wiki:waue/2009/0402

  1. Hadoop分佈式文件系統使用指南二
    1. 升級
    2. 退回
  2. bin/hadoop 的使用者用法
    1. archieve
    2. distCp
    3. fsck
    4. job
      1. -status
      2. -kill
      3. -list
      4. -history
    5. version
  3. bin/hadoop 的管理者用法
    1. balancer
    2. daemonlog
    3. dfsadmin
      1. -report
      2. -safemode
      3. -refreshNodes
      4. -finalizeUpgrade
      5. -upgradeProgress
      6. -setQuota
      7. -clrQuota
    4. namenode
    5. secondarynamenode
    6. tasktracker、datanode
  4. Hadoop Streaming 函式庫用法
  5. HDFS權限管理用戶

Hadoop分佈式文件系統使用指南二

升級

  • 由於換版本的話，資料夾內的conf設定檔也勢必被更改，因此目前作法為： 把conf 移至/opt/conf ，hadoop 0.16 與 hadoop 0.18用 ln 做捷徑代換。由於conf已不在hadoop_home內，因此記得匯入conf/hadoop-env.sh 
    $ source /opt/conf/hadoop-env.sh
  • 先看狀態 
    $ bin/hadoop dfsadmin -upgradeProgress status

There are no upgrades in progress.
  • 停止hdfs
    • 注意不可使用bin/stop-all.sh來停止 
      $ bin/stop-dfs.sh
  • 部署新版本的Hadoop
    • 注意每個node的版本都要統一，否則會出現問題
  • 啟動 
    $ bin/start-dfs.sh -upgrade

ps：之後有介紹到 bin/hadoop namenode -upgrade ，應該要查查看與 $ bin/start-dfs.sh -upgrade 有何不同

  • namenode管理網頁會出現升級狀態

退回

  • 停止集群 
    $ bin/stop-dfs.sh
  • 部署老版本的Hadoop
  • 退回之前版本 
    $ bin/start-dfs.sh -rollback
    ps：之後有介紹到 bin/hadoop namenode -rollback ，應該要查查看與 $ bin/start-dfs.sh -rollback 有何不同

bin/hadoop 的使用者用法

$ hadoop [--config confdir] [COMMAND] [GENERIC_OPTIONS] [COMMAND_OPTIONS]

[GENERIC_OPTIONS] :

-conf <configuration file> 指定應用程序的配置文件。
-D <property=value> 為指定property指定值value。
-fs <local|namenode:port> 指定namenode。
-jt <local|jobtracker:port> 指定job tracker。只適用於job。

archieve

  • archieve就是把資料壓縮成一個檔案，在壓縮的過程中，還會將被壓縮的目錄結構紀錄在index與masterindex內。
  • 由於每個上傳上去的檔案都被放在一個block中，因此我的input資料夾內共有四個檔，但是每個檔都會佔用一個block，用此方法就可以按照整個打包大小來分配共用去多少個block數。
  • hadoop archive -archiveName name <src>* <dest> 
    $ bin/hadoop archive -archiveName foo.har input/* output
09/04/02 14:02:30 WARN mapred.JobClient: Use GenericOptionsParser for parsing the arguments. Applications should implement Tool for the same.
09/04/02 14:02:30 INFO mapred.JobClient: Running job: job_200904021140_0001
09/04/02 14:02:31 INFO mapred.JobClient:  map 0% reduce 0%
09/04/02 14:02:44 INFO mapred.JobClient:  map 20% reduce 0%
09/04/02 14:02:49 INFO mapred.JobClient:  map 100% reduce 0%
09/04/02 14:02:56 INFO mapred.JobClient: Job complete: job_200904021140_0001
...略
  • 看har裡面的檔案結構 
    $ bin/hadoop dfs -lsr /user/waue/output/foo.har
  • 看har內檔案的內容 
    $ bin/hadoop dfs -cat /user/waue/output/foo.har/part-0
  • ps: 官方文件介紹的 hadoop dfs -lsr har:///user/hadoop/output/foo.har 會出現錯誤！ 
    lsr: could not get get listing for 'har:/user/waue/output/foo.har/user/waue' : File: har://hdfs-gm1.nchc.org.tw:9000/user/waue/output/foo.har/user/waue/input does not exist in har:///user/waue/output/foo.har

distCp

  • 是用於大規模集群內部和集群之間拷貝的工具
  • 使用Map/Reduce實現文件分發，錯誤處理和恢復，以及報告生成
  • 舉例為： 
    hadoop distcp hdfs://nn1:8020/foo/bar hdfs://nn2:8020/bar/foo

