONDescription
-submit job-fileSubmits the job.
-status job-idPrints the map and reduce completion percentage and all job counters.
-counter job-id group-name counter-namePrints the counter value.
-kill job-idKills the job.
-events job-id from-event-# #-of-eventsPrints the events‘ details received by jobtracker for the given range.
-history [all]jobOutputDirPrints job details, failed and killed tip details. More details about the job such as successful tasks and task attempts made for each task can be viewed by specifying the [all] option.
-list [all]Displays jobs which are yet to complete. -list all displays all jobs.
-kill-task task-idKills the task. Killed tasks are NOT counted against failed attempts.
-fail-task task-idFails the task. Failed tasks are counted against failed attempts.
-set-priority job-id priorityChanges the priority of the job. Allowed priority values are VERY_HIGH, HIGH, NORMAL, LOW, VERY_LOW

我们可以通过:mapred job -status <job_id> 来查看job的运行状态。

技术分享

pipes

运行管道job,稍后更新。

queue

获得与job交互的队列的信息,用法:./mapred queue -info default

技术分享

CLASSNAME

hadoop可以运行任何java class。

classpath

打印类路径。

管理命令

balancer

查看系统平衡情况,hdfs balancer [-threshold <threshold>] [-policy <policy>], 最直接的用法:./hdfs balancer
root@master:~/hadoop/bin# ./hdfs balancer
15/01/15 16:20:03 INFO balancer.Balancer: namenodes  = [hdfs://master:9000]
15/01/15 16:20:03 INFO balancer.Balancer: parameters = Balancer.Parameters[BalancingPolicy.Node, threshold=10.0, number of nodes to be excluded = 0, number of nodes to be included = 0]
Time Stamp               Iteration#  Bytes Already Moved  Bytes Left To Move  Bytes Being Moved
15/01/15 16:20:06 INFO net.NetworkTopology: Adding a new node: /default-rack/192.168.1.116:50010
15/01/15 16:20:06 INFO net.NetworkTopology: Adding a new node: /default-rack/192.168.1.189:50010
15/01/15 16:20:06 INFO balancer.Balancer: 0 over-utilized: []
15/01/15 16:20:06 INFO balancer.Balancer: 0 underutilized: []
The cluster is balanced. Exiting...
Jan 15, 2015 4:20:06 PM           0                  0 B                 0 B               -1 B
Jan 15, 2015 4:20:06 PM  Balancing took 3.215 seconds
root@master:~/hadoop/bin#

datanode

运行一个data node,Usage: hdfs datanode [-regular | -rollback | -rollingupgrace rollback]
COMMAND_OPTIONDescription
-regularNormal datanode startup (default).
-rollbackRollback the datanode to the previous version. This should be used after stopping the datanode and distributing the old hadoop version.
-rollingupgrade rollbackRollback a rolling upgrade operation.

daemonlog

dfsadmin

dfs管理工具,功能比较强大,参考官方文档。

Mover

一个新的数据移植工具,和balancer类似,定期扫描HDFS系统里的archive文件,来检查是否满足HDFS存储策略,如果不满足则将其副本移到另一个地方。

namenode

运行namenode,也是一个比较核心、强大的工具,参考官方文档。

secondarynamenode

运行HDFS次要name node。

hsadmin

管理HDFS history server

historyserver

启动history server,用法:./mapred historyserver

Java之美[从菜鸟到高手演练]之Hadoop常用命令