？？ 然而8020 port 在機器上沒有開，且不是應該檔案會均勻散佈在每個節點上嗎？怎麼還會知道nn1的節點上有這個檔要複製到nn2呢？

fsck

  • HDFS文件系統檢查工具 
$ bin/hadoop fsck /

.
/user/waue/input/1.txt:  Under replicated blk_-90085106852013388_1001. Target Replicas is 3 but found 2 replica(s).
/user/waue/input/1.txt:  Under replicated blk_-4027196261436469955_1001. Target Replicas is 3 but found 2 replica(s).
.
/user/waue/input/2.txt:  Under replicated blk_-2300843106107816641_1002. Target Replicas is 3 but found 2 replica(s).
.
/user/waue/input/3.txt:  Under replicated blk_-1561577350198661966_1003. Target Replicas is 3 but found 2 replica(s).
.
/user/waue/input/4.txt:  Under replicated blk_1316726598778579026_1004. Target Replicas is 3 but found 2 replica(s).
Status: HEALTHY
 Total size:	143451003 B
 Total dirs:	8
 Total files:	4
 Total blocks (validated):	5 (avg. block size 28690200 B)
 Minimally replicated blocks:	5 (100.0 %)
 Over-replicated blocks:	0 (0.0 %)
 Under-replicated blocks:	5 (100.0 %)
 Mis-replicated blocks:		0 (0.0 %)
 Default replication factor:	3
 Average block replication:	2.0
 Corrupt blocks:		0
 Missing replicas:		5 (50.0 %)
 Number of data-nodes:		2
 Number of racks:		1
The filesystem under path '/' is HEALTHY
  • 加不同的參數有不同的用處，如 
    $ bin/hadoop fsck / -files

/tmp <dir>
/tmp/hadoop <dir>
/tmp/hadoop/hadoop-waue <dir>
/tmp/hadoop/hadoop-waue/mapred <dir>
/tmp/hadoop/hadoop-waue/mapred/system <dir>
/user <dir>
/user/waue <dir>
/user/waue/input <dir>
/user/waue/input/1.txt 115045564 bytes, 2 block(s):  Under replicated blk_-90085106852013388_1001. Target Replicas is 3 but found 2 replica(s).
 Under replicated blk_-4027196261436469955_1001. Target Replicas is 3 but found 2 replica(s).
/user/waue/input/2.txt 987864 bytes, 1 block(s):  Under replicated blk_-2300843106107816641_1002. Target Replicas is 3 but found 2 replica(s).
/user/waue/input/3.txt 1573048 bytes, 1 block(s):  Under replicated blk_-1561577350198661966_1003. Target Replicas is 3 but found 2 replica(s).
/user/waue/input/4.txt 25844527 bytes, 1 block(s):  Under replicated blk_1316726598778579026_1004. Target Replicas is 3 but found 2 replica(s).
Status: HEALTHY
....（同上）
-move -delete -openforwrite
移動受損文件到/lost+found 刪除受損文件 印出寫打開的文件
-files -blocks -locations -racks
印出正被檢查的文件 印出區塊的資訊 印出每個區塊的位置 印出data-node的網絡拓撲結構

如： 

$ bin/hadoop fsck /user/waue/input/1.txt -files -blocks -locations

/user/waue/input/1.txt 115045564 bytes, 2 block(s):  Under replicated blk_-90085106852013388_1001. Target Replicas is 3 but found 2 replica(s).
 Under replicated blk_-4027196261436469955_1001. Target Replicas is 3 but found 2 replica(s).
0. blk_-90085106852013388_1001 len=67108864 repl=2 [140.110.138.191:50010, 140.110.141.129:50010]
1. blk_-4027196261436469955_1001 len=47936700 repl=2 [140.110.138.191:50010, 140.110.141.129:50010]
Status: HEALTHY
 Total size:	115045564 B
 Total dirs:	0
 Total files:	1
....（略）

job

  • 用以跟Map Reduce 的作業程序溝通
  • 在測試此指令之前，請確認已經先執行過mapReduce的程序過
  • 可到JobTracker:50030網頁來看程序的Jobid

-status

  • 查看工作狀態 
    $ bin/hadoop job -status job_200904021140_0001

-kill

  • 終止正在執行的程序，其id為 job_200904021140_0001 
    $ bin/hadoop job -kill job_200904021140_0001

-list

  • 印出所有程序的狀態 
    $ bin/hadoop job -list all

5 jobs submitted
States are:
	Running : 1	Succeded : 2	Failed : 3	Prep : 4
JobId	State	StartTime	UserName
job_200904021140_0001	2	1238652150499	waue
job_200904021140_0002	3	1238657754096	waue
job_200904021140_0004	3	1238657989495	waue
job_200904021140_0005	2	1238658076347	waue
job_200904021140_0006	2	1238658644666	waue

-history

  • 印出程序的歷史狀態 
    $ bin/hadoop job -history /user/waue/stream-output1

Hadoop job: job_200904021140_0005
=====================================
Job tracker host name: gm1.nchc.org.tw
job tracker start time: Thu Apr 02 11:40:06 CST 2009
User: waue
JobName: streamjob9019.jar
JobConf: hdfs://gm1.nchc.org.tw:9000/tmp/hadoop/hadoop-waue/mapred/system/job_200904021140_0005/job.xml
Submitted At: 2-四月-2009 15:41:16
Launched At: 2-四月-2009 15:41:16 (0sec)
Finished At: 2-四月-2009 15:42:04 (48sec)
Status: SUCCESS
=====================================
...略

version

  • 印出目前的hadoop 版本 
    bin/hadoop version

Hadoop 0.18.3
Subversion https://svn.apache.org/repos/asf/hadoop/core/branches/branch-0.18 -r 736250
Compiled by ndaley on Thu Jan 22 23:12:08 UTC 2009

bin/hadoop 的管理者用法

balancer

  • 用來讓hdfs的資料能均勻分散 
    $ bin/hadoop balancer

daemonlog

  • 不太知道用法？？ 
$ bin/hadoop daemonlog -getlevel gm1.nchc.org.tw:50070 dfshealth.jsp 

Connecting to http://gm1.nchc.org.tw:50070/logLevel?log=dfshealth.jsp
Submitted Log Name: dfshealth.jsp
Log Class: org.apache.commons.logging.impl.Log4JLogger
Effective level: INFO

dfsadmin

-report

  • 呈現報告 
    $ bin/hadoop dfsadmin -report

-safemode

enter leave get wait 
$ bin/hadoop dfsadmin enter
$ bin/hadoop dfsadmin -safemode leave

-refreshNodes

  • 更新Namenode 需要退出或加入的Datanode 的清單 
    $ bin/hadoop dfsadmin -refreshNodes

-finalizeUpgrade

  • 終結升級程序

-upgradeProgress

status details force
升級狀態 狀態的細節 強制升級操作進行

-setQuota

  • 目錄配額是對目錄樹上該目錄下的名字數量做硬性限制
  • 設定配額，數字代表個數 （如：我上傳了一個2個block的檔案可以上傳，但我上傳兩個檔案很小的檔上去卻不行）
  • 配額為1可以強制目錄保持為空
  • 重命名不會改變該目錄的配額 
    $ bin/hadoop dfs -mkdir quota
$ bin/hadoop dfsadmin -setQuota 2 quota
$ bin/hadoop dfs -put ../conf/hadoop-env.sh quota/
$ bin/hadoop dfs -put ../conf/hadoop-site.xml quota/

put: org.apache.hadoop.dfs.QuotaExceededException: The quota of /user/waue/quota is exceeded: quota=2 count=3
  • 檢查目錄的配額方法： "bin/hadoop fs -count -q <目錄> " 
$ bin/hadoop fs -count -q own
   none  inf  1    0     0 hdfs://gm1.nchc.org.tw:9000/user/waue/own
$ bin/hadoop dfsadmin -setQuota 4 own
$ bin/hadoop fs -count -q own
    4     3   1    0     0 hdfs://gm1.nchc.org.tw:9000/user/waue/own

-clrQuota

  • 清除之前設定的配額 
    $ bin/hadoop dfsadmin -clrQuota quota/

namenode

  • hadoop namenode [-format] | [-upgrade] | [-rollback] | [-finalize] | [-importCheckpoint]

secondarynamenode

  • 用法：hadoop secondarynamenode [-checkpoint [force]] | [-geteditsize]

tasktracker、datanode

  • 不可直接下 bin/hadoop tasktracker 這個指令，下此指令的節點 tasktracker:50060 網頁會出現錯誤訊息
  • 可以搭配 bin/hadoop-daemon.sh --config <hadoop_conf_dir> "start|stop" "datanode｜tasktracker" 來新加入或停止節點

Hadoop Streaming 函式庫用法

  • Hadoop streaming是Hadoop的一個工具， 它幫助用戶創建和運行一類特殊的map/reduce作業， 這些特殊的map/reduce作業是由一些可執行文件或腳本文件充當mapper或者reducer
  • 最簡單的透過shell執行stream的map reduce: 
    $ bin/hadoop jar hadoop-0.18.3-streaming.jar -input input -output stream-output1 -mapper /bin/cat -reducer /usr/bin/wc
    • 輸出的結果為： (代表 行、字數、字元數) 
      2910628 24507806 143451003

HDFS權限管理用戶

  • hdfs的權限有owner, group, other三種
  • 而用戶的身份取決於client上的使用者 （用 whoami），群組為（bash -c groups）
  • 相關的操作： 
    $ bin/hadoop dfs -mkdir own
$ bin/hadoop dfs -chmod -R 755 own
$ bin/hadoop dfs -chgrp -R waue own
$ bin/hadoop dfs -chown -R waue own
$ bin/hadoop dfs -lsr own
  • conf/hadoop-site.xml 可用參數： 
    dfs.permissions = true 
dfs.web.ugi = webuser,webgroup
dfs.permissions.supergroup = supergroup
dfs.upgrade.permission = 777
dfs.umask = 022
Last modified 16 years ago Last modified on Apr 26, 2009, 1:13:25 AM